JPS645765B2 - - Google Patents

Description

[Detailed description of the invention]

[Field of Application] The present invention relates to a digital filter arithmetic device, and in particular, the present invention relates to a digital filter arithmetic device, and in particular, the present invention relates to a digital filter arithmetic device that calculates periodic fluctuation components, so-called ripple components, included in a state quantity calculated based on a signal obtained by sampling an analog signal and converting it into a digital quantity. The present invention relates to a digital filter calculation device that can effectively remove the filter. [Prior Art] In conventional digital arithmetic devices that calculate a specific state quantity from a digital quantity obtained by sampling an analog signal at regular time intervals, when the input signal has a frequency component, the state quantity calculation result is Digital filtering is applied to remove ripples contained in the image. This filtering processing method is usually
A moving average value calculation method using the state quantity calculation result as an input signal is used. Furthermore, in order to improve the ripple removal effect at this time, for example,
No. 56-50462) "Moving average value digital arithmetic control device", there is a certain frequency component f _O of the input analog signal between the sampling time interval and the number of data used to calculate the moving average value.
It is known that settings are made so that the following conditions (1) to (3) are satisfied. Condition (1) The N pieces of state quantity data used to calculate the moving average value must fall within one cycle of the periodic component of the state quantities, when viewed equivalently. That is, Nθ=Mτ...(1) In other words, Nθ/τ is a positive integer M. Condition (2) The phase positional relationship of the N pieces of data used for calculating the moving average value is equivalently arranged at equal intervals. That is, N≠K・M...(2) In other words, N/M is not a positive integer. Condition (3) In the phase positional relationship of the N pieces of data used to calculate the moving average value, there is no data that has the same phase point. That is, K·θ≠τ (3) However, in the above formulas (1) to (3), the definitions of each symbol are as follows. θ=2πf _O /f _S τ=2πf _O /f _T M: Positive integer greater than or equal to 2 K: Positive integer greater than or equal to 1 f _S : Sampling frequency (Hz) f _T : State quantity (to moving average value calculation device) f _O : Fundamental frequency (Hz) of the analog input signal (input signal of A case is shown in which active power P _F1 is calculated using and AC current I as analog input signals. In the figure, 1 is an analog/digital converter, 2 is a sampling timing control device, 3 is a multiplier for calculating a state quantity (active power in this case), and 4 is a moving average value calculation device. Further, V is an AC voltage input signal, I is an AC current input signal, P is time series data which is an output signal of the multiplier 3, P _F1 is an output signal of the moving average value calculation device 4,
Each is shown below. In this example, the analog input signals V and I are V = V _n sin (ω _t ) I = I _n sin (ω _t −θ ₀ ) However, if ω = 2πf ₀ , the analog/digital signal at sampling time T = t _i Each output signal of converter 1 is as follows.
It becomes as shown in equations (4) and (5). V _T=ti = V _n sin (2πf ₀ /f _S・i) ...(4) I _T=ti = I _n sin (2πf ₀ /f _S・i−θ ₀ ) ...(5) i=1 , 2, ... Therefore, the output signal P of the multiplier 3 that calculates the state quantity is expressed by equation (6). P _T=ti =V _n I _n /2cosθ ₀ −V _n I _n /2cos (4πf ₀ /f _S・i
-θ ₀ ) ...(6) In this case, as is clear, the purpose of the moving average value calculation device 4 is to remove the double frequency component of the second term of the equation (6). Here, for example, if the fundamental frequency f ₀ of the analog input signal is 50Hz, then the fundamental frequency f _T of the state quantity is
is 2·f ₀ or 100Hz. At this time,
Assume that f _S =562.5 Hz and N=45 are selected as the sampling frequency f _S that satisfies the above-mentioned conditional expressions (1) to (3) for calculating the moving average value, and the number of data used for calculating the moving average value. Substituting these numerical values into the conditional expressions (1) to (3) above, first for equation (1), M=Nθ/τ=45・f _T /f _S =45×100/562.5=8, that is, M Since is a positive integer, it can be seen that equation (1) is satisfied. Next, by substituting N=45 and M=8 into equation (2), we get
Since N/M=45/8=5.625 and the quotient is not a positive integer, it can be seen that equation (2) is also satisfied. Finally, by substituting the above values of θ and τ into equation (3), we get
τ/θ=f _S /f _T =562.5/100=5.625, and since the quotient is not a positive integer, it can be seen that equation (3) is also satisfied. When the sampling conditions are set as above, as can be seen from equations (4) and (5) above, the analog input AC voltage, current V, and I are θ=2πf ₀ /f _S =
The signal is sampled at 32° intervals and supplied to the multiplier 3. In addition, the output of the multiplier 3, that is, the electric power, which is a state quantity, includes a frequency component twice that of the input voltage and current, as can be seen from equation (6), so the sampling interval of 32° mentioned above is a state quantity. For a certain amount of electricity,
This corresponds to a sampling interval of 32×2=64°. If the power waveform, which is a state quantity, is plotted at a sampling interval of 64 degrees, it will look like Figure 8 (only a portion is shown here), and each sampling point will be plotted at equal intervals of 8 degrees in terms of phase. It can be seen that there is a fixed distribution. In the figure, the number outside the circle is the sampling number i, and the number inside the circle is the corresponding phase angle. As is clear from the above explanation, the detection error included in the output P _F1 of the moving average value calculation device 4 at this time is such that the second term component shown in equation (6) becomes completely zero when f = 50Hz. . However, in other frequency ranges, the above three conditions do not hold, and the sampling phase of the state quantity, which is time series data, increases as the frequency deviation from 50Hz increases, so the error in the moving average value gradually increases. becomes larger. For example, when the power factor is 1.0, it will exist in the shaded area of the detection error vs. input frequency characteristic shown in FIG. In FIG. 2, the horizontal axis is the input frequency, and the vertical axis is the detected difference. In the conventional device described above, as shown in FIG. 2, the ripple removal effect by the moving average value calculation device 4 reaches its maximum at a certain frequency (for example, 50 Hz in the example of FIG. 2). The sampling frequency and the number of data used for calculating the moving average value are set so that the detection difference is minimized. Therefore, when it is necessary to detect state quantities over a certain frequency range (f ₁ ≦ f ≦ f ₂ ),
In practice, the following problems arose. In other words, since the detection error changes depending on the frequency as shown in Figure 2, the more the frequency of the input signal deviates from the value at which the detection error becomes the minimum value, the more the ripple component by the moving average value calculation device increases. This method has the disadvantage that errors due to incompleteness of the removal effect rapidly increase, making it impossible to obtain the detection accuracy required for practical use. [Object] It is an object of the present invention to provide a digital filter arithmetic device that eliminates the above-mentioned drawbacks and provides a good ripple removal effect over an arbitrary frequency range by calculating a moving average value. [Summary] The present invention focuses on the fact that the ripple component included in the moving average value calculation result described above is the frequency component of the input signal of the moving average value calculation device, and converts this ripple component into the second moving average value calculation device. The structure is such that it is removed by. The feature of the present invention is that the number of time series data used in the second moving average value calculation device, that is, the number of output signals of the moving average value calculation device, is the sum of the phase intervals of these number of time series data. is selected to be close to an integral multiple of the period of the time series data. Another feature of the present invention is that the frequency at which the ripple removal effect is maximized in the second moving average value calculation device is set to a value different from the center frequency of the first moving average value calculation device. The point is that the detection error characteristics are almost flattened over the frequency domain. [Example] An example of the present invention is shown in FIG. In this figure, parts equivalent to those in FIG. 1 are given the same symbols and numbers. In the figure, 4 is the first stage moving average value calculation device, 4A is the second stage moving average value calculation device, P _F1
P F2 indicates first time series data which is an output signal of the first moving average value calculation device 4, and P _F2 indicates second time series data which is an output signal of the second moving average value calculation device. In this embodiment, the number N of data used in the first-stage moving average value calculation device 4 and the sampling frequency f _S of the analog input signal are set so as to satisfy the above-mentioned conditional expressions (1) to (3). It is assumed that That is, for example, below the input frequency f ₀ is 50
Based on Hz, the sampling frequency f _S is 562.5Hz,
It is assumed that the number of data N is selected to be 45. In this state, the input frequency changes from f ₀ to f _A (= f ₀
+Δf), the result of the first moving average value calculation P _F1 is calculated by changing P _T in equation (6) to i
Since it is the average from = 1 to i = 45, (7)
It comes to be expressed by the formula. P _F1 = 1/45 ₄₅ 〓 ⁱ⁼¹ V _n I _n /2cosθ ₀ +1/45 ₄₅ 〓 〓 ⁱ⁼¹ V _n I _n /2cos (4π(f ₀ +Δf)/f _S・i−θ ₀ )… …(
7) The second item in equation (7) above, the moving average value of the double frequency component, is calculated when the phase positional relationship of each data is Δf
It cannot be completely removed because it changes depending on the change in Then, this component appears as a ripple component in the output signal (second time series signal) of the first-stage moving average value calculation device 4. At this time, the fundamental frequency of the ripple component is 2 of f _A.
The frequency component is 2f _A , and the phase interval of the data is defined by 2π(2f _A )/f _S . For this reason, the second stage moving average value calculation device 4A
In this case, the moving average value of time-series data whose frequency component is 2f _A and the data phase interval is 2π(2f _A )/f _S is determined. Now, when f _A = f _AO , let N _A be the number of data items used to calculate the moving average value so that this ripple can be completely removed. N _A is given by the relationship in equation (8). 2π・(2f _AO )/f _S・N _A = τ _A・k _A ……(8) Then, f _AO : The second stage where the ripple component becomes zero due to the calculation of the moving average value in the second stage Frequency of input signal P _F1 τ _A : Period of second stage input signal P _F1 N _A : Positive integer of 1, 2... k _A : Positive integer of 1, 2... In other words, N _A The total phase interval of the data is
It may be made equal to an integral multiple of the period of the time series data. Looking at this in the embodiment shown in FIG. 3, τ _A =2π. Therefore, if f _AO at this time is determined from the above equation (8), f _AO = k _A /2·N _A ·f _S. Table 1 shows k _A , N _A , the frequency f _AO at which the ripple component becomes zero, and the frequency deviation Δf _A from the center frequency f ₀ of the first-stage moving average value calculation device 4.
This shows an example of a trial calculation when the sampling frequency f _S is 562.5 Hz. As shown in Table 1, the moving average value calculation in the first stage

〔effect〕

As is clear from the above description, according to the present invention, it is possible to easily obtain detection accuracy that is improved by more than one order of magnitude over a wide frequency band as compared to the conventional method. Moreover, as a result, the frequency dependence characteristics of the state quantity detection results can be almost eliminated.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an example of a conventional digital filter calculation device, FIG. 2 is a detection error-input frequency characteristic diagram thereof, FIG. 3 is a block diagram showing an embodiment of the present invention, and FIGS. FIG. 7 is a detection error-input frequency characteristic diagram obtained by the digital filter arithmetic device of the present invention shown in FIG. 3, and FIG. 8 is a diagram showing an example of the sampling phase when the moving average error becomes zero. It is. 1... Analog/digital converter, 2... Sampling timing control device, 3... Multiplier,
4...First-stage moving average value calculation device, 4A...Second-stage moving average value calculation device.

Claims (1)

[Claims] 1. A first means for sampling an input analog signal at a constant period and converting it into a digital quantity, calculating a specific state quantity of the input signal based on the digital quantity, and converting the input analog signal into a first time series data. The second output as
a third means for calculating a moving average value of the first time series data of the second means for calculating the state quantity; and a third means for calculating a moving average value of the first time series data by the third means, In a digital filter arithmetic device having a fourth means for calculating a moving average value again, the moving average value is calculated by the third means for a certain predetermined frequency related to the frequency band used in the input analog signal. The phase angles (electrical angles) of the N pieces of time series data used in In addition, the sampling frequency of the first means is set so that the sum of the phase intervals of the N pieces of time series data is an integral multiple of the period of the output signal, and the moving average value calculation of the fourth means is performed. The number _N of time series data to be used is set such that the sum of the phase intervals of the N _A pieces of data is a value close to an integral multiple of the period of the second time series data that is the output signal of the third means. A digital filter calculation device characterized by: 2 The sum of the phase intervals of the N _A pieces of data is the third
A digital filter calculation device characterized in that the value is set to be closest to an integral multiple of the period of the second time series data which is the output signal of the means. 3. Claims 1 or 2, characterized in that a certain predetermined frequency of the input analog signal is equal to the center frequency of the frequency band in use.
The digital filter arithmetic device described in Section 1. 4. The digital filter calculation device according to claim 1 or 2, wherein a certain predetermined frequency of the input analog signal is different from the center frequency of the frequency band used.