JP2010216832A - Signal processing apparatus, image formation apparatus, and signal processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a signal processing apparatus which adopts a sampling period lower than a Nyquist frequency to appropriately restore the frequency of a peak value obtained by FFT transformation. <P>SOLUTION: The signal processing method samples data required for a frequency analysis at two different sampling periods (processing 201 to determination 206), checks whether a frequency giving an unmatched peak value exists(determination 208), searches for a peak value satisfying the relationship (fa-fb)=(Pa-Pb) if the frequency exists (determination 209), and complements (restores) the frequency Fr of the peak value using the relationship Fr=(fa-Pa)=(fb-Pb) (processing 211) if such a peak value is found (the result of determination 210 is YES). <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a signal processing apparatus, an image forming apparatus, and a signal processing method for detecting a state quantity that periodically changes a target and restoring the frequency of a peak value obtained by performing FFT conversion on the state quantity.

In recent years, image quality in image forming apparatuses is more demanded to have high image quality. Particularly in the production market, there is a high demand for preventing image deterioration, and output of abnormal images must be avoided as much as possible.
For this reason, there is an increasing need for failure prediction for monitoring an abnormal state of the apparatus due to aging or machine abnormality. By performing failure prediction, it is possible to perform repairs by a service person (maintenance worker) before outputting an abnormal image, or to make an advance response by automatic correction or degenerate operation in the apparatus.

One of the means for predicting such failure is to sample motor encoder data at regular intervals and analyze the frequency of the data to detect a failure precursor or identify the location of the failure. .
In conventional frequency analysis, aliasing occurs with half of the sampling frequency as the Nyquist frequency, and the correct peak near the analysis frequency is not known. Therefore, sampling is performed at a sampling frequency that is twice or more the analysis frequency band. It was.
However, since the conventional frequency analysis accuracy improvement technology has been supported by increasing the sampling frequency, circuit elements for performing the sampling process in a short time and at a high speed are required, and it is possible to support high-speed frequency analysis. A software system is required.
That is, there is a problem that increasing the sampling period increases the amount of software processing and increases the load on the CPU (central processing unit) that is the control means of the apparatus.

  In Patent Document 1, in order to solve the aliasing problem, in order to determine whether or not the peak waveform of the frequency spectrum is due to aliasing, sampling is performed at two different frequencies, and DFT (discrete FFT (high-speed FFT) is performed. (Fourier transform)) Disclosed is a technology that compares the frequency of peak values in the discrete frequency spectrum obtained as a result of the calculation to a predetermined value or more, and determines that the frequency satisfying a certain change in peak value is not a true peak but aliasing Has been.

  However, in the technique of Patent Document 1, since the peak value changes and a peak exceeding a certain value is only determined as aliasing, the original frequency cannot be analyzed with high accuracy. was there.

  The present invention has been made in view of such circumstances, and an object thereof is to provide a signal processing device, an image forming apparatus, and a signal processing method that can improve the accuracy of frequency analysis and reduce the CPU load. To do.

In order to achieve the above object, the present invention is a signal processing device for detecting a state variable that changes periodically, and restoring the frequency of the peak value obtained by FFT conversion of the state variable. A first sampling means for sampling and storing the state quantity at a first sampling frequency fa less than twice the frequency band; and a second sampling frequency fb lower than the first sampling frequency fa. A second sampling means for sampling and storing the state quantity; a first frequency analysis means for obtaining a peak value frequency by performing an FFT transform on the first sample quantity obtained by the first sampling means; Second frequency analysis means for obtaining a peak value frequency by performing FFT conversion on the second sample amount obtained by the second sampling means; and the first frequency A combination of different values of the frequency Pa of any first peak value acquired by the analyzing means and the frequency Pb of any second peak value acquired by the second frequency analyzing means described below, When there is a combination of the frequency Pa and the frequency Pb satisfying the expression (1), the peak value frequency Fr is calculated based on the following expression (2), and the calculated peak value of the frequency is determined as the first frequency. The frequency component is restored by adding it to the peak value of the frequency acquired by the analyzing means or the second frequency analyzing means, while the following formula (1) is used in a combination in which the values of the frequency Pa and the frequency Pb are different. When there is nothing satisfying the condition, there is provided a frequency component restoring means that does not restore the frequency component.
(Fa−fb) = (Pa−Pb) (1)
Fr = (fa−Pa) = (fb−Pb) (2)

Further, a plurality of combinations of the first sampling frequency fa and the second sampling frequency fb are set, and for each combination, the first frequency analysis means, the second frequency analysis means, and the frequency component restoration are performed. It is preferable that a plurality of frequency components can be restored by operating the means.
The object may be a motor, and the state quantity may be a pulse interval time output from a pulse encoder provided on a rotation shaft of the motor.

  The object is a motor, and the state quantity is a control amount representing a speed target value given to the motor.

An apparatus abnormality that includes the frequency component reconstructed by the frequency component reconstructing means and determines a device abnormality based on the peak value of the frequency acquired by the first frequency analyzing means or the second frequency analyzing means. When the determination means and the apparatus abnormality determination means determine the apparatus abnormality, a third sampling for sampling and storing the state quantity at a third sampling frequency such that the Nyquist frequency is higher than the analysis frequency band. And a third frequency analyzing means for acquiring a peak value frequency by performing FFT conversion on the third sample amount obtained by the third sampling means.
In the initial state of the apparatus, the processing result of the third frequency analyzing means is stored, and thereafter, until the apparatus abnormality determining means determines the apparatus abnormality, the first sampling means, the second sampling means, The sampling means, the first frequency analysis means, the second frequency analysis means, and the frequency component restoration means may be operated.

  Therefore, according to the present invention, it is possible to appropriately restore a frequency component necessary for frequency analysis by applying a sampling frequency of twice or less to the analysis frequency band, so that the CPU load is not increased so much. The effect that the accuracy of frequency analysis can be improved is obtained.

1 is an overall view illustrating a schematic configuration of a color copying machine including a color image forming apparatus using a general intermediate transfer belt configuration. It is a functional block diagram explaining the function of the color copying machine shown in FIG. 2 is a circuit block diagram illustrating a circuit configuration of a unit for controlling a motor that rotationally drives photosensitive drums 102 to 105. FIG. It is a functional block diagram for demonstrating sampling and preservation | save of the data for detecting an apparatus abnormal state. It is a timing chart for demonstrating the concept regarding the CPU processing time regarding the sampling of the data for detecting motor control and an apparatus abnormal state, and CPU processing amount. It is a graph for demonstrating an aliasing and a Nyquist frequency. It is a graph for demonstrating the principle of this invention. It is the flowchart explaining the outline of the speed control of the motors M1 and M2 performed by the CPU20. It is the flowchart figure explaining the outline of the detection method of apparatus abnormality.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
〔Example〕
FIG. 1 is an overall view illustrating a schematic configuration of a color copying machine including a color image forming apparatus using a general intermediate transfer belt configuration. In FIG. 1, an image forming unit using an electrophotographic process formed around the photosensitive drum is not shown for the sake of simplicity.

In FIG. 1, 101 is a scanner unit for reading an original in a color copying machine, and 102 to 105 are photosensitive drums for four colors of Y (yellow), C (cyan), M (magenta), and K (black), respectively. , 106 is a fixing unit for fixing a toner image transferred by the photosensitive drums 102 to 105 onto a transfer paper 112 (described later), and 107 is an image for each color formed on the photosensitive drums 102 to 105. In addition, an intermediate transfer belt for transferring to the transfer paper 112, a secondary transfer roller for transferring an image on the intermediate transfer belt 107 to the transfer paper 112, and an intermediate 114 arranged at a portion facing the secondary transfer roller 108. A repulsive roller 109 for generating / maintaining a nip between the transfer belt 107 and the secondary transfer roller 108, 109 is a skew of the transfer paper 112 A registration roller unit composed of a pair of registration rollers for carrying forward and transfer papers, 113 a paper feed unit on which the transfer paper 112 is stacked, 112 a transfer paper, 110 a transfer paper 112, and a secondary transfer roller A paper feeding roller for feeding the paper from the transfer unit consisting of 108 and the repulsive roller 114 to the transport unit, 111 a paper transport roller for transporting the transfer paper 112 sent from the paper feed roller 110 to the registration unit, and 115 for transferring and fixing the image. This is a paper discharge unit for discharging the transferred transfer paper 112. Reference numeral 116 denotes an intermediate transfer scale detection sensor that detects a scale formed on the intermediate transfer belt 107 and generates a pulse in synchronization with the conveyance of the intermediate transfer belt 107.
For the sake of explanation, the present embodiment uses a color copier equipped with the scanner unit 101. However, the present invention also applies to a printer device that receives image data from an external controller such as a personal computer device and forms an image. Can be applied.

FIG. 2 is a functional block diagram illustrating functions of the color copying machine shown in FIG.
In FIG. 2, an image read by the scanner unit 101 is transferred to the main control unit 1. The main control unit 1 includes an image formation control function for generating an image actually formed on the transfer paper 112 from an image read by the scanner unit 101, and a motor (illustrated) for forming the image on the transfer paper 112. It includes a timing generation control function for generating drive timing such as abbreviations and solenoids (not shown), and an operation unit control function for processing input from an operation unit (not shown).
The image forming control unit 2 is a part that mainly performs control related to driving of the photosensitive drums 102 to 105, and the image forming process that performs control related to the electrophotographic process such as charging, exposure, transfer, and the like. A control unit 4 is included. The intermediate transfer control unit 5 includes an intermediate transfer motor control unit 6 for intermediate transfer, an intermediate transfer belt control unit 7 for transferring a toner image on the photosensitive member onto the intermediate transfer belt, and a primary transfer control unit 8. The secondary transfer control unit 9 includes a secondary transfer motor control unit 10 and a secondary transfer control unit 11 that transfers the toner image on the intermediate transfer belt to the transfer paper 112. The fixing controller 12 controls the fixing function for fixing the toner image transferred onto the transfer paper 112 on the transfer paper. The sheet feeding / conveying control unit 13 performs control including a series of operations from feeding, conveying, and discharging of the transfer sheet 112.

In the color copying machine, an image read from the scanner 101 is sent to the main control unit 1, and the main control unit 1 generates image data of an image formed on the transfer paper 112. The generated image data is formed on the photosensitive drums 102 to 105 by the image forming control unit 2, and then an image is formed on the intermediate transfer belt 107 by the intermediate transfer control unit 5. Further, the image formed on the intermediate transfer belt 107 is transferred onto the transfer paper 112 when the transfer paper 112 from the paper feeding unit is conveyed between the intermediate transfer belt 107 and the secondary transfer roller 108.
During this time, the photoconductor motor, the intermediate transfer motor, and the secondary transfer motor are controlled by each control unit so that a normal image is formed on the transfer paper 112.
Then, the image transferred onto the transfer paper 112 passes through the fixing unit 106, and the transfer paper 112 is fixed and discharged to the paper discharge unit 115.

FIG. 3 is a circuit block diagram showing a circuit configuration of a unit for controlling a motor that rotationally drives the photosensitive drums 102 to 105. In this embodiment, a motor M1 that rotates the black photosensitive drum 105 that operates during monochrome copying and a motor M2 that rotates the other photosensitive drums 102 to 104 are provided separately. Two motors for driving 105 are provided.
20 is a CPU (Central Processing Unit) for managing / operating the entire apparatus, 21 is an NVRAM (volatile memory) for temporarily storing data necessary for controlling the apparatus, and 22 is an apparatus A ROM (read only memory; nonvolatile memory) 23 for storing a program and data for operating the motor is a motor drive control unit having a function necessary for motor control, and is an ASIC (Application Specific Integrated circuit (application-specific integrated circuit device). The motor drive control unit 23 generates, for example, a phase excitation signal for rotating the motors M1 and M2 under conditions instructed by the CPU 20, and an encoder signal (described later) for grasping the rotation speed of the motors M1 and M2. ) To detect changes. Reference numerals 24 and 25 denote motor drivers, and a current that allows the motors M1 and M2 to rotate based on a signal from the motor drive control unit 23 flows. Reference numerals 26 and 27 denote encoders that generate pulse signals in order to grasp the rotation states of the motors M1 and M2. The encoders 26 and 27 are attached to the rotating shafts of the motors M1 and M2 or a portion having accuracy of rotation and movement speed, and generate a pulse output corresponding to the speed.

FIG. 4 is a functional block diagram for explaining sampling and storage of data for detecting an abnormal apparatus state.
In FIG. 4, the motor drive control means 23A (motor drive control section 23) outputs instructions of the rotational speeds of the motors M1, M2 to the motor drive means 24A, 25A (motor drive circuits 24, 25) by PWM duty. Thus, the motor drive means 24A, 25A supplies the drive voltage according to the input PWM duty to the motors M1, M2, and rotationally drives the motors M1, M2.

Encoder pulses are generated from the encoder pulse generation means 26A and 27A (encoders 26 and 27) according to the rotation of the motors M1 and M2. Therefore, the pulse width measurement means 31 and 32 measure the pulse width, and the data sampling means 33 and 34 sample the encoder pulse width measured by the pulse width measuring means 31 and 32 at the timing designated by the data sampling timing generating means 35 and 36, and the encoder pulse width data of the sample value is data storage means. 37 and 38 are stored at an appropriate timing and used for frequency analysis by the frequency analysis means 39. The sampling period of the data sampling timing generation means 35, 36 is set by the CPU 20, so that the data sampling period of the data sampling timing generation means 35, 36 can be varied as will be described later.
The encoder pulse width data measured by the pulse width measuring means 31 and 32 is sent to the CPU 20 and used for management and control of the speed and position of the motors M1 and M2.

  Here, as the data sampling means 33 and 34 and the data sampling timing generation means 35 and 36, for example, a timer interrupt function for generating an interrupt at a constant cycle at the data sampling timing is used. It can be realized by sampling means by software that acquires and stores data in memory, or by means of hardware that performs sampling from pulse width data to data storage in memory by DMA activation with time-up of a fixed period Yes, and there is no particular specification regarding the means for realizing it.

FIGS. 5A to 5G show an example of a timing chart for explaining a concept relating to CPU processing time and CPU processing amount relating to sampling of data for detecting motor control and an apparatus abnormal state.
In this case, for the two motors M1 and M2, the CPU 20 executes a control interrupt process for the motor M1 and a control interrupt process for the motor M2, respectively, according to the interrupt signals generated at the periods T1 and T2. The control interrupt process for the motor M1 and the control interrupt process for the motor M2 need to be completed immediately before an interrupt signal of the next interrupt period is generated. As the control interruption process for the motors M1 and M2, for example, a process (described later) schematically shown in the flowchart of FIG. 8 is executed.
On the other hand, the data sampling timing generation means 35 and 36 generate sampling interrupts for signal sampling at the start timing of the sampling period, so the CPU 20 needs to start sampling interrupt processing at that timing. Also, this sampling interrupt process needs to be completed within a sampling period (sampling interrupt).

That is, when the occurrences of the respective interrupts overlap, priority is given to the control with the higher priority or the processing being performed first, and the other is waited until the processing is completed.
Therefore, in order to stably control the motors M1 and M2, as described above, each motor interrupt process needs to be completed within the control cycle interrupt. If the processing cannot be completed within the control interruption cycle, for example, the rotation unevenness of the motors M1 and M2 becomes large and the rotation position management cannot be performed correctly, which greatly affects the performance of the system.
In this way, the interrupt processing performed by the CPU 20 is quite busy as shown in FIG. 5G. For example, when the processing load of the sampling interrupt processing is reduced, the CPU 20 has a sufficient processing capacity. be able to.
Such a situation may always occur even if the CPU 20 is replaced with a higher processing capability, and it is preferable if the processing load on the CPU 20 can be reduced for each processing. There are circumstances.

When performing frequency analysis, there is an advantage that wideband analysis can be performed by acquiring data at a high sampling rate. However, if the speed is too high, the CPU load is unnecessarily increased and the operation of the device is increased. Adversely affect.
Therefore, it is necessary to set the data sampling period for frequency analysis as low as possible so as not to affect the control.

Next, a frequency analysis method according to the present invention will be described.
First, aliasing and the Nyquist frequency will be described with reference to FIGS. 6A to 6C (simple explanation of sampling theory).
For example, when a 1 KHz sine wave as shown in FIG. 6A is sampled at 100 KHz and FFT analysis (Fast Fourier Transform analysis; frequency analysis) is performed using the data, a peak appears only at a frequency of 1 KHz. Comes (see FIG. 6B). By judging this peak, it can be judged that the input data contains a 1 kHz periodic component.
Next, the sampling period is lowered to 1.5 KHz, and when the FFT analysis is performed on the data, the peak of 1 KHz, which is the frequency of the sine wave, is folded around 750 Hz, which is half of 1.5 KHz, and another peak at 500 Hz. Appears (see FIG. 6C).
This folding frequency is called the Nyquist frequency, and the frequency peak appearing at the folding position is called aliasing.
Due to this phenomenon, it is necessary to set a sampling period that is at least twice the frequency band to be analyzed.

Next, the principle of the present invention will be described with reference to FIGS.
For example, as shown in FIG. 7A, when the acquired data includes periodic frequency components of 1 KHz and 10 KHz, sampling is performed at a high sampling rate (for example, 100 KHz) and FFT analysis is performed according to the sampling theory described above. Then, peaks appear at frequencies of 1 KHz and 10 KHz as shown in FIG.
On the other hand, when the same input signal is sampled at a period of 16 KHz and FFT analysis is performed in the same manner, as shown in FIG. 7C, the peak at 1 KHz does not change, and the peak at 10 KHz is the Nyquist frequency (in this case). , 8 KHz) and appears as a peak at 6 KHz. This phenomenon is called aliasing as described above. If the abnormal state of the device is determined based on the result of this frequency analysis, it will be determined that the motor has a 6 KHz periodic component that does not actually occur, the related mechanism and control abnormality, and the correct abnormality Cannot detect. This phenomenon is called aliasing as described above.

Therefore, in the present invention, data acquisition is performed at a slower frequency of 14 KHz, for example, without making the sampling period constant. When this data is subjected to FFT analysis, as shown in FIG. 7 (d), the 1KHz peak does not change, and when sampled at 16KHz, the peak at 6KHz is now 4KHz, and the peak frequency changes. I understand. In this case, the Nyquist frequency is 7 KHz.
In other words, when sampling the signal of the frequency component that is the reference for the operation of the apparatus, a plurality of different frequencies among frequencies that are less than twice the frequency band to be sampled (analysis frequency band) and higher than the upper limit of the analysis frequency band are selected. By sampling as a sampling frequency and comparing the results of the FFT analysis, the presence / absence of aliasing and its frequency can be specified. Further, from here, the folded frequency is restored to the original higher-order frequency by using the sampling frequencies of the plurality of sampling frequencies.

The high-order frequency component is restored by the following procedure.
First, data is sampled at a certain frequency fa, FFT analysis is performed, and the result is saved. This data is assumed to be DATA (a). Data is sampled at the next frequency fb, FFT analysis is performed, and the data is similarly stored. This data is DATA (b).
The peak frequencies of the two DATA (a) and DATA (b) are compared, and the peak value where the frequencies do not match is confirmed. When the frequencies of the peak values that do not coincide with each other are Pa and Pb, it can be determined that the combination of peak values satisfying the following relational expression (1) is an aliased frequency.
(Fa−fb) = (Pa−Pb) (1)
Further, the restored frequency Fr is as shown in the following equation (2).
Fr = (fa−Pa) = (fb−Pb) (2)

For example, in the examples of FIGS. 7A to 7D, fa: 16 KHz, Pa: 6 KHz, fb: 14 KHz, Pb: 4 KHz,
(Fa-fb) = 16 KHz-14 KHz = 2 KHz
(Pa-Pb) = 6KHz-4KHz = 2KHz
Thus, the expression (1) is satisfied.
Further, the higher-order frequency Fr to be restored is in accordance with the above equation (2),
Fr = 16KHz-6KHz = 10KHz (= 14KHz-4KHz)
To 10 KHz.

That is, in this case, it is determined that a peak value exists at 10 KHz, and the 10 KHz peak value is added (restored) as a higher-order frequency with respect to the fundamental frequency of 1 KHz. At the same time, the frequencies Pa and Pb appearing in the frequency analysis result are deleted due to aliasing (hereinafter referred to as “aliasing frequency”). Therefore, when performing an abnormality determination based on the frequency analysis result, for example, it is determined that an abnormal signal component of 10 KHz is detected with respect to the basic period of 1 KHz of the motor M1, and an event corresponding to the abnormal signal component of 10 KHz, For example, it is determined that “wear of the gear of the drive system of the motor M1” has occurred.
Moreover, although the case where only the frequency signal of 10 KHz was included as a high-order frequency was demonstrated, also when the frequency signal of two or more high-order frequencies is included, the above-mentioned restoration of a high-order frequency is applied similarly. Can do.
In the above description, restoration of a high-order frequency (peak value) has been described using two sampling periods. However, the number of variable sampling periods is not limited as long as it is two or more. For example, when three sampling periods are used, three arbitrary combinations of two sampling periods can be made. Therefore, the above-described processing is executed for each combination, and the frequencies obtained by the respective processes are compared and matched. By selecting the one to be or close to, it is possible to restore the frequency with higher accuracy. In addition, it is possible to process the combinations of sampling cycles to be employed by narrowing down to about two.
Also, a plurality of combinations of two sampling periods are provided, and for each combination of sampling periods, the above-described processing is executed to restore the frequencies, and more frequencies can be restored by deleting the aliasing frequencies. As a result, determination accuracy such as abnormality determination processing using the frequency analysis result can be improved.

In addition, by applying the frequency analysis method of the present invention, when the frequency band to be analyzed is, for example, 0 to 2 KHz, a sampling period of 4 KHz or more has been conventionally required, but the sampling period may be set to 4 KHz or less. Thus, the sampling processing time that is a load on the CPU 20 can be reduced.
That is, when the frequency band (analysis frequency band) of a signal component to be detected is known, the signal component can be detected by applying a sampling frequency that is twice or less the upper limit value of the frequency band. Therefore, the processing load on the CPU 20 that performs frequency analysis can be significantly reduced, and as a result, more advanced frequency analysis can be realized.

Here, considering the applicable objects of the present invention, there are devices, apparatuses, facilities, etc. that have little disturbance, do not require frequent abnormality detection, and do not cause a serious accident even if a failure occurs. It is done. For example, a driving system of a driving motor such as a photosensitive member (photosensitive drum or photosensitive belt) of an image forming apparatus that uses an electrophotographic process (electrostatic photographic process) such as the above-described color copying machine, a paper conveyance system, and a ventilation fan A driving system of a motor for driving the motor, a cooling fan driving system of a data processing device such as a personal computer device, and the like. Furthermore, it can be applied to frequency analysis of vibration.
The failure or abnormality to be detected can be applied to an event in which the peak value of the frequency component obtained by the frequency analysis described above is known in advance when the failure or abnormality occurs.

FIG. 8 is a flowchart illustrating an outline of the speed control of the motors M1 and M2 (the motor M1 interrupt process and the motor M2 interrupt process shown in FIG. 5) performed by the CPU 20.
After turning on the power of the apparatus, it is confirmed whether or not there is an instruction to start the motor (NO loop in step S1). When motor activation is requested by an instruction from the host system or the user (the result of step S1 is YES), the target motor rotation speed is instructed to the motor drive control means 23A (motor drive control unit 23; ASIC). (Step S2). For example, when the motors M1 and M2 are DC brushless motors, the rotation speed is often indicated by PWM duty. That is, the motor drive control unit 23 (ASIC) has a timer function that can arbitrarily set the cycle and duty, and the CPU 20 instructs the motor rotation speed by setting the duty value by software.
After instructing the motors M1 and M2 to rotate, the encoder pulse widths output from the pulse width measuring means 31 and 32 are periodically monitored, and the motor speeds of the motors M1 and M2 are determined based on the encoder pulse widths. (Step S3).

If the motor speed matches the target speed (the result of step S4 is YES), the PWM timer value (speed instruction value) instructed to the motor drive control means 23A is not updated and the state is continued. Let it be maintained. If a motor stop request is generated from the system (the result of step S5 is YES), the PWM timer value (speed instruction value) instructed to the motor drive control means 23A is set to a value corresponding to the speed “0”. (Step S6), the process returns to step S1 to return to the standby state.
On the other hand, when the motor speed does not match the target speed (the result of step S4 is NO), when it is determined that the motor speed is faster than the target speed (the result of step S7 is YES), the speed instruction In the direction of decreasing the value, the PWM timer value instructed to the motor drive control means 23A is updated (step S8), and the process returns to step S5. Thereby, the rotational speeds of the motors M1 and M2 are lowered, and the state can be grasped by monitoring the encoder pulse interval.
If it is determined that the motor speed is lower than the target speed (the result of step S7 is NO), the PWM timer value instructed to the motor drive control means 23A is updated so as to increase the speed instruction value ( Step S9) and return to step S5. As a result, the rotational speeds of the motors M1 and M2 increase, and the state can be grasped by monitoring the encoder pulse interval.

  Although there are various control theories for calculating the speed instruction value, it is not specified here, and only the basic control outline has been described. With this flow, the motor can maintain the target rotation speed. In the above description, the motor speed instruction means using the timer built in the motor drive control means 23A, which is an ASIC, has been described as an example. However, if the timer built in the CPU 20 or the like can be substituted, it can be used. There is no particular limitation on the means of use.

FIG. 9 is a flowchart for explaining the outline of the apparatus abnormality detection method.
First, after starting the motor, the encoder pulse width data is sampled at the first sampling period (step S21). It repeats until acquisition of the number of data necessary for FFT analysis is completed (NO loop of step S22), and frequency analysis is executed after completion (the result of step S22 is YES) (step S23). In this frequency analysis process, the amplitude and frequency are obtained by FFT analysis, and the results are stored (step S24).
Next, the second sampling period is changed (step S25), and the encoder pulse width data is sampled in the same manner as described above (the result of step S26 is YES, steps S21 and S22). Similarly, frequency analysis is performed and the result is stored (steps S23 and S24).

After the frequency analysis results with two different sampling periods are obtained (YES in step S26), the results are compared (step S27). The results of varying the sampling period are compared, and if all the detected frequencies match (the result of step S28 is NO), it can be determined that there is no aliased peak, and it is particularly necessary to perform higher-order frequency restoration. There is no.
If a frequency that does not match can be detected (the result of step S28 is YES), whether or not aliasing is determined is determined according to the above-described equation (1) (steps S29 and S30). Here, the frequency that does not match the condition (the result of step S30 is NO) cannot be determined to be aliased, so the high-order frequency restoration processing is not performed. It is possible to cope with this by sampling data again and detecting it again as necessary.
For the frequency peak that matches the condition (the result of step S30 is YES), the frequency peak is converted into an actual frequency and restored according to the above-described equation (2) (step S31). By performing these restoration processes on all the frequencies that do not match, it is possible to restore the peaks of all necessary frequency bands.
Next, the frequency analysis result obtained at that time is checked to check whether or not the motors M1 and M2 are in a controllable state (step S32). When the result of step S32 is NO, the target motor M1 , M2 is stopped (step S33), the occurrence of an abnormality is notified to the host system, the encoder pulse width data at that time is stored for abnormality analysis (step S34), and the processing at this time is terminated. .
When the motors M1 and M2 are controllable and the result of step S32 is YES, the processes at this time are terminated without executing steps S33 and S34.

By the way, according to the above-described embodiment, it is possible to determine the apparatus state using the result of restoring the higher-order frequency component. However, the frequency may be the same depending on the frequency analysis processing, but the absolute value of the peak is not necessarily completely restored. It is not always possible. There is a possibility that the actual peak frequency is close to the Nyquist frequency, and the peak absolute value is somewhat lowered by the filter (window) processing before FFT analysis.
Therefore, the device abnormality is determined using the data after restoration of the high-order frequency component, and when the result is that the device is abnormal, the motor to be checked is limited and the sampling cycle is performed. Can be dealt with by re-measuring with the frequency set to more than twice the frequency band to be analyzed. By limiting the drive motor, it is possible to create a state in which the number of motor control interrupt processes is reduced, and by increasing the sampling period, peak absolute values can be obtained without worrying about aliasing.
Also, in the device that monitors device deterioration, when the initial state of the device is acquired, measurement is performed with a high sampling period, FFT conversion is performed based on the measurement result, the device state is determined using the result, and the initial state is set. save. When determining the subsequent deterioration, the sampling period is lowered, the peak value frequency is complemented to the FFT conversion result using the means of the present invention, and the abnormality is determined using the frequency analysis result. When an abnormality is determined, the abnormal state can be accurately determined by increasing the sampling period again and performing a more accurate frequency analysis. In addition, during the period up to the abnormality determination, the load such as sampling interrupt processing can be reduced, so that the control load of the CPU or the like can be reduced in the operation of the apparatus during normal operation.

In the above description, the encoder pulse data is used for monitoring the motor state. However, the present invention is not limited to this. For example, the motor speed instruction value (PWM pulse value), torque instruction voltage, or current may be used. Is possible. In addition, any signal that can determine the motor operating state can be applied.
Furthermore, by mounting the technology relating to the present invention on an image forming apparatus having a frequency analysis function for monitoring the state of the apparatus, it is possible to realize the abnormality analysis function of the apparatus while suppressing the CPU load.

  This invention is based on the frequency of a peak value obtained by detecting a state variable of a target, such as a rotating body such as a motor or a device that generates periodic vibration, and performing FFT conversion on the state quantity. Thus, the present invention can be applied to an apparatus that performs apparatus abnormality determination.

1: Main control unit 2: Image formation control unit 3: Photoconductor motor control unit 4: Image formation process control unit 5: Intermediate transfer control unit 6: Intermediate transfer motor control unit 7: Intermediate transfer belt control unit 8: Primary transfer control Unit 9: Secondary transfer controller 10: Secondary transfer motor controller 11: Secondary transfer controller 12: Fixing controller 13: Paper feed controller 20: CPU (Central Processing Unit)
21: NVRAM (non-volatile memory) 22: ROM (read-only memory)
23: Motor drive control unit 23A: Motor drive control means 24, 25: Motor drive circuit 24A, 25A: Motor drive means 26, 27: Encoder 26A, 27A: Encoder pulse generation means 31, 32: Pulse width measurement means 33, 34 : Data sampling means 35, 36: Data sampling timing generation means 37, 38: Data storage means 39: Frequency analysis means M1, M2: Motor

JP 07-218565 A

Claims (9)

A signal processing device that detects a state quantity that periodically changes, and restores the frequency of a peak value obtained by performing FFT conversion on the state quantity,
First sampling means for sampling and storing the state quantity at a first sampling frequency fa higher than the upper limit value of the analysis frequency band and lower than twice the upper limit value;
Second sampling means for sampling and storing the state quantity at a second sampling frequency fb lower than the first sampling frequency fa and higher than the upper limit;
First frequency analysis means for obtaining a peak value frequency by performing FFT conversion on the first sample amount obtained by the first sampling means;
Second frequency analysis means for obtaining a peak value frequency by performing FFT conversion on the second sample amount obtained by the second sampling means;
The value of the frequency Pa of any first peak value acquired by the first frequency analyzing means and the frequency Pb of any second peak value acquired by the second frequency analyzing means is When there is a combination of the frequency Pa and the frequency Pb satisfying the following formula (1) in different combinations, the peak value frequency Fr is calculated based on the following formula (2), and the calculated peak value of the frequency Fr Is added to the peak value of the frequency acquired by the first frequency analysis means or the second frequency analysis means to restore the frequency component, while the values of the frequency Pa and the frequency Pb are different. When there is no signal that satisfies the following formula (1), the signal processing apparatus includes a frequency component restoring unit that does not restore the frequency component.
(Fa−fb) = (Pa−Pb) (1)
Fr = (fa−Pa) = (fb−Pb) (2)   A plurality of combinations of the first sampling frequency fa and the second sampling frequency fb are set, and for each combination, the first frequency analysis means, the second frequency analysis means, and the frequency component restoration means are provided. The signal processing apparatus according to claim 1, wherein the signal processing apparatus restores a plurality of frequency components by operating.   The signal processing apparatus according to claim 1, wherein the target is a motor, and the state quantity is a pulse interval time output from a pulse encoder provided on a rotation shaft of the motor.   The signal processing apparatus according to claim 1, wherein the target is a motor, and the state quantity is a control quantity that represents a speed target value given to the motor. Device abnormality determination means for determining device abnormality based on the peak value of the frequency obtained by the first frequency analysis means or the second frequency analysis means, including the frequency component restored by the frequency component restoration means. When,
A third sampling means for sampling and storing the state quantity at a third sampling frequency such that a Nyquist frequency is higher than the analysis frequency band when the apparatus abnormality determination means determines an apparatus abnormality;
5. The apparatus according to claim 1, further comprising third frequency analysis means for obtaining a peak value frequency by subjecting the third sample amount obtained by the third sampling means to FFT conversion. The signal processing device according to claim 1.   In the initial state of the apparatus, the processing result of the third frequency analyzing means is stored, and thereafter, the first sampling means and the second sampling means until the apparatus abnormality determining means determines the apparatus abnormality. 6. The signal processing apparatus according to claim 5, wherein the first frequency analysis means, the second frequency analysis means, and the frequency component restoration means are operated.   An image forming apparatus comprising the signal processing apparatus according to claim 1 and having a function of determining an abnormality occurring in the apparatus. A signal processing method for detecting a periodically changing state quantity of a target and restoring the frequency of a peak value obtained by performing FFT conversion on the state quantity,
A first sampling step of sampling and storing the state quantity at a first sampling frequency fa higher than an upper limit value of the analysis frequency band and lower than twice the upper limit value;
A second sampling step of sampling and storing the state quantity at a second sampling frequency fb lower than the first sampling frequency fa and higher than the upper limit;
A first frequency analysis step of obtaining a peak value frequency by performing FFT conversion on the first sample amount obtained in the first sampling step;
A second frequency analysis step of acquiring a peak value frequency by performing FFT conversion on the second sample amount obtained in the second sampling step;
The frequency Pa of any first peak value acquired in the first frequency analysis step and the frequency Pb of any second peak value acquired in the second frequency analysis step are: When there is a combination of the frequency Pa and the frequency Pb satisfying the following formula (1) in different combinations, the peak value frequency Fr is calculated based on the following formula (2), and the calculated peak value of the frequency Fr Is added to the peak value of the frequency obtained in the first frequency analysis step or the second frequency analysis step to restore the frequency component, while the values of the frequency Pa and the frequency Pb are different. When there is no signal that satisfies the following formula (1), the signal processing method includes a frequency component restoration step that does not restore the frequency component.
(Fa−fb) = (Pa−Pb) (1)
Fr = (fa−Pa) = (fb−Pb) (2)   A plurality of combinations of the first sampling frequency fa and the second sampling frequency fb are set, and for each combination, the first frequency analysis step, the second frequency analysis step, and the frequency component restoration step are performed. The signal processing method according to claim 6, wherein the signal processing method restores a plurality of frequency components.

Images