JPH0715404B2 - Electronic balance - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an electronic balance, and in particular, for removing a noise component from time series data obtained by sampling a weighing signal including noise at every time T. Data processing technology.

<Prior art> Electronic balances use the analog output of the load-electricity converter as an A / D
The converter samples every predetermined time T, for example, every 0.2 seconds, and digitizes it, but conventionally, a method of reducing the noise component by arithmetically averaging this numerical data has been used. That is, when the sampled time series data is {Xn}, the averaged time series data {n} is Becomes

<Problems to be Solved by the Invention> On the other hand, the present inventor has proposed that the first time series data {Xn}
For example, every 4th data of this time series data is extracted and the cycle is an integral multiple, for example, 4 times the secondary time series data {Yn}
Make Here, a novel data processing method is proposed in which M is a positive integer of 2 or more and n0 is a positive integer by executing digital arithmetic means.

An electronic balance using this digital filter means exhibits excellent attenuation characteristics even for extremely low frequency noise, and can be realized by only computer software.
It has the advantages that the configuration is simple and that data processing does not require a long averaging time as in the conventional example.

However, when the time series variation of the sampled signal before digital filtering is large, when switching between different digital filters in sequence, if the initial value when the digital filtering is started is a biased value. For example, there is a problem that the effect remains for a considerably long time and discontinuity occurs in the value when the digital filter is switched.

Therefore, an object of the present invention is to provide an electronic balance that performs data processing using a digital filter, which can output a stable display value without discontinuity even when there is a large variation in the time series sampled first. Is to provide.

<Means for Solving Problems> The electronic balance of the present invention converts the load on the weighing pan into an electric signal, and uses the time-series data obtained by sampling the electric signal every predetermined time (Tx). In the device for displaying the load value, based on the first-order time series data {Xn} obtained by sampling at the above-mentioned predetermined time (Tx), based on the calculation content represented by the following formula (3), A first digital filter means for obtaining second-order time series data {Yn} whose cycle (Ty) is an integer multiple of the above-mentioned predetermined time (Tx), and its second
From the next time series data {Yn}, the cycle (Tz) is the above cycle (Tz).
y)) second digital filter means for obtaining third-order time-series data {Zn} that is an integer multiple of y), and whether or not the second-order time-series data {Yn} falls within a predetermined stability width and is stable. It is characterized in that it has control means for starting the calculation of the second digital filter means after the width has been reduced.

(However, M is a positive integer and M ≧ 2) In the present invention, it goes without saying that the integral multiple of the period of the time series data includes one.

<Operation> FIG. 1 shows a time chart of each time series data. The data is sampled every predetermined time Tx, for example, every 0.2 seconds, and the primary time series data {Xn} is obtained. The first digital filter means executes the operation of the equation (3) using the first-order time series data {Xn} to obtain the second-order time series data {Yn}. After it is determined that the second-order time series data {Yn} has fallen within a predetermined stability width, the operation of the second-order digital filter means is executed to obtain the third-order time series data {Zn}. The same data processing can be executed using the third and subsequent digital filter means, and the final time series data which is the calculation result of the final digital filter means is supplied to the weighing data.

When the first-order time-series data {Xn} obtained by sampling the electrical conversion signal of the load is changing, the second-order time-series data {Yn} is not stable, so the processing by the second digital filter means No, so the display value is not affected by the load sampling data in varying conditions over time.

<Embodiment> FIG. 2 is a block diagram showing the overall configuration of an embodiment of the present invention.

The load-electricity converter 1 converts the load W on the weighing pan 2 into an electric signal. The A / D converter 3 outputs digitally converted numerical value data Dn every predetermined time T, for example, every 0.2 seconds. In addition to the load W, noise is included in this. The data processing unit 4 includes a CPU 5, a ROM 6, and a RAM 7, and the RAM 7 has an X register (X _{0 to} X ₃ ), a Y register (Y _{0 to} Y ₁₆ ), and a Z register (Z which temporarily store time series data. _{0 to} Z ₁₆ ), W register (W ₀
To _W-16), and a counter C _1, C _2, and the flag INT1,2,3. The ROM 6 stores a program to be described later. The digital filter of the present invention is configured by this program and each of the above registers.

FIG. 3 shows a flow chart of the program of the embodiment of the present invention.

At the initial setting 11, the X register, Y register, Z register, W register, counters C ₁ and C ₂ and flags INT1, INT2 and INT3 are cleared. Next, when the sampling signal X is input from the A / D converter, it is multiplied by K = 16, for example (12). This 16-fold multiplication is executed by shifting left by 4 bits in the pure binary class. The product of K times is the primary time series {Xn}.

The reason why the sampling input X is multiplied by K is that the display value is completely returned to the original value when the input signal is a step signal or an impulse signal. It should be noted that the display value is not affected because it is divided by K after the series of digital filter processing is completed.

Initially, since the flag INT1 = 0, the process proceeds to step 13, and the data X is input in parallel to all the digits of the X and Y registers. This is to speed up the initial response of the digital filter.
At the same time, INT1 = 1 is set.

Since the flag INT = 1 after the second time, the routine proceeds to step 14, where the sampling input X is input to the minimum digit of the X register.

Calculation of primary digital filter in step 15 Is executed. This calculation is executed every time the sampling input X ₀ arrives, and the secondary time series {Yn} is created. At the same time, the contents of the X register and the Y register are shifted by one digit, the data of X ₃ and X ₁₆ are discarded, and at the same time, the counter C ₁ is incremented by ₁ . This counter C ₁ is for sampling one out of every four of the secondary time series {Yn} to create a tertiary time series {Zn}. INT2 = 0 in step 16
When, the contents of c ₁ are cleared.

The judgment step 17 judges whether or not the secondary time series {Yn} is stable. Specifically, the data of each digit of the Y register
The minimum value and the maximum value of Y ₁ , Y ₂ , ... Y ₁₆ are extracted, and when the difference is smaller than the predetermined amount A, it is judged that the stable width has been reached. The determination as to whether or not the data is stable may be made even if the data of all 16 digits is not complete, for example, if all of the 3 data are within a predetermined width. If it is judged that it is not stable yet, it is judged that there is no stable data, and the data that is changing in step 18. Is displayed.

On the other hand, when it is judged in the judgment step 17 that the time-series data {Yn} is stable, the process proceeds to step 19 and the average value of the stable data Y _{1 to} Y ₁₆ is calculated. Is calculated. When INT2 = 1 in step 20, stable data / 16 is displayed in step 22, and INT2 = 0
, The stable average value is Z ₁ , Z ₂ , Z _{3 of the} Z register.
And is set to INT2 = 1. On the other hand, if INT2 = 1 in judgment step 16, in step 23
When C ₁ = 4, the second digital filter is calculated in step 24. Is executed and stored in Z ₀ . At the same time, the contents of the Z register are shifted by one digit, the counter C ₂ is incremented by 1, and the counter C ₁ is cleared.

What should be noted here is that Y ₁ , Y ₅ , and
It means that the data of Y ₉ and Y ₁₃ and every four data are sampled. In this way, when sampling four pieces of data of the previous time series data and performing the calculation of the next digital filter, the maximum point (minimum point) of the frequency characteristic of the previous digital filter and the digital filter of the subsequent digital filter are sampled. There is an effect that the minimum point (maximum point) of the frequency characteristic coincides and the overall frequency characteristic becomes very good. Because the attenuation at the minimum point of the frequency of the digital filter is -∞ dB, the output is completely 0, and the total frequency characteristic also shows the attenuation at-.
This is because it becomes ∞ dB.

It is judged in step 26 whether the third time series data {Zn} obtained by the second digital filter is also stable, and after it is stable, the process proceeds to step 27, and similarly, the time series {Zn} is obtained. }, The stable average value is calculated.

Then, when INT3 = 1 in step 28, step 30
Stable data / 16 is displayed at, and when INT3 = 0, the stable average value is transferred to W ₁ , W ₂ and W ₃ of W register,
INT3 is set to 1. On the other hand, in decision step 25
When INT3 = 1, in step 31, when C ₂ = 4, in step 32, calculation of the third-order digital filter Is executed, the contents of the W register are shifted by one digit at the same time, and the counter C ₂ is cleared. This digital filter also samples four pieces of data from the previous time series.

When the time series is stable up to {Wn}, the process proceeds from decision step 33 to 34, 35, and the total average value of the W register multiplied by 1 / K = 1/16 is displayed. When the time series {Zn} is stable but the time series {Wn} is not yet stable, the process proceeds to the judgment step 26 to 27, 30 and the value based on the total average value of the Z register is displayed. When only the series {Yn} is stable, the process similarly proceeds from decision step 17 to 19, 22 and the value based on the total average value of the Y register is displayed.

FIG. 4 shows frequency characteristics of noise attenuation rate (dB) in time series {Zn} in this embodiment. What should be noted in this characteristic diagram is that, firstly, the amount of attenuation in the intermediate band excluding the vicinity of frequencies 0 and f ₀ is very large at around −50 dB. To a 2, f _0/16 output for each zero frequency points exist that fall into (attenuation -∞dB), therefore, is that the attenuation amount of the frequency near 0 is much larger than before.

<Effects of the Invention> According to the present invention, the response speed of the display is high and the noise at the ultra-low frequency can be greatly attenuated with a simple structure as compared with the conventional averaging process. Therefore, it is possible to easily remove the flickering of the display due to the wind and the slow swell of the display, which have been extremely difficult to remove.

Further, since it is confirmed that the output data of the previous digital filter is stable and the processing of the subsequent digital filter is started, the discontinuity of the display does not occur, and the characteristic of removing the noise component is further improved.

[Brief description of drawings]

FIG. 1 is a time chart of time series data according to the present invention. FIG. 2 is a block diagram showing the overall configuration of the embodiment of the present invention, and FIG. 3 (divided into a and b sheets) is a flow chart of the program of the embodiment of the present invention. FIG. 4 is a frequency characteristic diagram of the attenuation rate in the above embodiment.

Claims (2)

[Claims] 1. A balance for converting a load on a weighing pan into an electric signal and displaying the load value by time series data obtained by sampling the electric signal at every predetermined time (Tx). Based on the first-order time-series data {Xn} obtained by sampling for each Tx), the cycle (Ty) is the above-mentioned predetermined time (Tx) based on the operation content represented by the following formula.
A second digital time series data {Yn} which is an integer multiple of the second digital time series data {Yn} and its second time series data {Yn}.
The second digital filter means for obtaining the third-order time series data {Zn} whose cycle (Tz) is an integer multiple of the cycle (Ty) from n} and the second-order time series data {Yn} are predetermined. The electronic balance is provided with a control means for determining whether or not the stable width has fallen within the stable width and for starting the calculation of the second digital filter means after the stable width falls within the stable width. (However, M is a positive integer and M ≧ 2) 2. At least one digital device that obtains time-series data having a cycle that is an integral multiple of the cycle of the time-series data from the time-series data obtained by the digital filter means in the preceding stage after the second digital filter means. The filter means is provided, and the processing result by a digital filter having a large noise attenuation effect among these digital filter means is configured to be preferentially displayed, and the digital filter means is preferentially displayed. Electronic balance described.