JPH0654241B2 - Data processing device - Google Patents

Description

The present invention relates to a data processing device for correcting input data such as measured values and signal data.

[Conventional technology]

Conventionally, there is a measuring device that converts a measurement result into a digital value by an A / D converter and outputs the digital value. In such a measuring instrument, noise is generated due to the influence of the surroundings, and a slight variation occurs in the digital value. For example, when measuring the thickness of the sheet in microns, the variation of the digital value due to noise is removed. It is needed.

Therefore, the conventional measuring device stores the measurement results sequentially sampled in consideration of the influence of the noise.
An average of the measurement result sampled at a certain time point T and a predetermined number of measurement results sampled before the sampling time point T is obtained by the data processing device, and this average value is used as a correction value for the measurement result at the sampling time point T.

However, if the number of measurement results using the average calculation is increased in order to improve the accuracy of the measurement data, there is a problem to be solved that the capacity of the memory for storing the measurement results in the data processing device increases significantly. . This point will be described in detail.

In the conventional data processing device, the correction value at a certain time is calculated by the following equation.

here Is the measurement result after correction, x _n is the measurement result sampled at the nth time, N is the number of the measurement results (pickup number) used for the average calculation process, which is fixedly set and the data from the nNth time to the nth time is used. To be

The data processing device has N memories for storing the measurement result and in the equation (1). That is, N at the time of sampling of the previous measurement result
A memory is provided for storing the total of the individual measurement results.

The calculation processing procedure of the equation (1) using these memories will be specifically described with reference to FIGS. 4 and 5.

Here, in the equation (1), the number N of pickups is set to “3”, and the memories M1 to M3 are used as memory for storing the measurement result x _n.
M4 is the total value of the measurement results , And the memory M5 is further allocated to store the measurement result x _nN .

As shown in FIG. 4, when the first measurement result x ₁ is obtained, the data processing device stores the measurement result x ₁ in the memory M1. Since "3" measurement results cannot be obtained at this time, "0" is stored in the memories M2 to M5. Next, when the second measurement result x ₂ is obtained, the data processing device shifts the storage content of the memory M2 to the memory M3 and the storage content x ₁ of the memory M1 to the memory M2, and the memory M1 stores the same. Second
Memorize the _second measurement result x ₂ . In this way, the data processing device sequentially shifts and stores the obtained measurement results in the memories M1 to M3. When the second measurement result is obtained, the data processing device And stores x ₁ in the memory M4. Hereinafter, hereinafter, the data processing device shifts and stores the obtained measurement results in the memories M1 to M3. When the fourth measurement result x ₄ is obtained, the memory M1
_{~M3 x 4, x 3, x} 2 , respectively are stored in this x ₁ which has been stored in the memory M3 until is shifted stored in the memory M5 as the value of x _nN. Further, the total value of x ₁ + x ₂ + x ₃ is stored in the memory M4. Since x ₁ is stored as the value of x _nN , the data processing device calculates (value of memory M1 +
The value of the memory M4−the value of the memory M5) / 3 is calculated. This operation result Is output from the data processing device as a correction value for the _fourth measurement value x ₄ . Hereinafter, the data processing device shift-stores the values of the memories M1 to M3, the memory M4, and the memory M5 each time a new measurement result is obtained, and performs arithmetic processing based on the values of these memories.

Such a correction calculation is performed every time a new measurement result is obtained.
This is also called a moving average because the data used for the average calculation moves one after another as shown in FIG.

[Problems to be Solved by the Invention]

When the moving average described above is obtained by the data processing device, it is necessary to prepare the memory for the number of measurement results used for the moving average. In particular, in order to improve the accuracy of the output data, the specified average number (pickup number If 125), 256 times, 512 times, etc. are specified for a large number of times, the number of data memories for storing the measurement results to be averaged according to the specified average times also becomes large.

As a result, the data processing time increases as the data memory increases. For example, when the data sampling speed is 1 ms or less and 512 moving averages are performed, the moving average processing time for one time is about 0.5 seconds. When measuring the thickness and width of continuously flowing sheet material, specify the specified number of averages.
It is said that the data accuracy may be insufficient even if it is 512 times.

By the way, moving average of 512 2-byte data requires about 1 Kbyte of memory. When attempting to correct the measurement data using a 1-chip central processing unit (CPU) to reduce the size of the measuring instrument, the built-in memory of the 1-chip CPU currently on the market lacks the emery capacity. , It is necessary to add memory to the outside, which leads to a problem that the measuring device cannot be downsized.

In addition, the more memory the CPU has to access, the longer the computation time, resulting in a problem that the data sampling speed is limited.

Therefore, an object of the present invention is to solve the above-described problems, reduce the number of data memories to be prepared when performing moving average correction of measurement results, and obtain an approximate moving average value at high speed. It is to provide a data processing device.

[Means for Solving the Problems]

In order to achieve such an object, according to the present invention, a predetermined value N, which is an integer of 2 or more, and a variable n are set as n> N, and the nN-th input data sampled and input up to the previous time is input.
A moving average value obtained by performing noise correction on the input data x _n sampled at the n-th time using the _n-th input data from x _nN and _N input data of x _n-1 In the data processor that obtains the moving average value for the previous n-1th input data x _n-1 Then, Is used to calculate the moving average for the _nth input data x _n I asked for this time The value of will be calculated next time By substituting into the above equation as It is characterized by comprising a calculation means for sequentially calculating and outputting.

[Work]

The present invention focuses on the fact that there is no large error in the data values sampled at high speed, and when obtaining the moving average value, the latest input data value x _n is the total value of the input data used for the moving average sampled so far. In addition, since the moving average value calculated last time is subtracted to perform the moving average correction calculation, it is not necessary to store all the sampling data used for the moving average, and correction can be made compared to conventional data processing devices. Arithmetic can significantly reduce the number of memories used.

〔Example〕

 Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 1 shows a concrete configuration of the embodiment of the present invention.

In FIG. 1, 200 is a predetermined value N that is an integer of 2 or more and a variable n is n> N, and the nNth input data x _nN sampled and input from the input device 100 to the previous time is input.
N-1-th input data from, x _n-1 of N this time by using the input data n th the sampled input data x _n
Noise-corrected moving average value Is a data processing device that seeks.

300 is the moving average value for the previous n-1th input data x _n-1 Then, Using, the moving average value for the _nth input data x _n I asked for this time The value of will be calculated next time By substituting into the above equation as Is a calculation means for sequentially calculating and outputting.

FIG. 2 shows a concrete circuit configuration of the embodiment of the present invention. In FIG. 2, reference numeral 1 is a sensor for measuring an object to be measured, for example, measuring the thickness of a sheet, and outputting the measurement result as an analog signal. Reference numeral 2 is an amplifier that amplifies the analog signal obtained from the sensor 1 to a level at which it can be post-processed. Reference numeral 4 is an analog-digital (A / D) converter for converting the amplified signal obtained from the amplifier 2 from an analog value to a digital value.

A central processing unit (CPU) 4 executes the programs shown in FIGS. 3 (A) and 3 (B) to correct the measurement data. The A / D converter 3 may be built in the CPU 4.

The CPU4 has a random access memory (RAM) inside which stores the measurement data from the A / D converter 3, and FIG.
It has a read-on memory (ROM) that stores in advance a program for obtaining a moving average shown in (B). The result of the correction calculation by the CPU 4 is externally output as a digital signal as the corrected measurement result.

Further, in the RAM of the CPU 4, memories M11 to M13 for storing the calculation data of the correction calculation executed by the CPU 4 based on the control procedure of FIGS. 3A to 3B, a counter and the like are provided.

Next, the operation of this embodiment will be described with reference to the flow charts of FIGS. 3 (A) to (C). Figure 3 (A) is the CP shown in Figure 2.
The main control procedure regarding the correction processing of the measurement data executed by U4 is shown.

FIG. 3 (B) shows a detailed control procedure of the initial process (step S10) in the main control procedure of FIG. 3 (A). Third
FIG. 3C shows a detailed control procedure of the correction processing of the measurement result in the main control procedure of FIG.

In this example, when n ≦ N, the arithmetic expression for the moving average is When n> N Is used.

In FIG. 3 (A), when the CPU4 is powered on, the CPU4
Executes initialization processing of each memory in the CPU 4 in step S10. The CPU 4 shifts to the sub-procedure shown in FIG. 3 (B) and stores in the RAM the number of data (pickup number) N for the average calculation which is stored in the ROM of the CPU 4 in advance. In this example, this number of data is a fixed value, but if desired to be set arbitrarily, the number N of data may be input from the keyboard.

Next, the CPU 4 sets an initial value "0" in the counter. The counter is a counter used for counting the number of samplings of the sampling data and measuring whether or not the number n of sampling data has reached N, and is provided in the RAM in this example.

In addition, the CPU 4 finishes setting the initial value of the data used for the correction process, and proceeds to the next procedure (steps S11 to S13).

The CPU 4 inputs (data sampling) the measurement result sent from the A / D converter 3 at a constant cycle and performs a correction process for calculating a moving average of the measurement result (step S2
0-S30).

If the nth measurement result x _n is obtained, the CPU 4
Stores the input value x _n in the memory M11 in the RAM (step S31 in FIG. 3 (C)).

Next, the value of the internal counter of the CPU 4 is incremented by "1" to count the number of samplings. When it is determined that the counter value is equal to or less than the number N of sampling data, the arithmetic processing using the above-mentioned equation (3) is performed. When it is determined that the counter has counted the number of samplings n ≧ N + 1, the calculation using the equation (4) is performed (steps S32 to S33).

Steps S34 to S36 show the arithmetic processing steps of the equation (2). CP
U4 reads the total value of the sampling data so far from the memory M12 provided in the RAM, adds the latest data value x _n to the total value, and adds the total value of x _n to the memory.
Re-store in M12. Next, this total value is divided by the counter value n, and arithmetic processing based on the equation (2) is performed. The calculation result is stored in the memory M13 and is used for the correction calculation process of the next sampling data (steps S34 to S36).

On the other hand, when the sampling number n ≧ N + 1 indicated by the counter value, the latest measurement result X _n stored in the memory M11 is the N measurement values stored in the memory 12. Previously calculated moving average value Is subtracted and the result of the subtraction is stored in the memory 12 (steps S40 to S41).

Next, the value in the memory 12 is divided by the N value to calculate the moving average value at the nth time. The calculation result is stored in the memory 13 and used for the next moving average calculation. The moving average value thus calculated is externally output as a correction value for the sampled measured value (steps S42 to 48 → S37).
).

Next, in order to make a comparison with the conventional example, a specific calculation process under the same conditions as the calculation example described in the related art will be briefly described.

Here, the number of data used for the moving average is set to N = 3. When the first measurement result x ₁ (n = 1) is obtained, the measurement result is stored in the memory 11 and the value of the counter is set to “1” (steps S31 → S32).

Next, since the counter value (1) ≤ N (3), N is stored in the memory M12.
As the total value of the individual measurement results Is stored, and the result x ₁ of the moving average calculation is stored in the memory M13 and is output to the outside (steps S33 to S3).
7).

Subsequently, the moving average calculation is sequentially performed from the second time to the third time in the same procedure. At the end of the third moving average calculation, the measurement result x ₃ is stored in the memory M11 and x ₁ + is stored in the memory M12.
x ₂ + x ₃ total value, memory M13 The calculated value of is stored.

When the fourth measured value X 1 is obtained, the procedure is step S3.
1 → S32 → S34 → S40 and the latest measured value X is added to the total value so far, and the previous moving average value Is subtracted and the result of this calculation is Is stored in the memory M12 (steps S40 to S41).

Next, the moving average is calculated, and the moving average value is calculated. Is stored in the memory M13. The fifth and subsequent arithmetic processes are performed in the same procedure. Note that if the calculated value is sampled at high speed, then x ₁ is Is approximately equal to, and the result is the fourth moving average value described in the description of the prior art. It is possible to obtain almost the same value as (steps S42 to S4
8).

In addition, in the moving average processing in this embodiment, one counter memory for storing the number of samplings, one memory for storing the addition result (M12), and one memory (M13) for storing the average value 1
The moving average value can be obtained by preparing only a total of five pieces, one memory for storing the designated number N of average values and one memory (M11) for storing the latest measurement result.

Further, even if the specified number N of average values is several hundred, the number of memories required is five, and it is not necessary to prepare a memory for storing data for the number of average values as in the conventional case. Can significantly reduce the memory capacity of.
Furthermore, since the number of memories accessed by the CPU 4 is small, it is possible to shorten the processing time related to the moving average.
In this embodiment, when the number of sampling times n ≦ N, the above equation (3) is used, but the number N of pickups is used. Well, again It is okay.

In addition, although the present embodiment has described the example in which the measurement sampling data is noise-corrected, high-speed processing such as image processing data,
Needless to say, the present invention can be applied to the case where moving average processing is performed on the majority of data.

〔The invention's effect〕

As described above, according to the present invention, when calculating the moving average value, the latest input data value x _n is added to the total value of the input data found in the moving average sampled so far,
Since the moving average value calculated last time is subtracted to perform the moving average correction calculation, it is not necessary to store all the sampling data used for the moving average, and the correction calculation is used compared to the conventional data processing device. The number of memories to be used can be significantly reduced.

Further, as the number of memories decreases, a one-chip CPU can be used, which contributes to downsizing of the device, and the time for the arithmetic means to access the memory is shortened, and as a result, the arithmetic processing time is also shortened. Since the measurement values can be sampled at higher speed, the accuracy of the measurement values can be improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing the basic constitution of the embodiment of the present invention, FIG. 2 is a block diagram showing the concrete constitution of the embodiment of the present invention, and FIGS. 3 (A) to (C) are CPU4 of the embodiment of the present invention. FIG. 4 is an explanatory view showing a moving average calculation process of a conventional example, and FIG. 5 is an explanatory view showing a moving average calculation of a conventional example. 1 ... Sensor, 2 ... Amplifier, 3 ... A / D converter 3, 4 ... CPU.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Nobuo Hirata Inventor No. 1 Tanabe Shinden, Kawasaki-ku, Kawasaki-shi, Kanagawa Fuji Electric Co., Ltd. (72) Shigekazu Mitani 1 Tanabe Nitta, Kawasaki-ku, Kawasaki-shi, Kanagawa No. 1 within Fuji Electric Co., Ltd.

Claims (2)

[Claims] 1. A predetermined value N, which is an integer of 2 or more, and a variable n, where n> N, and the nNth input data Xn-N to the n-1th input data, which have been sampled and input until the previous time, x. A moving average value obtained by performing noise correction on the input data x _n sampled at the nth time using N input data of _n-1 In the data processor that obtains the moving average value for the previous n-1th input data x _n-1 Then, Using, the moving average value for the _nth input data x _n I asked for this time The value of will be calculated next time By substituting into the above equation as A data processing device comprising a calculation means for sequentially calculating and outputting. 2. The data processing apparatus according to claim 1, wherein the arithmetic means is the moving average value when n ≦ N. As And obtained when n = N The calculated value of A data processing device characterized by substituting when x _{N + 1} input data is processed as an initial value of.