JPH06317459A - Low frequency attenuator for digital metric register - Google Patents

Abstract

PURPOSE:To remove low frequency oscillatory components by determining a metric signal of a differentitation signal, from an FIR filter for differentiating the metric signal, corresponding to a peak value and then averaging the metric signals corresponding to even number of adjacent points of inflection of the differentiation signal. CONSTITUTION:A metric signal from a load cell 10 is passed through an LPF 12, an A/D converter 14, and a digital filter 16 for removing the low frequency components having high amplitude and then the signal is fed to a low frequency attenuator 18. In this regard, the differentiation value DELTAa(k) of a metric signal a (k) is determined through an FIR operation using a smoothed differentiation weight factor and then the value of P is determined according to formula I. If P<0, DELTAa(k-1) can be regarded as an extreme point of the differentiation signal and a {k-1-(N-1)/2} is represented as a value from which the low frequency oscillatory components are removed. N is an odd number. If P>=0, P1 is determined according to formula II and if P1<0, the value at that point is decided as a node. A corresponding metric signal and the average value of a {k-(N-1)/2} are then determined and indicated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for attenuating low frequency noise contained in a weighing signal, and more particularly to a device for attenuating a digital weighing signal.

[0002]

2. Description of the Related Art Generally, a weighing signal obtained by, for example, a load cell or the like may be affected by a disturbance such as floor vibration and may include a low frequency vibration component. In order to remove such a low frequency vibration component, an analog or digital low pass filter has been conventionally used.

[0003]

However, the low-frequency vibration component contained in the weighing signal includes a component having a small amplitude, for example, of about 5 Hz, which is to be removed by the low-pass filter described above. , A large phase delay occurs. Therefore, in a weighing machine that requires a long time for the filter output to reach a steady state and high performance (the number of items that can be weighed per unit time can be increased), such low-frequency vibration components are practically used. Is very difficult to remove, and the reality is that it is left alone.

However, the accuracy is extremely lowered only by the existence of such a low-frequency vibration component having a small amplitude.
Assuming that the true value of the weighing signal is P and the amplitude (maximum amplitude) of the low-frequency vibration component at the apex and the valley point is d, the accuracy of the weighing signal including the low-frequency vibration component is represented by ± P / d. . Even if the low frequency vibration component is reduced to 1/3, the accuracy is tripled.

An object of the present invention is to reduce the response delay and remove low frequency vibration components.

[0006]

To achieve the above object, the present invention provides a FIR filter using a smoothing differential weighting coefficient for differentiating a digital weighing signal obtained by digitizing a weighing signal containing low frequency noise. And first detecting means for sequentially detecting the peak value of the differential signal from the filter, and first output means for outputting the digital weighing signal corresponding to each peak value detected by the first detecting means. And second detection means for sequentially detecting the inflection points of the differential signal from the filter, and the digital weighing signal corresponding to an even number of adjacent inflection points detected by the detection means. The averaging means for averaging and the output means for outputting the output signal of the averaging means are provided. The invention may also be that the digital metering signal supplied to the filter has already been filtered by a low pass filter.

[0007]

According to the present invention, the filter means successively generate the differential signal of the digital weighing signal. The peak value of the sequentially generated differential signals is detected by the first detecting means. When the differential signal has a peak value, the digital weighing signal corresponding to the peak value of the differential signal is
A node that changes from positive to negative or from negative to positive. If the low frequency vibration component can be regarded as a sine wave, the digital weighing signal at the node is substantially a true value. Therefore, the low frequency vibration component can be substantially removed by outputting the digital weighing signal at this contact point by the first output means. Even if the low-frequency vibration component is not a sine wave, if it has already been filtered by another low-pass filter, the digital weighing signal at the node is regarded as a true value. it can.

Further, according to the present invention, the inflection points of the differential signals sequentially generated from the filter means are sequentially detected by the second detecting means. The digital weighing signal corresponding to each even number of adjacent inflection points in the differential signal corresponds to the apex or valley of the digital weighing signal when the low frequency vibration component can be regarded as a sine wave. Therefore, since each digital weighing signal corresponding to an even number of adjacent inflection points corresponds to a vertex and a valley point, by averaging them, this average value can be regarded as a true value. Therefore, by outputting this average value, the low frequency vibration component can be removed. Even in this case, even if the low-frequency vibration component is not a sine wave,
Already filtered by another lowpass filter,
Since the digital weighing signals at the vertices and the valleys are regarded as true values, the low frequency vibration component can be removed as in the above.

Further, in the present invention, the filter used to obtain the differential signal is F using the smoothed differential weighting coefficient.
Since it is an IR filter, there is little phase delay and a high-speed response can be obtained.

[0010]

EXAMPLE This example has a load cell 10 as shown in FIG. 2, and the weighing signal generated from this load cell 10 contains a low frequency vibration component. In order to remove unnecessary high frequency components when digitizing the weighing signal, the weighing signal is supplied to the analog filter 12 which is a low-pass filter. This analog filter 12
The output signal of is converted into a digital weighing signal by the analog / digital converter 14. This digital weighing signal is fed to a digital filter 16, where low-frequency signal components of large amplitude are removed and fed to a low-frequency attenuator 18 according to the invention. This low frequency attenuator 18
A low-frequency component having a small amplitude is attenuated by, and is displayed on the display device 20, for example. The output signal of the low-frequency attenuator 18 may be arithmetically processed to determine whether the item to be weighed has a desired weight or more.

The low frequency attenuator 18 can be composed of, for example, a microcomputer or a DSP (digital signal processing device). The processing performed by the low frequency attenuator 18 will be described with reference to the flowchart shown in FIG.

First, the filtered digital weighing signal a (k) is obtained from the digital filter 16 (step S).
2). Next, the differential value Δa (k) of this digital weighing signal a (k) is calculated (step S4).

The calculation of the differential value Δa (k) is performed by the FIR filter calculation using the smoothed differential weighting coefficient. The transfer function D (z) obtained by this filter calculation is
Represented by

[Equation 1] D (z) = (N-1) + (N-1) / 2Z ^-1 + ・ ・ ・ +0 ・ Z- ^{(N-1) / 2} -1 ・ Z ^{-((N-1 ) / 2 + 1)} -2. ^{((N-1) / 2 + 2)} -... (N-1) Z- ^{(N-1) The} tap coefficient N is an odd number.

For example, when N = 7, D (Z) is represented by the equation 2.

[Equation 2] D (Z) = 3 + 2Z ^-1 + Z ^-2 -Z ^-4 -2Z ^-5 -3Z ^-6 The delay of this filter is (π / 2
−ωT (N−1) / 2), and in the case of Equation 2, π / 2−
It becomes 3ωT. However, ω is an angular frequency and T is a sampling period.

FIG. 3A shows the digital weighing signals sequentially acquired in step S2, that is, the digital weighing signal a (k) from which the noise component having a large amplitude is removed by the digital filter 16, and FIG. Step S
Differential signal Δa (k) calculated in accordance with Equation 2 in 4
Indicates. Note that in FIG. 3A, the digital filter 1
6, the remaining low-amplitude low-frequency vibration component from which the large-amplitude noise component has been removed is represented by a sine wave, and this low-frequency vibration component is represented by sampling at 12 times the sampling frequency.

Next, the vertices or valleys of the differential signal Δa (k) are searched for based on steps S6 and S8. That is, the arithmetic operation of Equation 3 is performed in step S6.

## EQU00003 ## P = {Δa (k)-Δa (k-1)} × {Δa (k-1)-Δa (k-2)}

Then, in step S8, it is determined whether this P is negative. For example, reference numeral X1 in FIG. 3B indicates a valley point of the differential signal Δa (k). The value of this valley point is Δa (k−1), and the previous value thereof is Δa (k). −
2) If the next value is Δa (k), then {Δa (k) -Δ
Since a (k-1)} is a positive value and {Δa (k-1) -Δa (k-2)} is a negative value, P is a negative value. Similarly, if the value of the vertex X2 is Δa (k-1), the value immediately before that is Δa (k-2), and the value after that is Δa (k), then {Δa (k) -Δa ( k-1)} is a negative value, and {Δa (k-1) -Δa (k-2)} is a positive value. Therefore,
P becomes a negative value. Thus, when P becomes a negative value, Δ
It can be determined that a (k-1) is a valley point or a vertex of the differential signal.

When it is determined that the point is a vertex or a valley point (Yes in step S8), Δa (k-1- (N-
The value 1) / 2) is set to the value V from which the low-frequency vibration component is removed (step S10) and displayed (step S12).

Since the value of P becomes negative as described above, it is known that Δa (k-1) is the valley point or the peak of the differential signal. The valley points or vertices of the differential signal correspond to the nodes of the digital weighing signal a (k).

However, as described above, when the differentiation is performed according to the equation 1, a delay of π / 2-ωT (N-1) / 2 occurs between the original low frequency vibration component expressed as a sine wave. ing. In other words, ωT (N
There is a delay of -1) / 2. Considering this delay,
It outputs a (k-1- (N-1) / 2). For example, Y1 shown in FIG. 3A is displayed as V corresponding to the node X1 of the differential signal.

The values of the nodes of the digital weighing signal are
The value V obtained by removing the low-frequency vibration component is shown in FIG.
As is clear from the value of the node Y1 of, if the digital weighing signal a (k) is a sine wave as shown in FIG. 3 (a), the value of the node becomes a value very close to the true value of the digital weighing signal. Because it will be. If the low-frequency vibration component is not a sine wave, but is sufficiently filtered by the digital filter 16, the value of the node can be regarded as the metric true value.

Subsequent to the case where it is judged in step S8 that P is not negative, that is, when it is judged that it is not the peak or trough of the differential signal, the node of the differential signal is searched (steps S14 and S16). That is, P1 is calculated based on Equation 4 (step S14).

[Formula 4] P1 = Δa (k) × Δa (k-1)

For example, in FIG. 3 (b), reference numeral X3 represents a node of the differential signal, where the value at this point is Δa (k) and the previous value is Δa (k-1). P1 obtained by multiplying these is a negative value because it is a multiplication of a positive value Δa (k-1) and a negative value Δa (k). That is, Δa (k) and Δa (k-
Since the sign of 1) is opposite, P1 has a negative value.
This also applies to other nodes, for example, the node X4. Therefore, it is determined in step S16 whether the value of P1 is negative, and if negative, it is known that the value at that point is a node.

If the value of P1 is negative, first, the digital weighing signal Z corresponding to the current node value is stored as the value Z1 of the digital weighing signal corresponding to the previous node, and a
(K- (N-1) / 2) is stored as the value Z of the digital weighing signal corresponding to the current node, and the average value V of Z and Z1 is obtained (step S18).

Here, a (k- (N-1) / 2) is stored as the value Z of the digital weighing signal corresponding to the current node, as described above, where Δa (k) and a (k) are stored. This is because there is a (N-1) T / 2 phase delay between and. Therefore, in the case of the value X3 of the node in FIG. 3B, Y3 of FIG. 3A is stored as Z as a (k) corresponding to this, and in the case of the value X4 of the node, a corresponding to this is a.
As (k), Y4 in FIG. 3A is stored as Z.

The nodes of the differential signal correspond to the vertices or valleys of the digital weighing signal, as is clear from the comparison of X3, X4, Y3 and Y4 in FIGS. 3 (a) and 3 (b). Then, as shown in FIG. 3A, when the low-frequency vibration component is a sine wave, when the average of the peak value and the valley value of the digital weighing signal is calculated, the average value is extremely close to the true weighing value. It is a value close to. Even if the low-frequency vibration component is not a sine wave, if it is sufficiently filtered by the digital filter 16, the average value of the peak value and the valley value of the digital weighing signal is the true weighing value. Can be regarded as

Therefore, the digital weighing signal corresponding to the node of the previous differential signal is stored as Z1, and the average V of this and the digital weighing signal Z corresponding to the node of the current differential signal is obtained. When V is calculated in this way,
This is displayed (step S12). Then, the value of the next digital weighing signal a (k) is input and the same processing as above is performed, so the value of k is advanced by 1 (step S2).
2) and returns to step S2.

There are various techniques for differentiating the digital weighing signal. For example, one method is to obtain the difference between the respective digital weighing signals. But in case of difference,
Since the difference value is small in the vicinity of the vertices and valleys of the digital weighing signal, there is a problem that the vertices and valleys are likely to be erroneously discriminated due to the influence of slight external noise.

Further, the data string of each digital weighing signal a (k) is approximated by a function and the equation is analytically differentiated,
There is a method of obtaining a differential signal by subtracting the value of the equation, but such a technique requires a huge amount of calculation, which is not preferable. On the other hand, in the calculation using the smoothing differential weighting coefficient used in the present invention, the calculation amount is small and a relatively reliable result can be output.

In the above embodiment, the simple average of the values of the vertices and troughs of the digital weighing signal was calculated and output as the true value of the digital weighing signal, which has already been filtered by the digital filter 16. Therefore, the latest value is considered to be a value close to the metric true value. For example, when the filtering process is not performed by the digital filter 16, the values of two or more vertices and these vertices are determined. A simple average of the values of the same number of valley points may be obtained, or a root mean square value may be obtained.

[0031]

As described above, according to the present invention, the peak value and the inflection point of the differential signal generated by the filter means are respectively detected, and the node of the digital weighing signal is obtained by the peak value of the differential signal. The digital weighing signal of is output as a substantially true value. Further, since the vertex or trough of the digital weighing signal is obtained from the inflection point of the differential signal, the true value is output by averaging each digital weighing signal corresponding to an even number of adjacent inflection points. There is. In this way, since the inflection point and the peak value of the differential signal are respectively detected, the true value can be quickly obtained.

For example, when it is attempted to find the true value by detecting only the vertices and valley points of the digital weighing signal, for example, even if a vertex is found, until the next valley point is found,
The true value cannot be output, and in order to output the true value, at least 1/2 cycle of the low frequency vibration component must be waited. Even when only the nodes of the digital weighing signal are obtained, the true value of the digital weighing signal is obtained first, and the true value is obtained next is the next node that comes after 1/2 cycle of the low-frequency vibration component. It is when it is detected.

On the other hand, in the present invention, since the detection of the node of the digital weighing signal and the detection of the apex and the valley are used together, for example, the true value is output by the detection of the node, and then the 1/4 cycle is obtained. After that, the true value based on the detection of the peak value can be output, and a high-speed response can be obtained.

Moreover, since the digital filter using the smoothed differential weighting coefficient is used to detect the nodes, vertices and troughs of the digital metric signal, the differential signal can be obtained even if the calculation amount is small. Therefore, high-speed processing can be performed. Further, even when the number of taps is adjusted according to the property of the signal, the phase delay can be calculated, so that it is possible to determine almost correct nodes, vertices and troughs.

[Brief description of drawings]

FIG. 1 is a flowchart of a low frequency attenuator embodying the present invention.

FIG. 2 is a block diagram of a weighing device using the low frequency attenuator of the same embodiment.

FIG. 3 is an input / output waveform diagram of the low frequency attenuator of the same example.

[Description of Symbols] Step S4 Calculation of Δa (k) (differentiation means) Step S6 S8 Peak value detection means of differential signal Step S10 Output means of digital weighing signal corresponding to peak value of differential signal Step S14 S16 Node of differential signal Means for averaging and outputting digital weighing signals corresponding to the nodes of the differential signal

Claims (2)

[Claims] 1. An FIR filter using a smoothing differential weighting coefficient for differentiating a digital weighing signal obtained by digitizing a weighing signal containing low-frequency noise, and means for sequentially detecting peak values of differential signals from the filter. First output means for outputting the digital weighing signal corresponding to each peak value detected by the detection means, and detection means for sequentially detecting inflection points of the differential signal from the filter,
An averaging means for averaging the digital weighing signals corresponding to the even number of adjacent inflection points detected by the detecting means, and an output means for outputting an output signal of the averaging means are provided. Low frequency attenuator for digital weighing equipment. 2. The low-frequency attenuator for a digital weighing device according to claim 1, wherein the digital weighing signal supplied to the filter is already filtered by a low-pass filter. Attenuator.