JP3394302B2 - Adaptive filter for digital scale - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [0001] BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a frequency characteristic of a vibration component.
For digital weighing signal input that changes every weighing,
Digital filter adaptive filter to attenuate the vibration component
Related. [0002] 2. Description of the Related Art Conventional adaptive digital filters (hereinafter referred to as
This is called a response filter. ) Is mainly for audio processing and communication
Has been developed. This adaptive filter for voice processing and communication
Is a digital filter whose frequency characteristics change with time.
Signal, such as a frequency band where noise components are present or noise
For input of digital signal whose amplitude of sound component changes with time
On the other hand, filtering is performed according to changes in the frequency characteristics of the noise component.
Filtering by changing the wave transfer function over time
And thereby sufficiently attenuate unwanted noise components
The purpose is to make The adaptive fill
To change the filter's transfer function over time,
How to change the filter coefficient and change the filter order
Can be considered. However, the former filter coefficient is changed.
A further way is to change the filtering transfer function to the desired one
It is possible to change the latter filter order
The method limits the range over which the filter transfer function can be varied.
So that it is not always possible to change to the desired filtering transfer function.
I can't. Therefore, adaptive filters for speech processing and communications
Employs a method of changing the filter coefficient. [0003] As described above, the prior art
The adaptive filter rewrites the filter coefficients over time.
Changes the transfer function of the filter over time.
To complete the identification, i.e., to
Repeatedly to change the wave transfer function and filter
Many calculations are required, and the calculations involve correlation of errors.
Complicated matrix calculations such as matrices are required. As a result,
This adaptive filter is used for filtering by the adaptive filter
Microcomputer and digital signal processing equipment
(Hereinafter referred to as DSP), etc.
Therefore, the filtering processing time becomes long. That
Therefore, this conventional adaptive filter can be used as it is for a digital scale.
Weighing time is long and achieves high-speed weighing
There is a problem that you can not. In addition, the weighing time of a digital scale is reduced as much as possible.
For compression, that is, to perform a little faster filtering
Microcomputers and DSs with high processing capabilities
It is conceivable to use P, etc.
Digital weighing equipment is becoming more expensive due to rising costs of computers and DSPs.
However, there is a problem that the cost is increased. On the other hand, in a conventional adaptive filter, a filter
The coefficients are rewritten in time, but the filter order is
Fixed to a fixed value. This filter order is generally
Noise components with many frequencies and large amplitude noise.
Set to a value that can sufficiently attenuate sound components
Therefore, it is fixed to a relatively large value (high order).
However, the magnitude of this filter order is
Because the filter order is proportional, the larger the filter order, the longer the filtering time
Becomes longer. That is, a general digital weighing signal includes:
It has only a relatively limited frequency and a relatively small amplitude
Some include vibration components.
Using a conventional adaptive filter on the signal for a digital scale,
Filtering takes an excessive amount of time.
When the weighing speed of the digital scale slows down due to extra time
There is a problem. According to the present invention, the frequency characteristic of the vibration component is
Digital weighing signal input that may fluctuate
To attenuate the vibration component,
Update the filter order accordingly to change the filter transfer function.
To improve the weighing accuracy of the digital scale.
Adaptable filter for digital weighing that enables high-speed weighing
The purpose is to provide data. [0008] According to the present invention, there is provided a digital weighing apparatus suitable for a digital weighing machine.
The response filter is set in advance for each predetermined weighing cycle.
Initial number setting means for setting the number of lattice structures to the initial number
And the steady-state gain is 1 and each has a predetermined reflection
The grid structure having the coefficient is determined by the predetermined weighing cycle.
The initial number set by the initial number setting means for each
It is configured to be connected in series by the number of
Digital weighing signal is filtered and this filtered digital
A grid-type FIR filter that outputs a total weight signal;
Monitor the filtered digital weighing signal and
Of the vibration component contained in the processed digital weighing signal
At least a determination as to whether or not it is within the predetermined allowable attenuation range is made.
Judgment that the amplitude of the vibration component is within the allowable damping range
Sampling of digital weighing signal during one weighing cycle until
Determining means that is repeatedly performed for each ring;
Therefore, if the amplitude of the vibration component is outside the allowable damping range,
When it is determined, the lattice type FIR filter is configured
A predetermined number of said lattice structures in said lattice structure
Means for increasing the lattice structure for connecting
It is characterized by the following. [0009] The adaptive filter for a digital scale according to the present invention is
DC component corresponding to the weight of the weighing object of the digital weighing signal
The purpose is to extract
Its filtering transfer function, such as an adaptive filter for communications, etc.
Achieves sufficient attenuation over a wide stopband,
In addition, ensure linear phase characteristics, and
To achieve an amplitude characteristic such that the gain is as flat as possible.
It is not for the purpose. According to the present invention, the weight of an object to be weighed is measured.
Digital weighing signal is obtained by a lattice FIR filter
(Hereinafter referred to as an FIR filter),
The FIR filter is used for one sample of the input digital weighing signal.
Filters the coupled data and filters the filtered digital data.
And outputs a weighing signal (hereinafter, referred to as a filtered signal). Size
The determining means determines the amplitude of the vibration component contained in the filtered signal.
The determination whether or not within the specified allowable attenuation range
It is determined that the amplitude of this vibration component is within the allowable damping range
Digital weighing signal sampling during this one weighing cycle
Repeat for each ring. And the vibration of this vibration component
Each time the determining means determines that the width is outside the allowable attenuation range,
The grating structure increasing means is added to the grating structure of the FIR filter.
A predetermined number of lattice structures are connected in series. like this
In addition, the number of lattice structures constituting the FIR filter is increased.
Increase the damping rate to attenuate vibration components
Can eventually be included in the digital weighing signal
The amplitude of the vibration component can be attenuated within the allowable attenuation range.
Wear. However, every predetermined weighing cycle, F
Set the number of lattice structures of the IR filter to a predetermined initial number
At the beginning of this predetermined weighing cycle.
The sampling data of the input digital weighing signal
By FIR filter with child structure set to initial number
It is filtered. At this time, the digital input
The amplitude of the vibration component of the weighing signal is within the allowable attenuation range
In this case, the number of grating structures of the FIR filter is not increased.
No. In the FIR filter, each lattice structure is directly
It is connected to columns, and the stationary gay
The lattice structure of the FIR filter is
Even if the number of FIR filters is changed,
Is always 1 and can be constant. And each
The reflection coefficient of each grating structure is determined in advance,
Use this reflection coefficient for
There is no need to calculate the reflection coefficient when performing this. [0013] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described with reference to FIGS.
Will be explained. FIG. 2 shows an adaptive filter for a digital scale according to the present embodiment.
Block showing an electric circuit of the digital scale 10 having a filter
FIG. As shown in FIG.
Load detector 1, amplifier 2, analog / digital conversion
(Hereinafter referred to as an A / D converter) 3, a storage unit 4,
Operation control unit 5, input / output interface 6, display unit
7, an operation unit 8, and a communication control unit 9. Ma
FIG. 3 is a block diagram showing the basic configuration of the digital scale 10.
FIG. 3 is a simplified block diagram of FIG.
You. The digital scale adaptive filter 20 of FIG.
Are stored in the storage unit 4, the operation control unit 5, and the storage unit 4.
A program showing the reflection coefficient and the content of the filtering process
It is composed by The communication control unit 9 is provided with a digital weighing signal
Pull data (hereinafter referred to as sample data) will be described later.
The function to transmit to the reflection coefficient calculation device and the reflection coefficient calculation
It receives the reflection coefficient calculated by the device and
And a function to output to the interface 6.
is there. Then, it is sent to the communication control unit 9 of the digital scale 10.
The transmitted reflection coefficient is provided on the digital scale 10.
It is stored in the storage unit 4. Note that sample data is
The weighing object 11 is actually measured by the load detector 1 such as a load cell.
Digitized analog weighing signal obtained by weighing
A digital weighing signal, which is
Digital weighing signal that has not been filtered. The input / output interface 6 is a
Function to output data to the communication control unit 9,
Function to output the reflection coefficient input from the
Have. The arithmetic control unit 5 includes, for example, a microcomputer.
Data and DSP, etc.
The reflection coefficient and the filtering stored in the storage unit 4
Filtering based on the program that indicates the processing content
You. Then, the filtered digital weighing signal is converted to a weight value.
It is something to convert. The storage unit 4 includes, for example, a ROM and
It is composed of RAM etc. and various data such as reflection coefficient
Data can be written and deleted.
In addition, the storage unit 4 is represented by a flowchart shown in FIG.
The stored program is stored, and the arithmetic control unit 5
Filter the digital weighing signal input according to this program.
Can be wave treated. The display unit 7 is calculated by the arithmetic and control unit 5.
Display the weighed values, the contents input by the operation unit 8, etc.
Can be The operation unit 8 is operated by the operator.
By operating 8, the power of the digital scale 10 is turned off.
ON / OFF, and, for example, sample data
Input various commands such as commands to send data.
You can do it. The load detector 1 is provided, for example, from a load cell or the like.
The weight of the object 11 is weighed and an analog weighing signal is obtained.
Is output. This analog weighing signal
Signal corresponding to the weight of the object 11 to be weighed.
It contains a vibration component in addition to the flow component. Amplifier 2 is a load
A minute analog weighing signal from the detector 1 is input and
The analog weighing signal is amplified at a predetermined amplification rate, and A / D conversion is performed.
Output to the converter 3. Also, this amplifier 2
Is unnecessary high frequency when digitizing analog weighing signals.
A low-pass filter is added to attenuate wave components
ing. A / D converter 3 receives the output signal of amplifier 2
This signal is converted to a digital weighing signal and
5 is output. The digital scale adaptive filter 20 is
The storage unit 4, the arithmetic control unit 5, and the storage unit 4 shown in FIG.
A program showing the stored reflection coefficient and the contents of the filtering process
And an FIR having the structure shown in FIG.
It has a filter 22. This FIR filter 22
Is obtained by connecting the lattice structures 24 shown in FIG. 5 in series.
It is configured. And this lattice structure 24 is increased
Change the filtering transfer function, e.g.
Frequency (hereinafter referred to as notch frequency)
The decay rate is changed. Each of the FIR filters 22
Equations 1 to 4 show equations indicating the relationship between the parameters. [0019] (Equation 1) [0020] (Equation 2)   f (n)^(m)= ｛F (n)^(m-1)+ K_m・ G (n-1)^(m-1)｝ / (1 + k_m) [0021] (Equation 3)   g (n)^(m)= ｛G (n-1)^(m-1)+ K_m・ F (n)^(m-1)｝ / (1 + k_m) [0022] ## EQU4 ## f (n)⁽⁰⁾= G (n)⁽⁰⁾= X (n) Where k_mIs the reflection coefficient, and m (m = 1, 2,
..., M) is the filter order, and x (n) is the digital measurement signal
Data, n (n = 0, 1, 2,..., N)
The number of pulling. Also, in general, f (n)^(m)Is self-time
Forward prediction error of the regression model, g (n)^(m)Is an autoregressive model
This is called the backward prediction error. The FIR filter 22 transmits the filtered signal.
The function F (z) is represented by Expression 5. [0025] (Equation 5) A shown in Equation 5_i ^(m)Is the filter coefficient,
This filter coefficient a_i ^(m)Between the reflection coefficient km and
There is a relationship shown in Equation 6. [0027] (Equation 6)   a_i ^(m)= A_i ^(m-1)+ K_m・ A_mi ^(m-1)        ： I = 1,2, ・ ・ ・, m-1   a_i ^(m)= K_m                                  : I = m Then, the filtering is performed by the filtering transfer function F (z) of Formula 5.
Filtered signal which is the output of the wave processed FIR filter 22
y (n)^(m)Is represented by Expression 7. [0029] [Equation 7] y (n)^(m)= ｛F (n)^(m)+ G (n)^(m)｝ / 2 Therefore, Equations 4 and 7 are calculated from Equations 1 and 4.
This allows the input digital weighing signal x (n) to be
Filtered signal y (n), which is the output of FIR filter 22
^(m)Can be requested. And this filter order
m is the grating of FIG. 5 that constitutes the FIR filter 22 of FIG.
This corresponds to the number of connections of the structure 24. That is, FIR filter
Increasing the filter order of 22 by one is a lattice structure
24 by one, which results in a FIR
The filtering transfer function F (z) of the filter 22 changes. What
Note that the number of taps L of the filtering transfer function F (z) is L = m
+1. The digital weigher 10 is, for example, a combination weigher.
It's being used. When using a combination weigher,
Load detector 1, the same number of amplifiers 2 as load detector 1, and
And a multiplexer (not shown). First, the object to be weighed
When a weight of 11 is imposed on each load detector 1, an analog meter
A weight signal is generated, and this analog weighing signal is output to each load detector.
1 are input to the amplifiers 2 connected to the respective amplifiers 1. And this
The amplifier 2 amplifies the analog weighing signal. this
The amplifier 2 is provided with a low-pass filter.
High-frequency analog weighing signal
Minutes are removed. Analog weighing with high-frequency components removed
The signal is A / D converted via a multiplexer (not shown).
Is supplied to the converter 3 and converted into a digital weighing signal.
It is supplied to the control unit 5. This arithmetic control unit 5 is supplied
Of the digital weighing signal stored in the storage unit 4
Number and the program represented by the flowchart shown in FIG.
Filter processing is performed based on the And this digital
The vibration component of the weighing signal is attenuated by the arithmetic and control unit 5.
The DC component is extracted (filtered), and this DC component is extracted.
The minute is converted into a weight value and displayed on the display unit 7. Next, the digital scale adaptive filter 20
The operation of the filtering process will be described. This digital scale
The filter 20 has the structure of the FIR filter 22 as described above.
To increase the number of lattice structures 24 constituting the
This changes the filter transfer function F (z).
You. Also, the filtered signal which is the output of the FIR filter 22 is
No. y (n)^(m)Is to calculate Equation 1 to Equation 4 and Equation 7.
And is sought. First, the filtered signal y (n)^(m)When calculating
It is necessary to calculate the reflection coefficient km to be used in advance.
The procedure for calculating the reflection coefficient km will be described below.
You. This reflection coefficient km is obtained by the reflection coefficient calculation device.
Can be The reflection coefficient calculation device is, for example, a personal computer.
Computer system such as a personal computer.
Is done. This reflection coefficient calculating apparatus is a digital scale 10
Input sample data x (n) from
Difference value Δx (n) = x in the discrete-time sequence of x (n)
(n) −x (n−1) is configured inside the reflection coefficient calculation device.
To the FIR filter 30 shown in FIG. This F
The IR filter 30 has a transfer function H (z) shown in Expression 8.
And each parameter is a Burg algorithm
The relations are shown in equations 9 to 14 depending on the mechanism. Soshi
The filter order M of the FIR filter 30 is
The value is set to a value about large, for example, a value of about 40 order, and
By calculating the algorithm of Equation 14 from
The reflection coefficient km of the grating structure 24 (m = 1, 2,...,
M) is calculated. Then, the calculated reflection coefficient km
Is sent to the communication control unit 9 of the digital scale 10 shown in FIG.
And stored in the storage unit 4 through the input / output interface 6
Is done. Note that the sample data includes the vibration
The worst conditions where the components can be considered as digital weighing signals
Data, such as having many frequencies,
The digital weighing signal contains a vibration component with a large amplitude.
You. [0035] (Equation 8)   H (z) = ｛1 + a_m ^(m)+ (A₁ ^(m)+ A_m-1 ^(m)) Z^-1＋ ・ ・ ・                                       + (1 + a_m ^(m)) Z^-m｝ / 2 [0036] (Equation 9) [0037] (Equation 10)   a_i ^(m)= A_i ^(m-1)+ K_m・ A_mi ^(m-1)        ： I = 1,2, ・ ・ ・, m-1   a_i ^(m)= K_m                                  : I = m [0038] [Equation 11] f (n)^(m)= F (n)^(m-1)+ K_m・ G (n-1)^(m-1) [0039] (Equation 12) g (n)^(m)= G (n-1)^(m-1)+ K_m・ F (n)^(m-1) [0040] (Equation 13)   f (n)⁽⁰⁾= G (n)⁽⁰⁾= Δx (n): n = 1,2, ..., N-1 [0041] [Equation 14]   Δx (n) =-a₁ ^(m)・ Δx (n-1)-・ ・ ・ -a_m ^(m)・ Δx (n-m) + f (n)^(m) Where a_i ^(m)(I = 1, 2, ...,
m) is a filter coefficient, and m (m = 1, 2,..., M) is
The filter order, n (n = 0, 1, 2,..., N)
The number of samplings. The transfer function H (z) shown in Expression 8 is
Vibration with persistent vibration components included in sample data
Knock with a large attenuation rate only for the frequency where the component exists.
This allows the filter to be
The order is set to a general FIR filter, for example, FIR for sound etc.
F with a filter order less than the filter order of the filter
An IR filter can be configured. Also this transmission
The number of taps L of the function H (z) is L = m + 1. The FI constituting the adaptive filter 20
Filtered signal y (n) of R filter 22^(m)Is the reflection coefficient
Using the reflection coefficient km calculated by the calculation device,
From Equation 4 and Equation 7
You. Regarding the procedure of the filtering process of the adaptive filter 20,
This will be described with reference to the flowchart shown in FIG. First, the grating constituting the FIR filter 22
An initial number p of the structures 24 is set. That is, the filter
A number m = p is set (step S2). This step S
2 corresponds to the initial number setting means. Next, a digital weighing signal x (n) is obtained.
(Step S4). Then, by setting m = p, the FIR
Filtering is performed by the filter 22. Of the FIR filter 22
Output filtered signal y (n)^(m)Are Equations 1 to 4, and
It is calculated by calculating Equation 7 (Step S6).
In addition, in the calculation of the filtering process, the reflection coefficient calculating device is used.
Use the reflection coefficient km calculated in. And this filtering
Signal y (n)^(m)Is output (step S8). On the other hand, this filtered signal y (n)^(m)Monitor
And this filtered signal y (n)^(m)Fluctuation Δy (n)^(m)And
(Step S1).
0). Step S10 corresponds to a determination unit. What
Note that the fluctuation of the filtered signal Δy (n)^(m)Is represented by Equation 15
You. [0048] (Equation 15) The variation y of the filtered signal represented by the equation (15)
(n)^(m)Is larger than the allowable level V, that is, y
(n)^(m)> V, included in digital weighing signal
Judge that the vibration component is outside the allowable attenuation range,
The number m is increased by a predetermined number q. That is, m = m + q.
You. Thus, the lattice structure 24 of the FIR filter 22
Increases by q (step S14). And the next
Obtain sampling data x (n) of digital weighing signal
To perform the same processing as above,
Update the number n, ie, n = n + 1, and step
It returns to S4 (step S16). This step S
14 corresponds to a lattice structure increasing means. Further, the variation y (n) of the filtered signal in equation (15)^(m)
Is smaller than the allowable level V, that is, y (n)^(m)<
If V, the vibration component contained in the digital weighing signal is
Judge that it is within the allowable attenuation range and increase the filter order
Sampling data of the next digital weighing signal
In order to obtain the value x (n), the process is performed with n = n + 1.
The process returns to step S4 (step S16). After completion of this determination, a separate object
Sampling of digital weighing signal obtained by weighing 11
The same processing as described above is performed by acquiring the data x (n).
The determining means is included in at least the digital weighing signal.
Until the vibration component is determined to be within the allowable damping range.
That is, y (n)^(m)<One weighing cycle until V is satisfied
Data x of digital weighing signal
(N) is determined for each input, and this is repeated. Also Phil
If the order m increases and reaches the maximum value M,
The sampling data x (n) of the digital weighing signal of
FIR filter 22 of filter order M, ie, lattice structure
FIR filter having a structure in which M pieces are connected in series
22. Also, a predetermined number of weighing cycles (eg,
For example, of the numbers 1, 2, 3, ..., set in advance
At the end of the weighing cycle), the FIR filter 2
Reset the number of lattice structures 24 constituting 2 to the initial number p
You. That is, step S2 is performed for each predetermined weighing cycle.
repeat. Then, configure the FIR filter 22 again.
By increasing the number of lattice structures 24
To increase the damping rate, and ultimately,
Allowable attenuation of amplitude of vibration component included in digital weighing signal
It can be attenuated within a range. Note that this predetermined
Digital weighing signal x input at the beginning of the weighing cycle
(n) is an FIR image in which the lattice structure 24 is set to the initial number.
Filtered by the filter 22
The amplitude of the vibration component of the digital weighing signal x (n)
If it is within the range, the lattice structure 2 of the FIR filter 22
The number of 4 is not increased. The adaptive filter 20 according to the present embodiment includes
A FIR filter 22 in which structures 24 are connected in series,
In addition, since the steady-state gain of each lattice structure 24 is set to 1,
Therefore, even if the number of lattice structures 24 is changed,
The steady gain of the entire data 22 is always 1 and must be constant.
Can be. Then, the reflection coefficient of each grating structure 24 is
This reflection coefficient km is used in the filtering process.
Therefore, when a new lattice structure 24 is added,
There is no need to calculate the coefficient km. This allows this
The response filter 20 is a microcompi
Do not place a heavy load on the computer or DSP.
Therefore, the vibration component included in the digital weighing signal x (n)
Compared to the conventional adaptive filter 20 in a shorter time
Can be processed. The adaptive filter 20 has a predetermined
The lattice structure 2 of the FIR filter 22
The number m of 4 is set to a predetermined initial number p.
An FIR filter including the lattice structure 24 having the initial number p.
22 filters the data of the digital weighing signal x (n).
The degree of attenuation of the vibration component of the output result of the filtering process.
In the configuration where the lattice structure 24 is increased while monitoring
is there. With this configuration, the input digital weighing signal x
The vibration components contained in (n) can be sufficiently attenuated.
You. In addition, the minimum lattice structure corresponding to the amplitude of the vibration component
Structure 24, that is, an FIR filter 2 having a filter order m
2.Because the filter processing is performed in step 2, the filtering processing time
It can be kept to a minimum length. Therefore, the data provided with the adaptive filter 20
When the object 11 is weighed by the digital scale 10, a high-speed meter
Quantity and high precision weighing can be enabled. In the determination means, the fluctuation of the filtered signal
Δy (n)^(m)Is calculated by the formula shown in Expression 15.
Is a formula other than the formula (15), for example, a filtered signal y (n)
^(m)May be calculated by adding the moving average of
No. Also, the digital scale 1 having the adaptive filter 20
0 is not limited to the combination weigher. For example, one load detector 1 is installed.
Also used for digital scales and weight sorters
Can be. Next, the digital weighing signal x (n) shown in FIG.
Filter 2 of the present embodiment in the case of actually filtering
The operation of 0 will be described with reference to FIGS. FIG. 7 shows the digital balance 10 shown in FIG.
0 g of the object to be weighed 11 was weighed (however, 8
Digital weighing signal x (n) averaged)
X (n) before input to the adaptive filter 20.
You. The sampling time T is 5 msec. Soshi
The digital weighing signal x (n) shown in FIG.
Adaptive filter 2 for attenuating vibration components
0 was made. First, the object to be weighed 11 is placed on the digital scale 10.
The digital weighing signal x (n) with the boarding completed is
For example, the sampling number n = 150 to 250 shown in FIG.
(Hereinafter referred to as data sample time D)
Weighing signal x (n) as sample data x (n)
Calculate reflection coefficient of sample data x (n) from digital scale 10
Send to device. The reflection coefficient calculating device calculates the received
Based on the sample data x (n), the filter order M = 40
Substituting 8 into Equation 14 to calculate the reflection coefficient km,
The calculated reflection coefficient km is sent back to the digital balance 10. So
Then, the digital scale 10 receives the reflection coefficient km and
And stored in the storage unit 4 as the reflection coefficient km of the adaptive filter 20.
Remember. The adaptive filter 20 is stored in the storage unit 4.
Using the remembered reflection coefficient km, Equations 1 to 4 and the number
7 based on the input digital weighing signal x (n)
To perform the filtering process. Note that the initial number p of the lattice structures 24 is
Is p = 20. Further, the increased number q of the lattice structure 24
Is q = 1. FIG. 8 shows an adaptive filter 20 shown in Expression 5.
Is a diagram showing the amplitude characteristic of the filtered transfer function F (z) of FIG.
F shown₂₀, F₃₀, F₄₀Are respectively the filter order m = 2
These are amplitude characteristics at 0, 30, and 40. Smell
The angular frequency ω = 120 to 130 mrad / sec
The largest reduction in the vibration components included in the sample data
Shows the frequencies that require decay. And filter
As the order m increases, the notch frequency increases, and
It can be seen that the decay rate increases. In addition, filters
The gain in the DC component does not change even if the order m is changed.
Is recognized. Thus, the filter order m is changed.
, That is, the FIR filter 22 is configured.
By simply changing the grating structure 24, the filtering transfer function F (z)
Can be changed. FIG. 9 shows the digital weighing signal x of FIG.
For the input of (n), we have the filtering transfer function F (z)
Output response of the adaptive filter 20 and the filter order m
The filter transfer function F of Equation 5 fixed at m = 20 and 40
The figure which shows the output response of FIR filter 22 which has (z).
It is. 9 represents the output response of the adaptive filter 20.
Show. Also, Y₂₀And Y₄₀Is the filter order m
Is the output of the FIR filter 22 when m = 20 and 40
Indicates a response. In the figure, the response of Ym is
Before the sampling time D, that is, the sampling number n = 15
FIR filters with filter order m = 20 for up to 0
Acts as the data 22 and then the number of samples n increases
, The filter order m is increased, and n = 1
For FIR filters with filter order m = 40,
It is recognized that it is operating as the router 22. Then, with respect to Ym in FIG.
FIG. 10 is a graph showing an error from the true value of 1. Same figure
From the beginning of the data sample time D, ie, the sample
The vibration component is attenuated immediately after the ring number n = 150.
It is recognized that. That is, the digital weighing signal x (n)
It is recognized that the included vibration component is attenuated in a short time.
Can be Also included in the output of the adaptive filter 20
Vibration component and vibration of digital weighing signal x (n) shown in FIG.
In comparison with the components, the change of the digital weighing signal shown in FIG.
The dynamic Δd is up to about 88 g at the data sample time D.
In contrast, the output of the adaptive filter 20 shown in FIG.
The force fluctuation ΔF is found after the sampling number n = 150.
To a maximum of about 0.13 g. From this, the book
The adaptive filter 20 of the embodiment is configured to receive the input digital weighing signal.
The vibration components contained in the signal x (n) must be sufficiently attenuated
Is recognized. [0063] The adaptive filter for a digital scale according to the present invention has the following features.
Multiple grating structures with a constant gain of 1 connected in series
It consists of FIR filters and changes the number of connections of the lattice structure
In addition, the reflection coefficient of each grating structure is predetermined.
This reflection coefficient is used for filtering.
You. With this configuration, this digital scale adaptive filter
Configures this adaptive filter for filtering.
Puts a heavy burden on computer and DSP
Therefore, filtering can be performed at high speed.
Therefore, by using this adaptive filter,
This has the effect of enabling high-speed weighing of a digital balance. Then, the adaptive filter for a digital scale according to the present invention
At every predetermined weighing cycle, i.e.
Data string of digital weighing signal obtained by weighing the weighing object
Each time is input, the number of lattice structures of the FIR filter is predicted.
First, set the initial quantity to
Digital measurement signal by FIR filter
This is a configuration for filtering data of a signal. With this configuration
Of the vibration component contained in the input digital weighing signal
The width is small or the frequency at which this vibration component exists is small.
In other words, the amplitude of the vibration component is within the allowable damping range.
The number of lattice structures is the initial number,
Acts as a filter for numbers and, consequently, the height of the digital scale
There is an effect that quick measurement can be performed. Why
Therefore, the number of lattice structures is small, that is, the filter order is small.
This is because the filtering processing time of the FIR filter is short.
You. Also, included in the input digital weighing signal
The amplitude of the vibration component is large or
Frequency, that is, the amplitude of the vibration component
If it is outside the decay range, at least the amplitude of the vibration
Grid structure until the determining means determines that the current is within the attenuation range
Is increased by a predetermined number. This configuration
The amplitude of the vibration component of the input digital weighing signal
Is large or the frequency at which this vibration component exists is large.
The number of lattice structures should be larger than the initial number.
High-order FIR filters with a large filter order
Function as a filter, and as a result,
Digital weighing signal with vibration components having many frequencies
High weighing accuracy of the digital scale
There is an effect that is. However, in this case, the weighing time
Although it becomes slightly longer, the longer time
It can be made the minimum length according to the size
effective.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a lattice FI for a digital scale according to an embodiment of the present invention.
It is a flowchart which shows the operation procedure of the filtering process of an R filter. FIG. 2 is a block diagram showing an electric circuit of the digital scale including the digital scale adaptive filter of the embodiment. FIG. 3 is a block diagram illustrating a basic configuration of a digital balance including the digital scale adaptive filter of the embodiment. FIG. 4 is a lattice FI constituting an adaptive filter according to the embodiment.
It is a block diagram of an R filter. FIG. 5 is a block diagram of a lattice structure constituting the lattice type FIR filter. FIG. 6 is a block diagram of a grating type FIR filter included in the reflection coefficient calculation device of the embodiment. FIG. 7 is a waveform diagram of a digital weighing signal input to the digital weighing adaptive filter of the embodiment. FIG. 8 is a diagram showing an amplitude characteristic of a filtering transfer function of an FIR filter included in the digital weighing adaptive filter of the embodiment. FIG. 9 shows output responses of the digital scale adaptive filter of the embodiment and FIR filters having filter orders of 20 and 40. FIG. 10 is a diagram showing an output response of the digital weighing adaptive filter of the embodiment. [Description of Signs] 1 Load detector 4 Storage unit 5 Operation control unit 9 Communication control unit 20 Digital scale adaptive filter step S2 Initial number setting step of grid structure (filter order) S4 Digital weighing signal sample data x
(N) Acquisition step S6 Output y (n) ^(m) calculation step S8 Output y (n) ^(m) output step S10 Output vibration component Δy (n) ^(m) calculation step S12 Vibration component determination Step S14: Increase the lattice structure (filter order)

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-62-280625 (JP, A) JP-A-60-117115 (JP, A) JP-B7-9937 (JP, B2) JPB-6- 21814 (JP, B2) Patent 3057382 (JP, B2) Patent 3071825 (JP, B2) Patent 3251754 (JP, B2) Patent 2719722 (JP, B2) Patent 3300107 (JP, B2) US Patent 4789014 (US, A) ( 58) Fields surveyed (Int. Cl. ⁷ , DB name) G01G 23/37

Claims (1)

(57) [Claims 1] Initial number setting means for setting a predetermined number of lattice structures to an initial number for each predetermined measuring cycle, and a steady gain of 1 and A grating structure having a predetermined reflection coefficient is connected in series by the initial number set by the initial number setting means for each of the predetermined weighing cycles, and filters a digital weighing signal including a vibration component. A grating FIR filter for processing and outputting the filtered digital weighing signal; and monitoring the filtered digital weighing signal to determine the amplitude of the vibration component contained in the filtered digital weighing signal. The determination of whether or not the amplitude is within the allowable attenuation range is performed until the amplitude of the vibration component is determined to be within the allowable attenuation range. A determination unit that is repeatedly performed for each pulling; and when the determination unit determines that the amplitude of the vibration component is outside the allowable attenuation range, the determination unit further determines the amplitude of the vibration component in the lattice structure constituting the lattice type FIR filter. Means for increasing the number of the above-mentioned lattice structures in series, and an adaptive filter for a digital scale.