JP7807044B2 - Noise source identification device and noise source identification method - Google Patents

Description

Article 30, Paragraph 2 of the Patent Act applies. Publication date: March 1, 2021. Publication: Proceedings of the 2021 National Convention of the Institute of Electrical Engineers of Japan, Institute of Electrical Engineers of Japan. Publication date: March 5, 2021. Publication: Electromagnetic Compatibility Study Group (EMCJ), IEICE Technical Report, Vol. 120, No. 420, EMCJ2020-78, pp. 30-34, March 2021, Institute of Electronics, Information and Communication Engineers of Japan. Website posting date: September 15, 2021. Website address: https://cm2021.iri-tokyo.jp TIRI Cross Meeting 2021

This disclosure relates to a noise identification device and method for identifying noise sources and determining the magnitude of noise emitted from noise sources, which are necessary when implementing noise countermeasures (EMC countermeasures) for electronic devices, etc.

In recent years, various industrial products have been electrified, such as electric vehicles (EVs) powered by motors. These products are equipped with various electronic devices to achieve multi-functionality and miniaturization.
In addition, the above-mentioned products are equipped with, for example, a power supply unit configured using a power supply IC in order to operate electronic devices, etc., and it is becoming common for products to be equipped with multiple power supply ICs in order to operate multi-function products.
Some power supply ICs are designed to generate a predetermined voltage through switching operations. When such power supply ICs perform high-speed switching operations to output a stable voltage, high-frequency noise is generated.
When the above-mentioned products (such as power supply units) are equipped with multiple power supply ICs, the high-frequency noise generated by these power supply ICs may exceed the regulated values for radiated emissions tests (far-field measurements) set forth in EMC standards. For such products, it is necessary to implement appropriate noise countermeasures to deal with the high-frequency noise generated by each power supply IC.

In order to take appropriate measures against products that generate high-frequency noise as described above, it is important to identify the source of the noise.
A simple method for identifying a noise source is to use a magnetic field probe or the like to scan a board on which an electronic device (such as a power supply IC) is mounted to perform noise measurement (near-field measurement) and locate the noise source.
In the above measurement method, when multiple high-frequency noises are superimposed, it becomes difficult to identify the noise sources that generate each high-frequency noise.
Therefore, when a plurality of high-frequency noises are superimposed, the following techniques are available for identifying the source of each high-frequency noise.

For example, there is a measuring device in which a sensor scans the circuit board of an electronic device, determines the degree of mismatch (the difference in the amount of fluctuation in signal power) between a first signal power indicated by the output of this sensor and a second signal power indicated by the output of an antenna installed to the side of the electronic device, and identifies the source of electromagnetic interference using the sensor's position information and the degree of mismatch (Patent Document 1).
In other words, in this measuring device, the sensor detects the power transition in the near field of the electronic device (such as the phase of the electromagnetic waves in the near field), and the antenna detects the power transition in the far field of the electronic device (such as the phase of the electromagnetic waves in the far field), and the degree of mismatch is calculated from these detection results, and the location where the electromagnetic interference is occurring is identified.

Furthermore, as a technique for identifying the source of high-frequency noise, for example, a waveform of a received electromagnetic wave is used to sequentially generate waveform samples in which amplitude data and measurement times are associated at predetermined time intervals, and an amplitude probability distribution indicating the ratio of the total duration of the waveform to the duration during which the waveform exceeds a predetermined amplitude is calculated for each of the waveform samples as a feature of the received electromagnetic wave.
Thereafter, the similarity between each waveform sample is calculated, and the waveform samples are classified into a plurality of clusters based on this similarity.There is an electromagnetic wave discrimination device that identifies clusters formed by waveform samples that are interference waves using the number of waveform samples that make up each cluster, the frequency and time of reception of the waveform samples, etc. (Patent Document 2).

Furthermore, for example, there is an electromagnetic interference countermeasure support device that, when electromagnetic waves emitted from a device under test are measured using a spectrum analyzer, captures the spectrum (EMI measurement result) output from the spectrum analyzer and extracts candidate frequencies that are thought to be high-frequency noise (Patent Document 3).
This electromagnetic interference countermeasure support device is configured to select and process either a first high frequency detection method that detects a high frequency identified based on a gap value between frequency points, or a second high frequency detection method that detects a high frequency identified based on a value that is an integer fraction of the frequency.

As described above, when the electromagnetic interference countermeasure support device receives the EMI measurement results, it selects either the first or second RF detection method based on the EMI measurement results, and then processes the EMI measurement results using the selected RF detection method to extract candidate frequencies.
Thereafter, the source of the high frequency noise is identified by comparing the candidate frequencies with the clock signal or the like of the device under test that is known in advance.

JP 2013-53859 A Patent No. 5793961 Patent No. 5245946

Since the measurement device disclosed in Patent Document 1 is configured as described above, it is important to synchronize the measurement of near-field noise and the measurement of far-field noise in order to obtain a highly reliable mismatch degree. This requires a large-scale measurement system, which results in considerable costs for equipment, etc.
The electromagnetic wave discrimination device disclosed in Patent Document 2 is configured as described above, and therefore is unable to determine the magnitude of high-frequency noise in each cluster, making it difficult to identify noise frequencies that require countermeasures.
The electromagnetic interference countermeasure support device disclosed in Patent Document 3 is configured as described above, and therefore it is possible to extract multiple noise frequencies generated from a single power source. However, if multiple power sources each generate multiple noise frequencies, it becomes difficult to identify the power source that is the source of the high-frequency noise.

The present disclosure has been made to solve the above problems, and provides a noise source identification device and noise source identification method that can measure near-field noise and far-field noise asynchronously, identify multiple noise sources, and determine the noise magnitude radiated from each noise source.

A noise source identification device according to the present disclosure comprises a near-field noise measurement unit that measures near-field noise of each part of the object of measurement, a far-field noise measurement unit that measures far-field noise of the object of measurement, and an identification processing unit that identifies the source of radiated noise radiated from the object of measurement using the near-field noise waveform of each part of the object of measurement acquired from the near-field noise measurement unit and the far-field noise waveform of the object of measurement acquired from the far-field noise measurement unit. The identification processing unit generates a basic waveform corresponding to each part of the object of measurement using a portion of the far-field noise waveform that is in phase with the near-field noise waveform, compares the far-field noise waveform with the basic waveform to identify each portion of the far-field noise waveform associated with each part of the object of measurement, and performs a predetermined arithmetic process on each portion of the identified far-field noise waveform to determine the magnitude of radiated noise radiated from each part of the object of measurement.

The discrimination processing unit is also characterized in that it generates a near-field gating signal by binarizing the near-field noise waveform, generates a far-field gating signal by binarizing the far-field noise waveform, limits a search range corresponding to each part of the measurement object based on the cross-correlation between the near-field gating signal and the far-field gating signal, and sets the part of the far-field noise waveform included in the search range as the basic waveform.

The noise source identification device according to the present disclosure includes a near-field noise measurement unit that measures near-field noise of each part of the object to be measured, a far-field noise measurement unit that measures far-field noise of the object to be measured, and an identification processing unit that identifies the source of radiated noise emitted from the object to be measured using the near-field noise of each part of the object to be measured acquired from the near-field noise measurement unit and the far-field noise of the object to be measured acquired from the far-field noise measurement unit. The identification processing unit obtains the operating period of each part of the object to be measured from the near-field noise measured at each part of the object to be measured, calculates a matrix V that represents a spectrogram of the far-field noise, and calculates a matrix W that represents the frequency components and the time components of the matrix V. and a matrix H obtained by normalizing the matrix W from a predetermined objective function that indicates the difference between W and W and that takes into account the dependency of the matrix H on the operating period obtained from the near-field noise, the matrix H that minimizes the objective function when the matrix W is normalized is obtained, the waveform portion of the far-field noise waveform indicated by the matrix H that minimizes the objective function is set as the fundamental waveform corresponding to each part of the object of measurement, the far-field noise waveform is compared with the fundamental waveform to identify each waveform portion of the far-field noise waveform associated with each part of the object of measurement, and a predetermined arithmetic process is performed on each waveform portion associated with each part of the object of measurement to obtain the magnitude of the radiated noise emitted from each part of the object of measurement.

The identification processing unit is also characterized in that it labels the fundamental waveform corresponding to each part of the object to a part of the far-field noise waveform that contains the same waveform as the fundamental waveform, thereby identifying each part of the far-field noise waveform associated with each part of the object.

The identification processing unit also performs an FFT analysis of the far-field noise waveform to determine the magnitude of one radiated noise, removes any portion labeled with the fundamental waveform from the far-field noise waveform, performs an FFT analysis of the far-field noise waveform from which the any labeled portion has been removed to determine the magnitude of another radiated noise, and compares the magnitude of the one radiated noise with the magnitude of the other radiated noise to determine the magnitude of the radiated noise radiated from any portion of the object to be measured, and displays this on a display unit.

The noise source identification method according to the present disclosure is characterized by comprising: a first step in which a near-field noise measurement unit measures near-field noise of each part of the object to be measured; a second step in which a far-field noise measurement unit measures far-field noise of the object to be measured; a third step in which an identification processing unit acquires a near-field noise waveform from the near-field noise measurement unit; a fourth step in which the identification processing unit acquires a far-field noise waveform from the far-field noise measurement unit; a fifth step in which the identification processing unit generates a basic waveform corresponding to each part of the object to be measured using a portion of the far-field noise waveform that is in phase with the near-field noise waveform; and a sixth step in which the identification processing unit compares the far-field noise waveform with the basic waveform to identify each portion of the far-field noise waveform associated with each part of the object to be measured, and performs a predetermined arithmetic process on each portion of the identified far-field noise waveform to determine the magnitude of the radiated noise emitted from each part of the object to be measured.

The noise source identification method according to the present disclosure also includes a first step in which a near-field noise measurement unit measures near-field noise of each part of the object to be measured, a second step in which a far-field noise measurement unit measures far-field noise of the object to be measured, a third step in which an identification processing unit acquires the near-field noise from the near-field noise measurement unit, and a fourth step in which the identification processing unit acquires the far-field noise from the far-field noise measurement unit. The identification processing unit acquires the operating period of each part of the object to be measured from the near-field noise, calculates a matrix V representing a spectrogram of the far-field noise, and calculates a difference between the matrix V and the product of a matrix W representing a frequency component and a matrix H representing a time component, and the matrix H represents the near-field noise. a fifth step of determining, from a predetermined objective function that takes into account dependence on the operating cycle determined from far-field noise, the matrix H that minimizes the objective function when the matrix W is normalized, and determining, of the noise waveform of the far-field noise, the waveform portion indicated by the matrix H that minimizes the objective function as the fundamental waveform corresponding to each part of the object of measurement; and a sixth step of the identification processing unit comparing the far-field noise waveform with the fundamental waveform to identify each part of the far-field noise waveform associated with each part of the object of measurement, and performing predetermined arithmetic processing on each part of the identified far-field noise waveform to determine the magnitude of the radiated noise radiated from each part of the object of measurement.

According to the present disclosure, near-field noise and far-field noise can be measured separately (asynchronously) to identify multiple noise sources and determine the noise magnitude.

FIG. 1 is an explanatory diagram illustrating a schematic configuration of a noise source identification device in accordance with a first embodiment of the present disclosure. 2 is a flowchart showing the operation of the noise source identification device shown in FIG. 1 . 2 is an explanatory diagram showing a far-field noise waveform input to the discrimination processing unit of FIG. 1; FIG. 2 is an explanatory diagram showing an operation process performed by the identification processing unit of FIG. 1; FIG. FIG. 2 is an explanatory diagram illustrating an embodiment of the noise source identification device of FIG. 1. 6 is an explanatory diagram showing the schematic configuration and operation of the power supply unit of FIG. 5. 6 is an explanatory diagram showing measurement results of the power supply unit of FIG. 5. 10 is an explanatory diagram illustrating estimated values and actual measured values obtained by the discrimination processing unit of the second embodiment. FIG. FIG. 11 is an explanatory diagram showing the operation processing by the identification processing unit of the third embodiment. FIG. 2 is an explanatory diagram showing each row element h _k of a matrix H representing a time component.

The following describes one embodiment of this invention.

[First Embodiment]
FIG. 1 is an explanatory diagram illustrating a schematic configuration of a noise source identification device (identification device 1) according to a first embodiment of the present disclosure.
The identification device 1 includes a near-field probe 10 arranged near a measurement target, such as a power supply unit, and a near-field noise input unit 11 that amplifies the output signal of the near-field probe 10 and converts it into, for example, a digital signal.
The identification device 1 also includes a far-field antenna 12 that is placed, for example, several meters away from the power supply unit, and a far-field noise input unit 13 that, for example, amplifies and converts the output signal of the far-field antenna 12 into a digital signal.

The identification device 1 includes an identification processing unit 14 that uses, for example, digital signals output from the near-field noise input unit 11 and the far-field noise input unit 13, respectively, to identify noise generated by the measurement target (power supply unit).
The identification device 1 also includes a display unit 15 that displays the processing results output from the identification processing unit 14 .
The identification processing unit 14 includes, for example, a storage means (not shown) constituted by a non-volatile memory or the like, and a calculation means (not shown) constituted by a processor or the like that performs calculation processing, etc. using digital signals (data representing signals) stored in the memory or the like.
The identification processing unit 14 and the display unit 15 may be configured using, for example, a personal computer.

The near-field probe 10 is an electromagnetic field probe formed in a shape suitable for measuring near-field noise (a shape that allows easy approach to each part of the measurement target).
The near-field noise input section 11 is, for example, an electronic circuit configured to include a first high-frequency amplifier circuit that amplifies the output signal of the near-field probe 10, and a high-speed A/D converter (not shown) that converts the output signal of the first high-frequency amplifier circuit into a digital signal.
The far-field antenna 12 is, for example, a log-periodic antenna that is installed at a distance from the measurement target (such as a power supply unit) as described above.
The far-field noise input section 13 is an electronic circuit or the like configured to include a second high-frequency amplifier circuit that amplifies the output signal of the far-field antenna 12, and a high-speed A/D converter or the like (not shown) that converts the output signal of the second high-frequency amplifier circuit into a digital signal.

Next, the operation will be described.
FIG. 2 is a flowchart showing the operation of the identification device 1 shown in FIG.
The identification device 1 first brings the near-field probe 10 close to or into contact with the top surface or upper part (e.g., multiple power supply ICs mounted on a board or the like of the power supply unit) of the object to be measured (power supply unit), measures the near-field noise of each part (each power supply IC) of the object to be measured, and inputs the measured near-field noise to the identification processing unit 14 via the near-field noise input unit 11 (S101).
Next, the far-field noise of the measurement object is measured using a far-field antenna 12 or the like installed at a predetermined distance from the measurement object, and input to the identification processing unit 14 via the far-field noise input unit 13 (S102).
When measuring the near-field noise and far-field noise, power or the like is supplied to the measurement target to keep the measurement target in an operating state.

In the above-described process S101, the signal input to the identification processing unit 14 is a digital signal converted from the near-field noise waveform (voltage waveform) measured by the near-field probe 10. Hereinafter, this digital signal will be referred to as the "near-field noise waveform."
In the above-mentioned step S102, the signal input to the discrimination processing unit 14 is a digital signal converted from the far-field noise waveform (voltage waveform) measured by the far-field antenna 12. Hereinafter, this digital signal will be referred to as the "far-field noise waveform."
The discrimination processing unit 14 stores the near-field noise waveform input in step S101 in the memory or the like described above, and also stores the far-field noise waveform input in step S102 in, for example, the memory or the like described above.
The discrimination processing unit 14, which has input (stored) each near-field noise waveform and far-field noise waveform, discriminates the portion of the far-field noise to which the near-field noise generated by each power supply IC contributes (S103).
In step S103, the identification processing unit 14 of the first embodiment uses the change in switching time (ON/OFF time) of each power supply IC in the near field as an identification source signal, and identifies noise sources in the far field using this identification source signal.

Here, when the power supply IC is, for example, a DC-DC converter using PWM control, the output voltage value of the power supply IC is determined by the switching time (ON/OFF time) of the power supply IC.
That is, if the output voltages of the power supply ICs are different, the switching times of the power supply ICs will also be different, so that the power supply ICs can be identified based on the changes in the switching times of the power supply ICs (such as differences in operation timing).

Fig. 3 is an explanatory diagram showing a far-field noise waveform input to the discrimination processing unit 14 of Fig. 1. This diagram shows an example of a far-field noise waveform 200 input to the discrimination processing unit 14 by the far-field antenna 12 and the far-field noise input unit 13.
The vertical axis of the waveform graph shown in Figure 3 represents the amplitude (voltage) of the far-field noise waveform, while the horizontal axis represents the time point (any point in time) when the waveform was measured, and the width of the waveform along the horizontal axis represents the period of the waveform, etc.
The far-field noise waveform 200 shown in FIG. 3 represents the far-field noise of the measurement target measured using the far-field antenna 12, the far-field noise input unit 13, etc. as described above, and is input to the identification processing unit 14.
3 also shows the identification source signal superimposed on the far-field noise waveform 200, with areas 201, 202, 203, and 204 in the figure being portions (significant portions) of the identification source signal. Note that the insignificant portions of the identification source signal are portions connecting the areas 201, 202, 203, and 204, and are shown as a voltage level of 0 [V] in FIG.

As described above, in the operational processing of step S101, the identification device 1 measures the near-field noise radiated from each power supply IC using the near-field probe 10, and stores the near-field noise waveform input from the near-field noise input unit 11 to the identification processing unit 14 in, for example, a memory provided in the identification processing unit 14.
Furthermore, as the operational processing of step S102, the far-field noise of the measurement target is measured using the far-field antenna 12, and the far-field noise waveform 200 input to the identification processing unit 14 via the far-field noise input unit 13 is stored in, for example, a memory or the like provided in the identification processing unit 14.

Next, the identification processing unit 14 performs the following operation process in step S103.
The discrimination processing unit 14 detects, for example, rising portions (e.g., rising edges) that exceed a predetermined amount of change (that change sharply and significantly) and falling portions (e.g., falling edges) that exceed a predetermined amount of change (that change sharply and significantly) from the near-field noise waveform stored in the aforementioned memory or the like, and generates a binary signal (near-field gating signal) (of a gating waveform whose value changes in these rising portions and falling portions) using the detected rising portions and falling portions.
Furthermore, the discrimination processing unit 14 detects, for example, rising portions (e.g., rising edges) that exceed a predetermined amount of change (that change sharply and significantly) and falling portions (e.g., falling edges) that exceed a predetermined amount of change (that change sharply and significantly) from the far-field noise waveform stored in the aforementioned memory or the like, and generates a binary signal (far-field gating signal) (of a gating waveform whose value changes in these rising and falling portions) using the detected rising and falling portions.

Thereafter, the identification processing unit 14 performs a cross-correlation analysis between the binary signal generated from the near-field noise waveform and the binary signal generated from the far-field noise waveform, performs processing to align the phase of the near-field noise waveform and the phase of the far-field noise waveform, and generates an identification source signal (binary signal of the gating waveform).
As mentioned above, the identification source signal is generated using a signal obtained by binarizing the near-field noise waveform. Since the near-field noise is generated under the influence of the switching operation of each power supply IC, it includes changes related to the ON/OFF operation of each power supply IC. In other words, by comparing the near-field noise of each power supply IC, it is possible to identify each power supply IC.
From the above, it can be seen that the significant portions of the identification source signal correspond to the respective power supply ICs. That is, the identification source signal is a signal generated so that the power supply ICs can be identified by the significant portions.

When comparing the above-described identification source signal with far-field noise waveform 200, specifically, for example, as shown in Figure 3, when the identification source signal is superimposed on far-field noise waveform 200, the significant portions of the identification source signal, i.e., areas 201, 202, 203, and 204, include portions of far-field noise waveform 200 where the amplitude is large. Note that, although four areas 201, 202, 203, and 204 are shown here as examples of significant portions of the identification source signal, the number of significant portions included in the identification source signal is not limited to four (four locations).

In other words, the identification source signal specifies (limits the search range) the parts (parts of the far-field noise waveform 200) in which areas 201, 202, 203, and 204 correspond (are associated) with the power supply ICs, respectively, and include the near-field noise generated by each power supply IC.
The identification processing unit 14 compares the far-field noise waveform 200 stored in a memory or the like with the identification source signal (areas 201, 202, 203, 204) generated as described above, to identify the noise source (power supply IC) of each part of the far-field noise waveform 200.

In the operational processing of step S103, the identification processing unit 14 identifies the noise source of each part (large amplitude part) of the far-field noise waveform 200, and then generates a basic waveform (classifier) for each power supply IC (S104).

FIG. 4 is an explanatory diagram showing the operation process by the identification processing unit 14 of FIG.
For example, when the power supply unit to be measured is configured with four power supply ICs, the identification processing unit 14 generates four basic waveforms (classifiers) from the far-field noise waveform 200. In the following description, the four power supply ICs provided in the power supply unit will be referred to as the "first power supply, second power supply, third power supply, and fourth power supply."

The significant portions of the identification source signal (for example, areas 201, 202, 203, and 204) correspond to either the first power source, the second power source, the third power source, or the fourth power source, respectively.
The discrimination processing unit 14 sets the waveforms included in each area (noise waveforms in a range limited by the gating waveform of the discrimination source signal) of the far-field noise waveform 200 as basic waveforms (classifiers).
Specifically, for example, a noise waveform included in an area corresponding to the first power supply among areas 201, 202, 203, and 204 is set as the basic waveform of the first power supply.
In addition, the noise waveform included in the area corresponding to the second power supply is set as the basic waveform of the second power supply, the noise waveform included in the area corresponding to the third power supply is set as the basic waveform of the third power supply, and the noise waveform included in the area corresponding to the fourth power supply is set as the basic waveform of the fourth power supply.

Here, the identification source signal is generated by performing cross-correlation analysis between the binary signal generated from the near-field noise waveform and the binary signal generated from the far-field noise waveform as described above, so as to align the phase of the near-field noise waveform with the phase of the far-field noise waveform.
That is, each fundamental waveform is generated using a portion of the far-field noise waveform that is in phase with the near-field noise waveform.

After performing the operation process of step S104, the identification processing unit 14 identifies (labels) the far-field noise waveform 200 for each fundamental waveform of the power supply IC (S105).
Specifically, for example, the far-field noise waveform 200 stored in the memory or the like of the identification processing unit 14 is searched for a portion that includes a waveform similar to the fundamental waveform of the first power supply (for example, similar changes in amplitude or period), and also searched for a portion that includes a waveform similar to each of the fundamental waveforms of the second power supply, the third power supply, and the fourth power supply. The portions that are found by this search and include a waveform similar to each fundamental waveform are labeled. Note that this labeling is performed so that it can be identified which power supply IC the fundamental waveform of is similar to in the operation processing that will be performed later.

Thereafter, the identification processing unit 14 determines the magnitude (contribution) of the radiation noise for each power supply IC (first power supply, second power supply, third power supply, fourth power supply) and displays this on the display unit 15 or the like (S106).
When determining the magnitude (contribution) of the radiation noise of each power supply IC, which is related to the magnitude of the radiation noise of the measurement target, the portion that was labeled in step S105 is removed from the far-field noise waveform 200, and this waveform is subjected to, for example, FFT (Fast Fourier Transform) analysis as a predetermined arithmetic process to determine the magnitude of the radiation noise radiated from each power supply IC.

In detail, for example, first, the above-described FFT analysis is performed on the noise waveform when the first power supply, the second power supply, the third power supply, and the fourth power supply are all operating, i.e., the far-field noise waveform 200, to determine the magnitude of the radiation noise (noise magnitude for each frequency).
Next, the portion labeled with the basic waveform corresponding to any one of the power supply ICs (first power supply, second power supply, third power supply, or fourth power supply) is removed from the far-field noise waveform 200, and the above-mentioned FFT analysis is performed. The magnitude of the radiation noise obtained at this time (the magnitude of the other radiation noise) is compared with the magnitude of the radiation noise of the above-mentioned far-field noise waveform 200 (the magnitude of the one radiation noise), and the magnitude of the radiation noise radiated from the above-mentioned arbitrary power supply IC is determined.
Specifically, the difference is calculated between the magnitude of the radiation noise obtained by performing FFT processing on the far-field noise waveform 200 and the magnitude of the radiation noise obtained by performing FFT processing on the far-field noise waveform 200 from which the arbitrary labeling portion has been removed. This difference is the magnitude of the radiation noise emitted from the arbitrary power supply IC.

Furthermore, for example, by comparing the magnitudes of the above two radiation noises on the display screen of the display unit 15, the contribution of any power supply IC (any of the first power supply, second power supply, third power supply, or fourth power supply) to the radiation noise of the measurement target (power supply unit) can be visually seen.
Specifically, for example, a spectrum in which the vertical axis represents the noise level (for example, a voltage value indicating the magnitude of the noise) and the horizontal axis represents the noise frequency is displayed on the screen of the display unit 15 .
This spectrum display screen displays the radiated noise when all power supply ICs (first power supply, second power supply, third power supply, and fourth power supply) are operating, and the radiated noise when, for example, the first power supply is excluded (when the second power supply, third power supply, and fourth power supply are operating). The difference between these two radiated noise (spectrums) represents the magnitude (contribution) of the radiated noise radiated from the first power supply.

Furthermore, as described above, when the contribution of the radiation noise emitted from any power supply IC is known, the order of influence (rank of the first, second, third, and fourth power supplies) at each frequency (any frequency) of the far-field noise can be determined.
That is, it is possible to determine the power supply IC (the first power supply, the second power supply, the third power supply, or the fourth power supply) for which noise countermeasures should be implemented with priority at each frequency of far-field noise.

Fig. 5 is an explanatory diagram showing an embodiment of the identification device 1 of Fig. 1. This diagram shows a schematic configuration (schematic configuration of a measurement system) when the near-field probe 10 and the far-field antenna 12 of the identification device 1 are installed in an anechoic chamber 20 and noise from a power supply unit 21 is measured.
5 does not show, for example, the components of the identification device 1 (the identification processing unit 14, the display unit 15, etc.) that are installed outside the anechoic chamber 20.
The measurement target (power supply unit 21) is placed on a mounting stand 22, for example, 80 cm high, and the far-field antenna 12 is installed at a position 3 m away from the mounting stand 22. The far-field noise input unit 13 connected to the far-field antenna 12 is installed outside the anechoic chamber 20.

6 is an explanatory diagram showing the schematic configuration and operation of the power supply unit 21 of FIG. 5. The power supply unit 21 includes, for example, a plurality of power supply ICs that output a predetermined DC voltage by switching operation (by PWM control). The power supply unit 21 illustrated here includes, as the power supply ICs, a first power supply 31 that outputs a voltage of 7 V, a second power supply 32 that outputs a voltage of 8 V, a third power supply 33 that outputs a voltage of 9 V, and a fourth power supply 34 that outputs a voltage of 10 V.
As shown in FIG. 6, each of the power supply ICs has a different switching operation (ON/OFF duty ratio).

When measuring the near-field noise of the power supply unit 21, the near-field probe 10 is brought close to each of the first power supply 31, second power supply 32, third power supply 33, and fourth power supply 34, and the near-field noise of each power supply IC is measured individually (the operational processing of the aforementioned step S101 is performed).
When measuring the near-field noise and far-field noise of the power supply unit 21, power for operating the power supply unit 21 is supplied to the power supply unit 21, and a load (e.g., a load resistor) is connected to the output terminal of the power supply unit 21, for example, to keep the power supply unit 21 in a rated operating state.

Fig. 7 is an explanatory diagram showing the measurement results of the power supply unit 21 in Fig. 5. In this figure, the upper part shows the far-field noise waveform 300 of the power supply unit 21 measured by the far-field antenna 12 in Fig. 5, and the lower part shows the noise source identification results performed by the identification processing unit 14 on the far-field noise waveform 300.
The identification device 1 measures the near-field noise of each power supply IC as described above using the near-field probe 10 in Fig. 5, etc., and stores the near-field noise waveform in a memory, etc. Thereafter, the far-field antenna 12, far-field noise input unit 13, etc. in Fig. 5 are used to measure the far-field noise of the power supply unit 21 (performing the operational processing of step S102 described above).

The identification processing unit 14 of the identification device 1 processes each measured near-field noise waveform (stored in a memory or the like) and the far-field noise waveform 300, and labels the far-field noise waveform 300 for each basic waveform (steps S103 to S105).
Specifically, the identification processing unit 14 sets, for example, a basic waveform (classifier) 301 for the first power source 31, a basic waveform (classifier) 302 for the second power source 32, a basic waveform (classifier) 303 for the third power source 33, and a basic waveform (classifier) 304 for the fourth power source 34. Next, labeling of each of the basic waveforms (classifiers) is performed as shown in the lower part of FIG.
Thereafter, the magnitude (contribution) of the radiation noise of the first power source 31, the second power source 32, the third power source 33, and the fourth power source 34 is calculated (step S106).

The near-field noise and far-field noise were measured asynchronously using the identification device 1, and the magnitude (contribution) of the radiated noise from each power supply IC was determined. The results were similar to those obtained when near-field noise and far-field noise were measured synchronously using conventional devices (radiated emission tests).

As described above, the identification device 1 of embodiment 1 can asynchronously measure the near-field noise and far-field noise of the object to be measured (power supply unit 21) and identify waveform portions contained in the far-field noise waveform that are associated with each component of the object to be measured (first power supply 31, second power supply 32, third power supply 33, fourth power supply 34). Furthermore, because it is possible to asynchronously measure near-field noise and far-field noise, it is possible to prevent the measurement system from becoming too large.

Furthermore, by identifying the waveform portions as described above, it is possible to determine the magnitude of each radiated noise radiated from the first power source 31, the second power source 32, the third power source 33, and the fourth power source 34. Therefore, it is possible to determine the ranking of the noise sources (first power source 31, second power source 32, third power source 33, and fourth power source 34) that have a large influence on the far-field noise, and to identify the noise sources for which effective noise countermeasures can be taken. In other words, it is possible to identify the noise level of each power source that contributes to the far-field noise.
Furthermore, it is possible to identify the power supply noise generated by each noise source (first power supply 31, second power supply 32, third power supply 33, fourth power supply 34).
In addition, the noise frequencies generated by each noise source (first power source 31, second power source 32, third power source 33, fourth power source 34) can be determined, and frequency bands for which noise countermeasures are effective can be identified.

[Embodiment 2]
The noise source identification device (identification device 1) according to the second embodiment of the present disclosure has the same configuration as the noise source identification device (identification device 1) according to the first embodiment, but differs in the operational processing from step S103 onwards shown in FIG. 2 .
Here, the description of each part of the identification device 1 of embodiment 2 that is configured in the same way as the identification device 1 of embodiment 1 will be omitted. Also, detailed description of the operation processing that is the same as that of the identification device 1 of embodiment 1 will be omitted, and the operation processing that is characteristic of the identification device 1 of embodiment 2 will be described.

The identification device 1 of the second embodiment is installed, for example, as shown in FIG. 5, and constitutes a measurement system for measuring near-field noise and far-field noise of the power supply unit 21.
The power supply unit 21 to be measured here is configured, for example, as shown in FIG. 6, and includes four power supply ICs (first power supply 31, second power supply 32, third power supply 33, and fourth power supply 34) that each output a predetermined voltage through switching operation.
The operation of the identification device 1 of the second embodiment, which identifies the noise source of the far-field noise of the power supply unit 21 in such a measurement system, will be described below.

2, the identification device 1 of the second embodiment measures the near-field noise (voltage-time waveform) of each part (each power supply IC) to be measured and inputs the measured near-field noise to the identification processing unit 14 via the near-field noise input unit 11. The near-field noise measured here is the switching noise of each power supply IC, and the identification device 1 of the second embodiment measures the near-field noise of each power supply IC so as to be able to recognize the noise waveform (voltage-time waveform) generated during the voltage output switching operation when the output transitions from OFF to ON (rising edge), and the noise waveform (voltage-time waveform) generated during the voltage output switching operation when the output transitions from ON to OFF (falling edge).
When measuring the near-field noise of each power supply IC as described above, and when measuring the far-field noise, a load is connected to the power supply unit 21, and all power supply ICs etc. provided in the power supply unit 21 are set to their rated operating state.

As will be described later, when determining an estimated value of far-field noise, only the noise waveform at the rising edge may be used, but it is preferable to determine the estimated value of far-field noise using the noise waveform at the falling edge as well as the noise waveform at the rising edge.
Here, a processing operation for estimating far-field noise using a noise waveform during rising operation will be described.

In step S102, the identification device 1 of embodiment 2 measures the far-field noise of the measurement object using a far-field antenna 12 or the like installed at a predetermined distance from the measurement object, and inputs the measured far-field noise to the identification processing unit 14 via the far-field noise input unit 13.
The discrimination processing unit 14 of the second embodiment stores the near-field noise input in step S101 in the memory or the like described in the first embodiment, and also stores the far-field noise input in step S102 in the above-mentioned memory or the like. Note that the near-field noise, far-field noise, etc. used in the following operation processing are those stored in the above-mentioned memory or the like.
Here, the voltage-time waveform of the near-field noise of the first power supply 31 stored in a memory or the like is denoted by x ₁ (t), the voltage-time waveform of the near-field noise of the second power supply 32 is denoted by x ₂ (t), the voltage-time waveform of the near-field noise of the third power supply 33 is denoted by x ₃ (t), and the voltage-time waveform of the near-field noise of the fourth power supply 34 is denoted by x ₄ (t). Also, the far-field noise waveform (voltage-time waveform) stored in a memory or the like is denoted by y _all (t).

Next, the discrimination processing unit 14 of the second embodiment extracts a noise waveform generated by the switching operation of the power supply IC from the near-field noise waveform (voltage-time waveform) stored in the memory or the like.
Specifically, for example, the near-field noise waveform x ₁ (t) of the first power supply 31 stored in a memory or the like is compared with a signal or the like that indicates the switching operation (switching timing, i.e., ON/OFF cycle) of the first power supply 31 detected from the first power supply 31 during operation, and the noise waveform (voltage-time waveform) that is generated by the switching operation (generated at the time of rise) contained in the near-field noise waveform x ₁ (t) is identified and extracted.
The discrimination processing unit 14 of the second embodiment compares the extracted noise waveform with the far-field noise waveform y _all (t), searches for a portion of the far-field noise waveform y _all (t) that is in phase with, for example, the extracted noise waveform, and sets the noise waveform extracted by this search as the fundamental waveform of the first power supply 31. In this way, the fundamental waveform of the first power supply 31 is generated.

Similarly, a noise waveform generated by the switching operation of the second power supply 32 is extracted from the near-field noise waveform x ₂ (t) of the second power supply 32 stored in a memory or the like, and the portion of the far-field noise waveform y _all (t) that is in phase with this noise waveform is set (generated) as the basic waveform of the second power supply 32. Similarly, basic waveforms are generated from the near-field noise waveform x ₃ (t) of the third power supply 33 and the near-field noise waveform x ₄ (t) of the fourth power supply 34.

Thereafter, the discrimination processing unit 14 of the second embodiment adds the period information of the near field to the fundamental waveform to generate an estimated value of the far-field noise.
Here, for example, a signal or the like indicating a switching operation detected from the above-mentioned first power supply 31 is defined as periodic information of the near field of the first power supply 31. Similarly, a signal or the like indicating a switching operation detected from the second power supply 32 is defined as periodic information of the near field of the second power supply 32, a signal or the like indicating a switching operation of the third power supply 33 is defined as periodic information of the near field of the third power supply 33, and a signal or the like indicating a switching operation of the fourth power supply 34 is defined as periodic information of the near field of the fourth power supply 34.
The periodic information of the near field of each power source described above corresponds to the identification source signal described in the first embodiment, and identifies the position (portion) of the far-field noise waveform y _all (t) that labels the fundamental waveform of each power source.

The discrimination processing unit 14 of the second embodiment uses the periodic information of the near field of the first power source 31 to label the far-field noise waveform y _all (t) with the fundamental waveform of the first power source 31, and extracts the portion of the far-field noise waveform y _all (t) that includes the fundamental waveform of the first power source 31.
The discrimination processing unit 14 of the second embodiment regards the portion extracted by the above labeling as an estimated value _y1 (t) of the far-field noise. The estimated value _y1 (t) of the far-field noise obtained by such an operational process is an estimated value of the magnitude (intensity) of the contribution of the first power supply 31 to the far-field noise generated by the power supply unit 21.

The identification processing unit 14 of embodiment 2 performs the same operational processing as when determining the estimated value _y1 (t) of the far-field noise of the first power source 31 to determine the estimated value _y2 (t) of the far-field noise of the second power source 32, and also determines the estimated value _y3 (t) of the far-field noise of the third power source 33 and the estimated value _y4 (t) of the far-field noise of the fourth power source 34.

Thereafter, the discrimination processing unit 14 of embodiment 2 performs data conversion processing such as FFT analysis on the far-field noise estimate y ₁ (t) representing the voltage-time waveform and the far-field noise waveform y _all (t), etc., in the same manner as described in embodiment 1, to obtain the level estimate Ay ₁ (f) of the radiation noise in the far field. The level estimate Ay ₁ (f) represents the level (estimate) at which the radiation noise from the first power supply 31 contributes to the far-field noise.
Similarly, the far-field noise estimate y ₂ (t) and the far-field noise waveform y _all (t) are subjected to data conversion processing such as FFT analysis to obtain the far-field radiation noise level estimate Ay ₂ (f).
In addition, an estimated value _Ay3 (f) of the level of radiation noise in the far field is obtained from the estimated value _y3 (t) of the far field noise and the far field noise waveform _yall (t), etc., and an estimated value _Ay4 (f) of the level of radiation noise in the far field is obtained from the estimated value _y4 (t) of the far field noise and the far field noise waveform _yall (t), etc.

The level estimation value _Ay2 (f) represents the level (estimated value) at which the radiation noise of the second power source 32 contributes to the far-field noise. The level estimation value _Ay3 (f) represents the level (estimated value) at which the radiation noise of the third power source 33 contributes to the far-field noise, and the level estimation value _Ay4 (f) represents the level (estimated value) at which the radiation noise of the fourth power source 34 contributes to the far-field noise.
Thereafter, the identification device 1 of the second embodiment displays the level (estimated value) of each power supply IC's contribution to the far-field noise calculated as described above on the display unit 15, for example, in step S106.

8 is an explanatory diagram showing estimated values and actual measured values by the discrimination processing unit 14 of embodiment 2. Fig. 8 shows graphs representing the noise level in the far field of each power supply IC as a function of noise frequency, in which the top graph shows the level estimated value _Ay1 (f) of the first power supply 31 and the actual measured value by the existing EMI receiver, the second graph from the top shows the level estimated value _Ay2 (f) of the second power supply 32 and the actual measured value by the EMI receiver, the third graph from the top shows the level estimated value _Ay3 (f) of the third power supply 33 and the actual measured value by the EMI receiver, and the bottom graph shows the level estimated value _Ay4 (f) of the fourth power supply 34 and the actual measured value by the EMI receiver.

The identification device 1 according to the second embodiment is configured to measure the near-field noise of each power supply IC constituting the power supply unit 21, extract a noise waveform generated during switching operation (during rising operation) from the measured near-field noise, and, based on the extracted noise waveform, obtain an estimated value of the level of radiation noise of each power supply IC that contributes to the far-field noise from the far-field noise waveform y _all (t).

As can be seen from the graphs in FIG. 8 , the estimated values (levels) of the far-field noise contributed by each power supply IC, obtained by the identification device 1 of embodiment 2, and the actual measured values of the far-field noise measured by the EMI receiver under the same conditions as those of the above-described identification device 1 (when each power supply IC was operated independently) were almost the same.
When determining the level of the far-field noise shown in FIG. 8, the identification device 1 of embodiment 2 measures the near-field noise of each power supply IC while the first power supply 31, the second power supply 32, the third power supply 33, and the fourth power supply 34 are operating (all power supply ICs are operating), and similarly operates all power supply ICs to measure the far-field noise, and then determines an estimated value.
8, the far-field noise level measured by the EMI receiver is shown in the top graph as measured when the first power supply 31 is operated independently, and in the second graph from the top as measured when the second power supply 32 is operated independently. Similarly, the far-field noise measured when the third power supply 33 is operated independently is shown in the third graph from the top, and in the bottom graph as measured when the fourth power supply 34 is operated independently.

As described above, according to the identification device 1 of embodiment 2, similarly to the identification device 1 of embodiment 1, it is possible to asynchronously measure the near-field noise and far-field noise of the measurement target (power supply unit 21) and identify the noise levels that the first power supply 31, the second power supply 32, the third power supply 33, and the fourth power supply 34 each contribute to the far-field noise. Furthermore, because it is possible to asynchronously measure the near-field noise and the far-field noise, it is possible to prevent the measurement system from becoming large in size.
In addition, the noise frequencies generated by each noise source (first power source 31, second power source 32, third power source 33, fourth power source 34) can be determined, and frequency bands for which noise countermeasures are effective can be identified.

[Third embodiment]
The noise source identification device (identification device 1) according to the third embodiment of the present disclosure has the same configuration as the noise source identification device (identification device 1) according to the first embodiment, etc., but differs in the operational processing such as step S103 shown in FIG. 2 .
Here, for the identification device 1 of embodiment 3, a description of each unit configured in the same way as the identification device 1 of embodiment 1, etc. will be omitted. Also, a detailed description of the same operational processing as the identification device 1 of embodiment 1, etc. will be omitted, and an operational processing that is characteristic of the identification device 1 of embodiment 3 will be described.

2 , the identification device 1 of the third embodiment measures near-field noise of each part (each power supply IC) of the object to be measured, and inputs the near-field noise to the identification processing unit 14 via the near-field noise input unit 11. Furthermore, in step S102, the far-field noise of the object to be measured is measured using a far-field antenna 12 or the like installed at a predetermined distance from the object to be measured, and inputs the far-field noise to the identification processing unit 14 via the far-field noise input unit 13.
The discrimination processing unit 14 of the third embodiment stores the near-field noise input in step S101 in the memory or the like described in the first embodiment, and also stores the far-field noise input in step S102 in the above-mentioned memory or the like. Note that the near-field noise, far-field noise, etc. used in the following operation processing are those stored in the above-mentioned memory or the like.

FIG. 9 is an explanatory diagram showing the operation process by the identification processing unit 14 according to the third embodiment.
As described above, the near-field noise measured for each power supply IC is stored in a memory, etc. For example, when the near-field noise of an arbitrary power supply IC stored in a memory, etc. is plotted along the time axis, it will have a waveform like that shown in graph A in Figure 9.
As shown in graph A, a power supply IC generates periodic noise in the near field in response to its switching operation. If, for example, an autocorrelation function is applied to this periodic near-field noise, the autocorrelation shown in graph B in Fig. 9 is obtained. By calculating the average of the intervals (lags) between the peaks in graph B, the switching period _Tk (operating period) of the power supply IC can be estimated. In other words, by calculating, for example, the autocorrelation of the near-field noise, the switching period _Tk of the power supply IC or the like that is generating the near-field noise can be estimated.
Furthermore, for example, the frequency components of near-field noise can be determined by Fourier transform, and the switching period T _k (operating period) of the power supply IC can be estimated from the period in which peak values appear.

Even if elements such as power supply ICs are of the same type, there are individual differences and slight differences in the switching period _Tk . Therefore, by calculating the autocorrelation as described above and estimating the switching period _Tk , it is possible to identify each element such as a power supply IC.
The identification processing unit 14 of the third embodiment treats the switching period T _k obtained by calculating the autocorrelation of the measured near-field noise as a feature of each element (noise source) such as a power supply IC.

The identification processing unit 14 of the third embodiment obtains a matrix V representing a spectrogram of the far-field noise acquired from the far-field noise input unit 13, a matrix W representing the frequency components, and a matrix H representing the time components, and performs calculation processing using these matrices to separate and identify the portion of the far-field noise contributed by each power supply IC.
Specifically, nonnegative matrix factorization (hereinafter referred to as NMF) is used to separate (identify) the noise components contributed by each power supply IC from the measured far-field noise. At this time, a matrix H (time component) that depends on the switching period _Tk of the near-field noise is used.
In order for each row of the matrix H used in NMF to represent the activation of switching noise, it is necessary that each row has periodicity, and that this periodicity matches the switching period T _k determined from the near-field noise described above. Therefore, an algorithm that satisfies this condition is generated.

Here, the number of power supply ICs that are switching noise sources is assumed to be K, and the matrix H representing the time component is assumed to be composed of (K×M) elements.
The number of bases after factorization by NMF is set to the same as the number of power supply ICs.
The switching period T _k obtained as described above is denoted by T _k (1≦k≦K), and h _km is the element in the kth row and mth column of the matrix H.

FIG. 10 is an explanatory diagram showing each row element h _k of the matrix H representing the time component.
In order for each row element h _k of the matrix H to include the switching period T _k , that is, for the row vector H _k to depend on the switching period T _k , the value of the autocorrelation R _k in the following equation (1) must become large.

When the spectrogram of far-field noise is represented by a matrix V, in NMF, the matrix V can be expressed as approximating the product of a matrix W representing the frequency components (base frequencies) of the far-field noise and a matrix H representing the time components of the far-field noise.
Therefore, in order to identify the portion of the far-field noise that is contributed by the power supply IC that generates each near-field noise, the difference between the matrix V and the matrix W·matrix H (the product of the matrix W and the matrix H) is minimized, and the negative autocorrelation of each row vector H _k of the matrix H is minimized.
Specifically, matrices W and H that minimize an objective function such as the following equation (2) are obtained.

Here, matrix V is composed of (N×M) elements, and matrix W is composed of (N×K) elements. α is a weight (constant) related to the periodicity of row vector H _k . ν _nm represents the nth row, mth column element of matrix V, and (WH) _nm represents the nth row, mth column element of the product of matrix W and matrix H. All elements of matrix W and matrix H are non-negative. K is a natural number smaller than M and N.
In the above equation (2), the upper part indicates the difference between the matrix V and the product of the matrix W and the matrix H, and the lower part indicates the terms related to each row vector H _k of the matrix H, which indicates that the matrix H depends on the switching period T _k .

The identification processing unit 14 of embodiment 3 performs partial differentiation on each element of the matrices W and H in equation (2), and uses the gradient method to find the matrices W and H that minimize L in equation (2) (the difference between matrix V and matrix W × matrix H).
At this time, the matrix W is normalized so that the value of each column is 0 or 1, and the matrix H that minimizes the objective function of equation (2) is obtained.
In this way, the matrix H that depends on the switching period T _k of any power supply IC is obtained, and the noise portion (portion indicated by the matrix H) contributed by the power supply IC is separated and identified from the far-field noise.

The identification processing unit 14 of the third embodiment performs the above-mentioned separation processing for each power supply IC provided in the measurement target, thereby separating (identifying) the noise waveforms contributed by each power supply IC from the far-field noise waveform 200 described in the first embodiment, for example.
Specifically, for each power supply IC, the autocorrelation of the measured near-field noise waveform is calculated to obtain the switching period T _k (average value), and the matrix H described above is calculated from equation (2) including this switching period T _k .
The matrix H calculated for each power supply IC can identify, for example, the position where the near-field noise (switching noise) generated by each power supply IC exists in the far-field noise waveform 200 shown in FIG. 3 (the portion of the far-field noise waveform 200 shown along the time point axis to which the near-field noise contributes).
That is, the identification processing unit 14 of the third embodiment uses the matrix H calculated for each power supply IC (switching period T _k ) to separate and identify the portion of the far-field noise waveform 200 (the noise waveform of the far-field noise measured by the far-field noise input unit 13) to which the near-field noise related to each power supply IC contributes.

As described above, after identifying the portion of the far-field noise waveform 200 to which each power supply IC contributes, the identification processing unit 14 of the third embodiment performs each operation process from step S104 onwards as described in the first embodiment.
In addition, since the identification processing unit 14 of the third embodiment identifies each part of the far-field noise waveform 200 using the matrix H corresponding to each power supply IC in step S103, in step S104, a basic waveform (classifier) is generated from the part identified by each matrix H.
As described above, the discrimination processing unit 14 of the third embodiment knows the switching period T _k of each power supply IC. Therefore, for example, the portion of the far-field noise waveform 200 indicated by the matrix H may be used as a reference to generate a basic waveform (classifier) using a waveform within a range limited by the switching period T _k .

In the above-described first to third embodiments, a power supply IC was used as an example of each part of the measurement object. However, if the measurement object includes elements (electronic devices, etc.) that generate noise with different periods, the identification device 1 can identify the radiated noise emitted from each part of the measurement object (elements, electronic devices, etc.) and determine the magnitude of each radiated noise.

REFERENCE SIGNS LIST 1 Identification device 10 Near-field probe 11 Near-field noise input unit 12 Far-field antenna 13 Far-field noise input unit 14 Identification processing unit 15 Display unit 20 Radio wave anechoic chamber 21 Power supply unit 22 Installation stand 31 First power supply 32 Second power supply 33 Third power supply 34 Fourth power supply

Claims (4)

a near-field noise measurement unit that measures near-field noise from each part of the object to be measured that may be a periodic noise source ;
a far-field noise measurement unit that measures the far-field noise of the measurement target;
an identification processing unit that identifies a noise source of radiation noise radiated from the object to be measured, using the near-field noise waveform of each portion of the object to be measured acquired from the near-field noise measurement unit and the far-field noise waveform of the object to be measured acquired from the far-field noise measurement unit;
Equipped with
The identification processing unit
a fundamental waveform generating means for generating fundamental waveforms corresponding to the respective portions of the object to be measured using portions of the far-field noise waveform that are in phase with the near-field noise waveform of each portion of the object to be measured;
a far-field noise waveform portion identification means for comparing the far-field noise waveform with each fundamental waveform corresponding to each portion of the object to be measured , and identifying each portion of the far-field noise waveform associated with each portion of the object to be measured;
a radiation noise magnitude calculation means for performing a predetermined calculation process on each part of the identified far-field noise waveform to determine the magnitude of the radiation noise radiated from each part of the measurement object;
Equipped with
the basic waveform generating means generates near-field gating signals by binarizing the near-field noise waveform corresponding to each part of the object to be measured, generates far-field gating signals by binarizing the far-field noise waveform, aligns the phase of the near-field noise waveform and the phase of the far-field noise waveform from the cross-correlation between each near-field gating signal and the far-field gating signal, limits a search range for the far-field noise waveform corresponding to each part of the object to be measured, and sets parts of the far-field noise waveform included in the search range corresponding to each part of the object to be measured as the basic waveform that serves as a classifier for each part of the far-field noise waveform,
the far-field noise waveform portion identification means searches the far-field noise waveform for each fundamental waveform corresponding to each portion of the object to be measured, labels portions including the same waveform as the fundamental waveform, and identifies each portion of the far-field noise waveform associated with each portion of the object to be measured;
the radiation noise magnitude calculation means performs an FFT analysis of the far-field noise waveform to determine the magnitude of one radiation noise, removes a portion labeled with any of the fundamental waveforms from the far-field noise waveform, performs an FFT analysis of the far-field noise waveform from which the arbitrary labeled portion has been removed to determine the magnitude of another radiation noise, and determines the difference between the magnitude of the one radiation noise and the magnitude of the other radiation noise, thereby determining the magnitude of the radiation noise radiated from any portion of the object to be measured.
A noise source identification device comprising: a near-field noise measurement unit that measures near-field noise from each part of the object to be measured that may be a periodic noise source;
a far-field noise measurement unit that measures the far-field noise of the measurement target;
an identification processing unit that identifies a noise source of radiation noise radiated from the object to be measured using the near-field noise waveform of each portion of the object to be measured acquired from the near-field noise measurement unit and the far-field noise waveform of the object to be measured acquired from the far-field noise measurement unit;
Equipped with
The identification processing unit
an operating period acquisition means for acquiring an operating period T _k of each part of the object to be measured from a near-field noise waveform measured at each part of the object to be measured ;
a basic waveform generating means for determining a matrix V representing a spectrogram of the far-field noise waveform, determining a matrix H that minimizes the objective function when the matrix W is normalized from an objective function of equation (1) that indicates the difference between the product of the matrix V and a matrix W representing frequency components and a matrix H representing time components and that takes into account that the matrix H depends on the operating period T _k determined from the near-field noise waveform, and setting a waveform portion of the far-field noise waveform that is indicated by the matrix H that minimizes the objective function as a basic waveform that serves as a classifier for each portion of the far-field noise waveform corresponding to each portion of the object to be measured;
a far-field noise waveform portion identification means for comparing the far-field noise waveform with each fundamental waveform corresponding to each portion of the object to be measured, and identifying each portion of the far-field noise waveform associated with each portion of the object to be measured;
a radiation noise magnitude calculation means for performing a predetermined calculation process on each part of the identified far-field noise waveform to determine the magnitude of the radiation noise radiated from each part of the measurement object;
Equipped with
the far-field noise waveform portion identification means searches the far-field noise waveform for each fundamental waveform corresponding to each portion of the object to be measured, labels portions including the same waveform as the fundamental waveform, and identifies each portion of the far-field noise waveform associated with each portion of the object to be measured;
the radiation noise magnitude calculation means performs an FFT analysis of the far-field noise waveform to determine the magnitude of one radiation noise, removes a portion labeled with any of the fundamental waveforms from the far-field noise waveform, performs an FFT analysis of the far-field noise waveform from which the arbitrary labeled portion has been removed to determine the magnitude of another radiation noise, and determines the difference between the magnitude of the one radiation noise and the magnitude of the other radiation noise, thereby determining the magnitude of the radiation noise radiated from any portion of the object to be measured.
A noise source identification device comprising :
a first step in which a near-field noise measurement unit measures near-field noise of each part of the measurement object that may be a periodic noise source;
a second step of measuring the far-field noise of the measurement target by a far-field noise measurement unit;
a third step in which an identification processing unit acquires a near-field noise waveform from the near-field noise measurement unit;
a fourth step in which the identification processing unit acquires a far-field noise waveform from the far-field noise measurement unit;
a fifth step in which the discrimination processing unit generates each basic waveform corresponding to each part of the object to be measured using each part of the far-field noise waveform that is in phase with the near-field noise of each part of the object to be measured;
a sixth step in which the identification processing unit compares the far-field noise waveform with each basic waveform corresponding to each part of the object to be measured, identifies each part of the far-field noise waveform associated with each part of the object to be measured, and performs arithmetic processing on each part of the identified far-field noise waveform to determine the magnitude of radiation noise radiated from each part of the object to be measured;
and
in the fifth step, near-field gating signals are generated by binarizing the near-field noise waveform corresponding to each part of the object to be measured, far-field gating signals are generated by binarizing the far-field noise waveform, the phases of the near-field noise waveform and the far-field noise waveform are aligned based on the cross-correlation between each near-field gating signal and the far-field gating signal, a search range for the far-field noise waveform is limited to correspond to each part of the object to be measured, and parts of the far-field noise waveform included in the search range corresponding to each part of the object to be measured are set as the basic waveforms that serve as classifiers for each part of the far-field noise waveform,
In the sixth step, each fundamental waveform corresponding to each part of the object to be measured is searched from the far-field noise waveform, portions including the same waveform as the fundamental waveform are labeled to identify each portion of the far-field noise waveform associated with each part of the object to be measured, an FFT analysis of the far-field noise waveform is performed to determine the magnitude of one radiated noise, portions labeled with any of the fundamental waveforms are removed from the far-field noise waveform, an FFT analysis of the far-field noise waveform from which the any of the labeled portions has been removed is performed to determine the magnitude of another radiated noise, and a difference between the magnitude of the one radiated noise and the magnitude of the other radiated noise is obtained, thereby determining the magnitude of the radiated noise radiated from an arbitrary part of the object to be measured.
A noise source identification method comprising: a first step in which a near-field noise measurement unit measures near-field noise of each part of the measurement object that may be a periodic noise source;
a second step of measuring the far-field noise of the measurement target by a far-field noise measurement unit;
a third step in which an identification processing unit acquires the near-field noise waveform from the near-field noise measurement unit;
a fourth step in which the identification processing unit acquires the far-field noise waveform from the far-field noise measurement unit;
a fifth step in which the identification processing unit acquires an operating period T _k of each part of the object of measurement from the near-field noise waveform , calculates a matrix V representing a spectrogram of the far-field noise, calculates a matrix H that minimizes the objective function when the matrix W is normalized from an objective function of equation (2) that indicates the difference between the matrix V and a product of a matrix W representing a frequency component and a matrix H representing a time component and that takes into account that the matrix H depends on the operating period T _k obtained from the near-field noise waveform, and sets a waveform portion of the far-field noise waveform that is indicated by the matrix H that minimizes the objective function as a basic waveform that serves as a classifier for each portion of the far-field noise waveform that corresponds to each part of the object of measurement;
a sixth step in which the identification processing unit compares the far-field noise waveform with each basic waveform corresponding to each part of the object to be measured, identifies each part of the far-field noise waveform associated with each part of the object to be measured, and performs a predetermined arithmetic process on each part of the identified far-field noise waveform to determine the magnitude of radiation noise radiated from each part of the object to be measured;
and
In the sixth step, each fundamental waveform corresponding to each part of the object to be measured is searched from the far-field noise waveform, portions including the same waveform as the fundamental waveform are labeled to identify each portion of the far-field noise waveform associated with each part of the object to be measured, an FFT analysis of the far-field noise waveform is performed to determine the magnitude of one radiated noise, portions labeled with any of the fundamental waveforms are removed from the far-field noise waveform, an FFT analysis of the far-field noise waveform from which the any of the labeled portions has been removed is performed to determine the magnitude of another radiated noise, and a difference between the magnitude of the one radiated noise and the magnitude of the other radiated noise is obtained, thereby determining the magnitude of the radiated noise radiated from an arbitrary part of the object to be measured.
A noise source identification method comprising :