JPH0353814B2 - - Google Patents

Description

[Detailed description of the invention] [Technical field to which the invention pertains]

The present invention relates to a digital arithmetic correction method for a filter circuit in which a waveform delayed by a filter provided in an analog signal input circuit is converted from analog to digital and then compensated.

[Prior art and its problems]

Generally, when quantizing an analog input signal,
In order to remove the effects of harmonics contained in this input signal, it is said to be desirable to install a low-pass filter whose center frequency is less than half the sampling frequency. In addition, when detecting the effective value of an AC signal such as a commercial frequency, a square calculation circuit or a combination of a full-wave clear current circuit and a smoothing circuit is used.
A low pass filter is also used in this case to remove ripples from the waveform. FIG. 6 is a block diagram showing a conventional example of an analog input section of a digital arithmetic unit. An analog signal is input from an input terminal 1, and an analog filter 2 provided for each terminal removes noise and high frequencies. After high-frequency components are removed, the signal passes through a multiplexer 7 and a sample-and-hold circuit 8, is converted into a digital quantity by an analog-to-digital converter 4, and is further input to a digital arithmetic unit 10. The aforementioned low-pass filter is used for this analog filter 2, and its center frequency is set low, but since the time constant of the analog filter 2 is large, unnecessary high frequency components such as high frequencies are removed. At the same time, there will be a time delay in the transient response of the fundamental wave component that is originally desired to be detected, so if an analog signal inputted after passing through the analog filter 2 is used, there will be problems such as a loss of system response. Therefore, conventionally, when determining the center frequency of this low-pass filter, the balance between the frequency band to be removed and the response time of the system determined by the filter time constant at that time is considered. A filter alone cannot simultaneously satisfy the two contradictory conditions of having a high-speed response to the fundamental wave component and removing high frequencies. On the other hand, in the field of digital measurement and control, digital filters that use quantized data instead of analog filters have been realized, but they have the same problems as analog filters in terms of balance between center frequency and response. In addition, when detecting the AC effective value, it is possible to calculate it by digital calculation from the sample data of the AC waveform without using analog circuits, but this requires high-speed sampling and calculation, so other control It becomes difficult to balance this with calculation. In particular, in systems where the effective value or average value of an AC voltage waveform is detected using a full-wave rectifier circuit or an effective value calculation circuit and a filter, and PID control is performed using the quantized data,
The time constant of the analog filter has a large effect on the transient characteristics of the system. This is because it is necessary to increase the filter time constant in order to remove ripples from the waveform. As mentioned above, in the conventional technology, when attempting to remove high frequency components by installing an analog filter on the analog signal input to the digital system, this filter function simultaneously delays the system response, and there is no effective way to solve this problem. The problem is that there is no means to do so.

[Purpose of the invention]

The purpose of this is to take advantage of the inherent function of analog filters, which is to remove high frequency components, while also making digital measurements possible.
It is an object of the present invention to provide a digital calculation correction method for a filter circuit that can improve the response and transient characteristics of a control device.

[Key points of the invention]

This invention attempts to compensate for the delay element included in an analog filter provided in an analog signal input circuit by using sample data obtained by quantizing this analog signal. The purpose is to compensate for the time delay caused by this analog filter by determining the transfer function of this analog filter and performing digital calculations on quantized sample data based on the inverse function of this transfer function.

[Embodiments of the invention]

FIG. 1 is a block diagram showing the principle of the present invention, and based on this FIG. 1, the principle of the present invention will first be explained below. In FIG. 1, an analog signal input to an input terminal 1 passes through an analog filter 2 and appears at a filter output terminal 3. This filter output signal is converted into a digital quantity by an analog-to-digital converter 4, subjected to correction calculation processing in a correction calculation circuit 5, and then taken out from a correction output terminal 6. The signals and transformations in the configuration shown in FIG. 1 are expressed by a Laplace function and a transfer function using the Laplace function as follows. V _i (s)... Analog input signal to input terminal 1 V _a (s)... Analog filter output signal V _p (s) appearing at filter output terminal 3... Corrected output signal F (s) appearing at correction output terminal 6... Transfer function D(s) of analog filter 2...Transfer function of correction calculation circuit 5 Using the above representation, the following relational expression can be obtained. V _a (s)=F(s)・V _i (s) ...(1) V _p (s)=D(s)・V _a s = D(s)・[F(s)・V _i ( s)] ...(2) The data output from the analog-to-digital converter 4 and input to the digital arithmetic unit uses the analog filter output signal V _a (s) instead of the true input signal V _i (s). Since we are observing, normally this V _a
(s) is used to perform digital calculations,
As already described, this V _a (s) includes the transfer function F(s) of the analog filter as shown in equation (1), which is disadvantageous. Therefore, a correction calculation circuit 5 is provided to correct the signal V _a (s) input to this digital calculation device, and its transfer function D
By performing the calculation (s), a corrected output signal V _p (s) shown by equation (2) is obtained. Here D(s)
If the transfer function is selected so that it has the relationship shown in equation (3) below, then from equation (2) and (3) D(s)・F(s)=1...(3), we get the following: Obtain equation (4). V _p (s)=V _i (s) (4) In other words, the analog filter 2 is theoretically removed. In the present invention, the transfer function D of the correction calculation circuit 5
By appropriately selecting (s), the fundamental wave component of the analog input signal V _i (s) is reproduced,
Compensate for the delay element of analog filter 2. Considering the above in terms of transient response in the time domain, the analog filter output f _a
(t) becomes the following equation (5) from equation (1), but in the transient response f _a (t) = 1/2πj∫〓 ^+∞ 〓 _-∞ F(s)・V _i (s)・
e ^-st dt (5) It can be seen from equation (5) that the transfer function F(s) of analog filter 2 distorts the filter output. On the other hand, the transfer function D which has the relationship of equation (3)
Corrected output by applying (s) to the filter output
As f _p (t), the following equation (6) is obtained from equation (2). f _p (t)=1/2πj∫〓 ^+∞ 〓 _-∞ D(s)・F
(s)・V _i (s)・e ^-st dt=1/2πj∫〓 ^+∞ 〓 _-∞ V _i (
s)・e ^-st dt=f _i (t)...(6) From this equation (6), it is clear that the correction output f _p (t)
is equal to the input f _i (t) even in the transient response. That is, the delay element of the analog filter 2 can be eliminated. Since the transfer function F(s) of the analog filter 2 can be known in advance, the transfer function (hereinafter abbreviated as correction function) D(s) of the correction calculation circuit 5 can also be made clear. However, this correction function D
(s), if the correction value f _i (t) is reproduced in perfect agreement with the filter input value f _i (t) according to the theoretical formula, then the purpose of installing the analog filter 2 should be removed. Since harmonics and ripples are also emphasized, this correction function D(s) must be selected with sufficient consideration of its effects. Next, a method for realizing the correction function D(s) by digital calculation will be described below. Generally, the transfer function F(s) of analog filter 2 can be expressed as a rational function of s as shown in equation (7) below, and the higher the order of s, the sharper the cutoff F(s)=b _p・s ^m +b ₁・s ^m-1 +b ₂・
s ^m-2 ＋…＋b _n ／s ⁿ ＋a ₁・s ^n-1 ＋a ₂・s ^n-1 ＋…＋a _o ……(7
) cutting characteristics can be obtained. Here, a ₁ , a ₂ ... in equation (7)
a _o and b ₀ , b ₁ , b ₂ ...b _n are constants, and s is s=
j・2π・f=jω. The correction function D(s) is also a rational function of s and can be expressed by the following equation (8). D(s)=A ₀・S ⁿ +A ₁・S ^n-1 +A ₂・S ^n-2
+…+A _o /S ^m +B ₁・S ^m-1 +B ₂・S ^m-2 …+B _n …(8) Here, A ₀ , A ₁ , A ₂ …A _o and B ₁ , B ₂ …B _n is a constant. When using these equations (7) and (8), if the correction function D(s) is of a high order, it will be difficult to operate the above-mentioned equation (2), but some examples of the operation will be described. (b) When the constants B ₁ , B ₂ ...B _n are zero: In this case, equation (8) can be rewritten as equation (9) below. this
Equation (10) is obtained from equations (9) and (2). D(s)=A ₀・S ^nm +A ₁・S ^nm-1 +A ₂・S ^nm-2 +…+A _o・S ^-m ……(9) V _p (s)=D(s)・[F (s)・V _i (s)〕＝A ₀・S ^nm・F _a (s)
+A ₁・S ^nm-1・F _a (s) +…+A _o・S ^-m・F _a (s)……(
10) In general, the form S ⁱ・F _a (s) or S ^-i・F _a (s) means the differentiation or integration of the function f _a (t), so this equation (10) is the filter output f _a (t)
This means that it is the sum of the products obtained by performing differentiation and integration on , and this is based on the Laplace transform shown in equations (11) to (13) below. S ^-n・F(s)→∫〓 ₀ dτ ₁ ∫〓 ¹ ₀ dτ ₂ …∫〓 ^n-1 ₀ f(τ _o )・dτ _o ……(11) S ⁿ・F(s) + _{o- 1} 〓 ^k=0 S ^nk-1・f ^(k) (0 ₊ )→d ⁿ /dt ⁿ f(t) ...(12) S ⁿ・F(s)→d ⁿ /dt ⁿ f(t) − _o-1 〓 〓 ^k=0 f ^(k) (0 ₊ )・1/(n−k−1)!・t ^(nk-2) ……
(13) Therefore, if the correction function D(s) can be expressed by the polynomial of equation (9), it is corrected by performing differential and integral operations on the filter output f _a (t) according to equations (10) to (13). can be realized. (b) When the correction function D(s) can be decomposed into partial fractions: In this case, equation (8) can be expressed as equation (14) below. D(s)=H ₁ (s)/S-α ₁ +H ₂ (s)/S-α ₂ +...
+H _n (s)/S-α _n ...(14) Here, α ₁ , α ₂ ...α _n are constants, and H ₁ (s), H ₂
(s),...H _n (s) is a polynomial in S. The following equation (15) is obtained from equation (14) and equation (2). V _p (s)=D(s)・[F(s)・V _i (s)]
= _n 〓 〓 ⁱ⁼¹ H _i (s)/S−α _i [F(s)・V _i (s)]……(15
) In this equation (15), A _i (s)・[F(s)・V _i
(s)] is a polynomial in s, so A _i (s) is a polynomial in s, so
The above-mentioned example (a) shows that the value can be obtained by performing differential integration on the filter output f _i (s). Therefore, equation (18) is obtained using equations (16) and (17) below. A _i (s)・[F(s)・V _i (s)] Inverse Laplace transform――――――――→ f _b (t)
……(16) F(s)・G(s) inverse Laplace transform――――――――→ ∫ ^t _p f(t−τ)・g(τ)dτ ……(17) _n 〓 ^{i= 1} H _i /S−α _i [F(s)・V _i (s)]→ _n 〓 ⁱ⁼¹ ∫ ^t _p e〓

Claims (1)

[Claims] 1 Input the analog signal that has passed the filter,
A digital calculation correction method for a filter circuit, characterized in that in a system for quantizing and processing sample data, a digital calculation is performed on the quantized sample data based on an inverse function of the transfer function of the filter.