JP2005216213A - System and method for failure diagnosis - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a system and a method for failure diagnosis, which can simply perform failure diagnosis for control targets and controllers and specify failure parts from remote places without loss of time, and perform failure diagnosis during normal playback operations without stopping factory lines. <P>SOLUTION: The failure diagnosis system has a command generation part 11, which generates input data for a control target 12 including a motor, a controller 10, which is provided with a data storage part 13 stored with the input data and output data from the control target 12, and a control model 22 which simulates the control target 12. The system is constituted of a failure diagnosis part 20, which is provided with a failure determination part 23 that diagnoses failures for the control target 12 to specify failure parts from the control model 22, the input data and the output data, and a communication pass 14 that connects the controller 10 with the failure diagnosis part 20. The failure diagnosis part 20 is placed outside or inside of the controller 10. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

The present invention relates to a failure diagnosis system and a failure diagnosis method for identifying and diagnosing a failure portion to be controlled including a motor such as a robot, a servo, and an NC device.

When performing failure detection or failure diagnosis of a control target including a motor such as a robot, a servo, or an NC device installed in a user's factory in a remote place, there may be a case where the user side operator cannot sufficiently cope with it. This is because the structure of the control target and the control device and the operation system have become complicated due to the computerization accompanying the performance improvement of the machine tool, and it requires a huge amount of time for the workers on the user side who do not have the know-how of fault diagnosis. There was a case.
For this reason, it is necessary for workers on the manufacturer's side or workers on the manufacturer's service center near the user's factory to go to the site, but it takes time, labor and cost.
In addition, when a manufacturer's worker performs fault detection or fault diagnosis at the user's factory, it is necessary to stop the factory line where the control target is installed, which has an impact on user productivity. It was. Also, even when a large measuring instrument was brought into the factory line, it was necessary to stop the factory line as well. Even with physically narrow lines, it was sometimes difficult to bring measuring instruments.

To solve such a problem, for example, as shown in FIG. 16, a machine tool is interactively connected by connecting a management computer owned by the manufacturer and a machine tool owned by the user or a terminal device thereof through a commercial communication line. There are methods for performing periodic diagnosis, failure diagnosis, and repair of failure (see, for example, Patent Document 1).
In this method, when there is a periodic diagnosis request or failure diagnosis request for a machine tool from the user side, commercial communication from the management computer with the periodic diagnosis or failure diagnosis program owned by the manufacturer to the machine tool or its terminal A program for periodic diagnosis or failure diagnosis is sent via the line. Information necessary for periodic diagnosis or failure diagnosis generated as a result of executing this program in the machine tool is fed back to the management computer on the manufacturer side again via the commercial communication line.
Based on the information required for periodic diagnosis or failure diagnosis by the management computer, the dedicated analysis program is executed to automatically analyze the machine tool periodic diagnosis or failure diagnosis program, and the analysis result is sent to the user machine tool or its terminal. Send out via commercial communication line.
In addition, a display device for a periodic diagnosis and failure diagnosis program is provided in the machine tool on the user side or its terminal and the management computer on the manufacturer side, and it is necessary for periodic diagnosis, failure diagnosis and failure repair in an interactive manner via this display device. It is configured to transmit information.

Also, an apparatus is disclosed that stores a waveform of robot motion data and extracts a feature amount compared with a reference waveform in order to diagnose a robot motion waveform (see, for example, Patent Document 2).
In this apparatus, a data file that stores a history of robot motion data, a waveform feature extraction unit that extracts the feature amount of the captured motion data waveform compared to the reference waveform, and a model from the reference waveform and feature amount Standard data creation unit that creates standardized data, standard data adjustment unit that adjusts standard data according to operation data, and comments for waveform diagnosis that are set in advance by comparing the adjusted standard data and operation data The robot controller is connected to a workstation having a waveform diagnosis unit for performing waveform diagnosis of operation data while looking up the table and communicating with the robot controller.

JP 05-035751 A Japanese Unexamined Patent Publication No. 07-160323

However, the method of performing periodic diagnosis, failure diagnosis, and repair of machine tools in an interactive format requires the execution of periodic diagnosis or failure diagnosis programs and dedicated analysis programs, and the controlled objects and control devices change. There was a problem that could not be handled. In addition, the problem of having to stop the factory line when performing periodic diagnosis and failure diagnosis still remained.
Also, in the interactive format, when there is no user side worker, it is not possible to cope with the loss of time, the control target cannot perform any processing during playback operation, periodic diagnosis and failure diagnosis It was not possible to deal with the automation of and the identification of the failure part. Furthermore, it is necessary to always execute a complicated dedicated analysis program in order to execute failure diagnosis, and there is a problem that it requires CPU power and enormous calculation time.
On the other hand, in the method of storing the robot motion data waveform and extracting the feature value compared with the reference waveform, the feature value is extracted from the robot motion data by comparing with the reference waveform, and a comment table for waveform diagnosis is created. Although it is looked up, since it is only extraction of operation data and automatic judgment of its waveform, it has not been configured to be able to identify a faulty part as in the above method. Further, since the waveform of the operation data can be obtained only by a predetermined operation, it is not possible to cope with a periodic diagnosis or a failure diagnosis during playback operation by a work program created by a user-side worker. Further, like the above method, there is a problem that CPU power and enormous calculation time are required to extract data.

In view of this, the present invention can easily perform failure diagnosis and control of a control object or control device even from a remote location without loss of time, and perform failure diagnosis even during normal playback operation without stopping the factory line. It is an object of the present invention to provide a failure diagnosis system and a failure diagnosis method capable of performing the above.

The present invention is configured as follows in order to solve the above-described problems.
The invention according to claim 1 is a control device including a command generation unit that generates input data to a control target including a motor, a data storage unit that stores the input data and output data of the control target, and A fault diagnosis unit having a control model simulating a control target, and having a fault determination unit for diagnosing a fault of the control target from the control model, the input data, and the output data and identifying a fault part; and A failure diagnosis system comprising a communication path connecting an apparatus and the failure diagnosis unit.
The invention according to claim 2 is characterized in that the input data is a command to the controlled object, and the output data is a response of the controlled object to the command. It is.
According to a third aspect of the present invention, the data storage unit transmits the input data and the output data to the failure diagnosis unit, and the failure determination unit includes the output data and the control model for the input data. 3. The fault diagnosis system according to claim 1, wherein the fault diagnosis is performed by comparing the output data.
In the invention according to claim 4, when the failure determination unit diagnoses the control target as a failure,
The control object and the control model are divided into a plurality of modules according to their hardware or software configuration, and the data storage unit is configured to input data to the divided control object and each of the divided control objects. The output data is transmitted to the failure diagnosis unit, and the failure determination unit compares each output data with each output data of the divided control model for each input data, thereby the failure of the control target. 4. The fault diagnosis system according to claim 1, wherein a part is specified.
According to a fifth aspect of the present invention, the control device or the failure diagnosis unit includes a display screen that displays the input data, the output data, and the output data of the control model. It is a failure diagnosis system of description.
The invention according to claim 6 is characterized in that the control device or the failure diagnosis unit includes a process of specifying a failure part and a display screen for displaying the specified failure part in characters or graphics. The failure diagnosis system according to any one of claims 5 to 5.
The invention according to claim 7 is characterized in that the failure diagnosis unit includes a history storage unit, and the history storage unit records the specified failure part and displays the failure history on the display screen. The fault diagnosis system according to claim 5 or 6.
In the invention according to claim 8, when a plurality of candidates are listed when the failure diagnosis unit identifies the failure part to be controlled, a failure history stored in the history storage unit or a preset order The fault diagnosis system according to claim 7, wherein a fault site is specified by:
According to a ninth aspect of the present invention, the history storage unit stores a difference value between the output data and the output data of the control model at regular intervals, and a time difference between the difference values and a preset threshold value. 9. The fault diagnosis system according to claim 7, wherein a failure of the control target is predicted by comparing the value and a warning is displayed on the display screen.
According to a tenth aspect of the present invention, in a failure diagnosis method for a control system that controls a control target including a motor, input data to the control target and output data of the control target are stored, and the control target is simulated. The input data is input to the control model, and the output data and the output data of the control model are compared to diagnose a failure of the control target.
The invention according to claim 11 is a control system failure diagnosis method for controlling a control object including a motor, storing input data to the control object and output data of the control object,
The input data is input to a control model simulating the control object, the output data and the output data of the control model are compared to diagnose a failure of the control object, and when the failure is diagnosed, the control object And dividing the control model into a plurality of modules according to their hardware or software configuration, storing each input data to the divided control object and each output data of the divided control object, and the division Each of the input data is input to the control model, and the output data of the divided control target is compared with the output data of the divided control model to identify the faulty part of the control target. It is a feature.
According to a twelfth aspect of the present invention, in a failure diagnosis method for a control system for controlling a control target including a motor, a difference value between the output data and the output data of the control model is stored at a certain period, and the difference value is stored. The failure diagnosis method according to claim 11, wherein a failure of the control target is predicted by comparing a time difference between the time difference and a preset threshold value.

According to the failure diagnosis system according to claims 1, 2, and 3, it is necessary to easily perform failure diagnosis and identification of a failure to be controlled or a control device from a remote place or at a site without time loss, and to stop the factory line. No failure diagnosis can be performed even during normal playback operation. In addition, since only a simple waveform comparison is required, no special know-how is required, and the failure part can be automatically identified.
According to the fault diagnosis system of the fourth aspect, the control unit and the control model are divided, the input / outputs are compared and compared, and the process is repeated until the faulty part is specified, thereby narrowing down the faulty part in more detail. In addition, when dividing the control target and control model into multiple parts, by storing the input / output data of the software or hardware modules divided in units of modules, it is possible to separate and diagnose hardware failures and software failures It is possible to shorten the time for specifying the faulty part.
According to the failure diagnosis system of claim 5, the operator confirms the result of the location and condition where the failure part is specified by displaying the command and response of the input / output data and the output from the control model on the screen. it can. It can also be done intuitively when adjusting the threshold for failure determination.
According to the fault diagnosis system according to claim 6, the process of identifying the faulty part and displaying the specified faulty part on the screen in characters or graphics, the operator can immediately recognize the details of the fault, The time for preparing for the response can be shortened.
According to the failure diagnosis system of the seventh aspect, the failure site diagnosis can be easily set by storing the failure site, creating a database, and presenting it to the operator.
According to the failure diagnosis system according to claim 8, when a plurality of parts are listed as candidates while specifying the failure part, by giving priority to the specification of the failure part in the failure history or a preset order, The know-how accumulated by skilled workers can be used, and the failure part can be identified in a short time.
According to the failure diagnosis system of the ninth aspect, the failure prediction can be easily performed by grasping the time change of the difference value between the output from the control model and the output from the control target.
According to the failure diagnosis method of claim 10, the failure diagnosis of the control object and the control device and the identification of the failure part can be easily performed from a remote place or at the site without time loss, and it is not necessary to stop the factory line. Fault diagnosis can be performed even during playback operation. In addition, since only a simple waveform comparison is required, no special know-how is required, and the failure part can be automatically identified.
According to the failure diagnosis method of the eleventh aspect, failure parts can be narrowed down in more detail.
According to the failure diagnosis method of the twelfth aspect, it is possible to easily predict a failure of a control target or a control device from a remote place without loss of time, and it is not necessary to stop the factory line. In addition, failure prediction can be performed during normal playback operation.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

The first basic configuration of the present invention will be described with reference to FIG. 1, and the flow of processing will be described with reference to FIG.
Here, the failure detection unit 20 on the remote manufacturer side performs the failure detection and failure diagnosis method of the control object 12 including the motor arranged on the user side and the control device 10 that controls the control object. The process flow will be described in three steps. Each step corresponds to the first to third steps of the processing flow shown in FIG.

(1) First Step As shown in FIG. 1, command data such as a position command and a speed command is input from the command generator 11 in the control device 10 to the controlled object 12 (hereinafter, this input data is used as actual machine input data). And).
The control object 12 operates based on actual machine input data, and outputs response data such as a position response and a speed response to the command generation unit 11 (hereinafter, this output data is referred to as actual machine output data).
The actual machine input data and the actual machine output data are stored in the data storage unit 13 every predetermined period (for example, several tens of ms) during the operation of the controlled object 12 (hereinafter, the actual machine input data and the actual machine output data will be described). Are used as input / output data).
Here, when the control object 12 has a structure that can be divided into a plurality of modules, input / output data of each module is stored in the data storage unit 13 in advance. For example, as shown in FIG. 5, when three modules 12a, 12b, and 12c are connected in series, the first actual machine output data of the module 12a is used as the actual machine input data to the module 12b, and the module 12b The second actual machine output data is used as actual machine input data to the module 12c, and the actual machine input data from the command generator 11 and the actual machine output data 12a, 12b, 12c of the module to be written are used as third actual machine output data of the module 12c. Acquired and stored as input / output data.

(2) Second step The input / output data is transmitted through a communication line 14 such as a telephone line or a commercial communication line such as the Internet, a dedicated line such as a LAN, or wireless communication, for example, a predetermined cycle (for example, 1 The data is transmitted to the failure diagnosis unit 20 arranged on the manufacturer side every work or hour). Here, the model information of the control object 12 and the contents of the work program are attached to the data initially transmitted to the failure diagnosis unit 20, and the control model 22 in the failure diagnosis unit 20 is referred to as the control object 12. Make the same setting.

(3) Third Step The failure diagnosis unit 20 includes a control model 22 that simulates the control object 12 and a failure determination unit 23 that detects and diagnoses a failure. Based on the input / output data and the control model 22 The failure determination unit 23 performs failure detection and diagnosis.
In the failure diagnosis, the actual machine input data included in the input / output data is input to the control model 22, the response data output from the control model 22 (hereinafter, this output data is referred to as model output data) is obtained, and the failure determination unit 23 This is done by obtaining the difference value with the actual machine output data included in the input / output data.
Since the control model 22 reflects the motion characteristics of the motor including the characteristics of the mechanism of the controlled object 12, the same output as that of the controlled object 12 is obtained when the same command is input.
For this reason, the difference value between the actual machine output data and the model output data is small when there is no failure in the controlled object 12, and the difference value is large when there is a failure. Therefore, the absolute value of this difference value is set in advance. Compared with the threshold value, if it is larger than the threshold value, it is determined as a failure, so that the failure can be easily detected without requiring know-how and a complicated analysis program.

Further, as described above, when the control target 12 is composed of a plurality of modules, as shown in FIG. 6, the control model is also divided into a plurality of parts in advance, and the actual machine input data is input to the control model 22a. First model output data output from the control model 22a is obtained, an absolute value of a difference value from the first actual machine output data is obtained by the first data comparison unit 231 in the failure determination unit 23, and set in advance. Compare to the threshold value.
Similarly, the second actual machine input data is input to the control model 22b, the second model output data output from the control model 22b is obtained, and the second data comparison unit 232 in the failure determination unit 23 receives the second model output data. The absolute value of the difference value from the actual machine output data is obtained, compared with a preset threshold value, the third actual machine input data is input to the control model 22c, and the third model output from the control model 22c is output. Output data is obtained, and the third data comparison unit 233 in the failure determination unit 23 obtains the absolute value of the difference value from the third actual machine output data, and compares it with a preset threshold value.
As described above, when the control target 12 is composed of a plurality of modules, it is possible to complete the specification of the faulty part when the fault is detected.

The second basic configuration of the present invention will be described with reference to FIGS. 3, 5 and 6, and the flow of processing will be described with reference to FIG.
In the first basic configuration, the control target 12 is configured from a plurality of modules in advance. In the second basic configuration, the control target 12 and the control model 22 are broken down into modules in the process of fault diagnosis. A process for identifying a part is provided.
The failure detection and failure diagnosis method will be described in 7 steps. Each step corresponds to the first to seventh steps of the processing flow shown in FIG. Since the process up to the failure detection is the same as that in FIG. If a failure is detected, the process proceeds to the next fourth step.

(4) Fourth step When the failure determination unit 23 diagnoses a failure, the process enters the process of identifying the failure part. As shown in FIG. 3, a command for dividing the control object 12 and the control model 22 into N pieces (N: natural number) is sent from the failure determination unit 23. When the division command is sent, the location where the response data of the control target 12 and the control model 22 is acquired is changed to N location, and the internal state quantities of the control target 12 and the control model 22 are compared in more detail.
Specifically, N is determined by the number of components that constitute the control object 12 and that have detectors attached so that the output can be detected. The number is transmitted from the control device 10 to the failure diagnosis unit 20 in the second step.
In addition, in order to reduce the data transfer amount between the control device 10 and the failure diagnosis unit 20 as much as possible, the control object 12 and the control model 22 are first divided into N pieces, and then a plurality of modules are further specified in the process of specifying the failure part in detail. Take the procedure of dividing into
As an example of dividing into three hardware modules, the control object 12 is divided into a first control object 12a to a third control object 12c as shown in FIG. Similarly, the control model 22 is divided into a first control model 22a to a third control model 22c as shown in FIG. Hereinafter, a case where the control object 12 and the control model 22 are divided into three will be described.

(5) Fifth Step The input / output data of each of the control objects 12a, 12b, and 12c divided into three is stored in the data storage unit 13 by executing the operation again in response to the operation command from the host. .
Alternatively, every time the operation is executed, all the input / output data of the control objects 12a, 12b, and 12c is stored, and only the response data of the control object 12c that is the output terminal is detected before the failure diagnosis unit 20 before the failure detection in the second step. Even after the failure is detected, the three response data of each control object 12 stored in advance can be transmitted to the failure diagnosis unit 20 to avoid re-operation of the control object 12 after the failure occurs. good.

(6) Sixth Step Similar to the second step, three pieces of input / output data output from the control objects 12a, 12b, and 12c are transmitted to the failure diagnosis unit 20 via the communication path 14.

(7) Seventh Step Similar to the third step, the identification of the failure part of the controlled object 12 is repeated based on the input / output data and the control model 22 divided into three.
Specifically, as shown in FIG. 6, the actual machine input data included in the input / output data is input to the first control model 22a, and the first model output data output from the first control model 22a is obtained. . Further, the first model output data is input to the second control model 22b, and the second model output data output from the second control model 22b is obtained. Further, the second model output data is input to the third control model 22c, and the third model output data output from the third control model 22c is obtained.
In the failure determination unit 23, the absolute value of the difference value between each of the three model output data and the three actual machine output data included in the input / output data is taken and compared with a preset threshold value. Identify the site.

When there are a plurality of modules whose absolute value of the difference value exceeds the threshold value and the failure part cannot be specified, the failure part is specified for a module having a high priority according to the priority order described later, and the fourth step is performed. Returning, the module with higher priority is further divided, and the process for specifying the faulty part is repeated.
When the failure part can be identified, the failure diagnosis is terminated, and the operator who operates the control device 10 on the user side and the operator who operates the failure diagnosis unit 20 on the manufacturer side with the display unit to be described later are in trouble. Inform the site and replacement method.

Since the above processing can be performed by comparing the response waveforms of the actual machine and model to normal operation commands, there is no need to perform special operations for fault detection or fault diagnosis, and the factory line is stopped until the fault location is specified. Failure diagnosis can be performed from a remote location. Also, since it is a simple waveform comparison, it can be easily automated. Furthermore, the amount of communication data between the control device 10 and the failure diagnosis unit 20 can be reduced until normal failure detection.

A first specific embodiment of the present invention will be described with reference to FIGS. In this embodiment, the control target is a vertical articulated 6-axis robot 30 that drives each axis by a motor via a speed reducer as shown in FIG. 7, and the robot 30 has a welding jig 31 at its hand. An operation pendant 15 for an operator to operate the robot 30 is connected to the control device 10 that controls the robot 30.
In this embodiment, the failure part of the robot 30 is a speed reducer connected to the fourth axis motor from the root, and assuming that this speed reduction gear skipped due to wear, this failure part is regarded as the remote manufacturer's side. The process specified by the failure diagnosis unit 20 arranged in FIG. For the sake of simplification, the description will be given focusing on the control system for the fourth joint axis.
Further, as shown in FIG. 8, the control object 12 corresponding to each axis of the robot 30 includes a motor control unit 121 that generates a torque command or a current command for the motor 122 from an operation command output from the command generation unit 11, and a torque command or A motor 122 driven by a current command; a first detector 123 that detects a state quantity of the motor 122; a speed reducer 124 that doubles the driving force of the motor 122; a mechanism 125 connected to the speed reducer 124; The second detector 126 is configured to detect 125 state quantities.
As described in the processing flow of FIG. 4, the fault site identification process is divided into seven steps. In the first to third steps, fault detection and rough fault site identification are executed, and in the fourth to seventh steps, detailed processing is performed. Execute fault location identification. Details will be described below.

(1) First Step First, the process from the occurrence of no failure to the occurrence of the failure and the detection of the failure will be described.
For example, when welding work is performed using a robot arranged on the user side, the robot 30 sends a position command stored in the control device 10 in advance to the control object 12 from the command generation unit 11, so that the robot 30 performs the welding work as taught. I do.
As shown in FIG. 8, the position command is sent to the motor control unit 121 in the controlled object 12, and the motor control unit 121 outputs a torque command or a current command to the motor 122 that is a drive source.
The output of the motor 122 is a position and becomes an input to the speed reducer 124. The output of the speed reducer 124 is also a position and becomes an input to the mechanism 125. The output of the mechanism 125 also becomes a position and can be detected by the second detector 126. The output of the second detector is returned to the command generator 11.
Here, the position command for the controlled object 12 is set as actual machine input data, and the position response which is the output of the second detector 126 is set as actual machine output data. The actual machine input / output data is stored in the data storage unit 13 at predetermined intervals (here, the calculation cycle time of the torque command or current command) as input / output data for six axes.

(2) Second step Input / output data is transmitted to a remote place at a predetermined fixed period (here, one workpiece is welded) different from the first step by a communication path 14 such as the Internet. It is transmitted to the failure diagnosis unit 20 on the manufacturer side. Here, in order to set the control model 22 in the failure diagnosis unit 20 to the same setting as the control object 12 in the data transmitted first, the model information of the robot (inertial value of each of the six axes, calculation cycle time, etc.) The contents of the work program are attached.

(3) Third step The failure diagnosis unit 20 creates a control model 22 simulating the control target 12 as shown in FIG. 9 at the start of failure diagnosis. The control model 22 includes a motor control model 221 that simulates the motor control unit 121, a motor model 222 that simulates the motor 122, a speed reducer model 224 that simulates the speed reducer 124, and a mechanism 125. It is comprised with the mechanical model 225 which simulated.
Here, in order to reduce the amount of calculation, the mechanical model 225 is treated as a rigid body until a failure is detected, and after the failure is detected, it is treated as a multi-inertia model in order to identify the failed part. Therefore, it calculates as a rigid body model here.
By inputting actual machine input data θref to the control model 22, model output data θfbm that is an output of the motor model 222 can be obtained. This calculation is performed by the following (Expression 1) to (Expression 3). (Equation 1) is the same calculation as the software calculated by the motor control unit 121 of the actual machine, and the control parameters may be the same, so that a complicated analysis algorithm is unnecessary.

Trefm = J * Kv * (Kp * (θref − θfbm) − • θfbm) (Equation 1)
here,
Trefm: Model torque command J: Motor rotor inertia + mechanism inertia (including reduction gear)
Kv: Speed loop gain
Kp: Position loop gain θref: Position command from command generator 11 (actual machine input data)
θfbm: Model position response (model output data)
・ Θfbm: Model speed response

・ Θfbm = Σ (dt * Trefm / J) (Formula 2)
here,
Σ: integral
dt: Calculation cycle time

θfbm = Σ (dt * · θfbm) (Equation 3)

Here, the calculated value of the inertia J of the rigid body model may be periodically sent to the failure diagnosis unit 20 as a result of the calculation by the motor control unit 121 of the actual machine, or the actual machine position response θfb included in the input / output data. From the above, it may be calculated on the failure diagnosis unit 20 side. Similarly, for viscous friction, the state value of the actual machine and the model may be made more consistent by periodically sending the friction value measured in the actual operation to the failure diagnosis unit 20 side.
In the failure diagnosis, the calculation of (Equation 1) to (Equation 3) is performed based on the position command θref which is actual machine input data in the input / output data to obtain the position response θfbm of the control model 22, and the failure determination unit 23 is performed by obtaining a difference value Δθfb between the position response θfb of the actual machine which is the actual machine output data and the position response θfbm of the control model 22 in the input / output data.
When there is no failure, the difference value Δθfb is reduced, and when there is a failure, the difference value Δθfb is increased. Therefore, the absolute value of the difference value Δθfb is compared with a preset threshold value. If it is larger, it is possible to detect the failure by determining that it is a failure.
When the failure is diagnosed, the occurrence of the failure is notified to the worker who operates the manufacturer-side failure diagnosis unit 20 and the user who operates the user-side control device 10 by the display unit described later. The process proceeds to the failure part specifying mode after the fourth step. If no failure is detected, the first to third steps are repeated.
In the present embodiment, the actual machine position response θfb deviates from the normal value due to the tooth skip of the fourth-axis real machine speed reducer 124, so that the absolute value of the difference value Δθfb becomes larger than a preset threshold value. A failure on the fourth axis is detected.

(4) Since the failure is diagnosed by the fourth step failure determination unit 23, the failure part is identified and the control unit 12 and the control object 12 are controlled by the division command from the failure determination unit 23 as shown in FIGS. The model 22 is divided into three.
When the control object 12 and the control model 22 are divided by the division instruction, the division is performed in units of software or hardware modules.

When there are a number of minimum unit modules whose output can be detected in the control object 12, first, in order to roughly identify the faulty part, first, a module in units of parts in which a plurality of minimum unit modules are combined. create. The failure part is roughly identified by comparing the output of the module for each part and the output of the model, and the module for the specified part is again divided into modules of the minimum unit to identify the detailed failure part.
In order to execute this method, the software or hardware module to be divided in advance is determined according to the previous failure location accumulated in the manufacturer's database and the manufacturer's know-how, and the data storage Held in the unit 13. Further, it is attached to data (such as robot model information and work program contents) transmitted to the failure diagnosis unit 20 at the beginning of the above-described failure diagnosis, and stored in the failure determination unit 23.
Also, when multiple pieces of software are installed on a single hardware module, it is divided into software module units, and each input / output variable is monitored, resulting in parameter settings and coding errors in each control module. Abnormalities can also be detected.

Here, when the control object 12 is divided by its software configuration or hardware configuration, a “motor control unit” configured by software and hardware and a “motor, speed reducer, and mechanism” configured only by hardware. It can be divided into two parts.
Furthermore, when divided into parts, “motor, speed reducer and mechanism” are divided into two parts, “motor” and “speed reducer and mechanism”, which are divided into three parts.
In the control target 12, the first control target 12a in FIG. 5 is a motor control unit 121 composed of an electronic board and a program, and the second control target 12b in FIG. 5 is a motor 122 and a first detector which are movable parts. 123, and the third controlled object 12c in FIG. 5 is the speed reducer 124, the mechanism 125, and the second detector 126.
Further, in the control model 22, the first control model 22a in FIG. 6 is a motor control model 221 that simulates the motor control unit 121, the second control model 22b in FIG. 6 is a motor model 222, and the first control model 22 in FIG. 3 is a multi-inertia model of the speed reducer model 224 and the mechanical model 225.

(5) Fifth Step The input / output data of each module of the control object 12 divided into three is stored by executing the operation again in response to the operation command. Here, the first actual machine output data, which is the output data of the first controlled object 12a, is a torque command and can be obtained by (Equation 1). The second actual machine output data that is the output data of the second controlled object 12b is a motor position response, and can be obtained by measuring the output of the encoder that is the first detector 123. The third actual machine output data, which is the output data of the third controlled object 12c, is a mechanical position response, and is a value obtained by second-order integration of the value of the acceleration sensor arranged at the hand of the robot as the second detector 126. Etc. The hand position of the robot can be obtained not only by the acceleration sensor but also by a camera image or a laser sensor arranged outside.

(6) Step 6 Similar to step 2, the communication path 14 diagnoses the failure of the three input / output data (actual machine input data and first to third real machine output data) output from each part of the control target 12. To the unit 20.

(7) Seventh Step Similar to the third step, the failure part of the control target 12 is specified based on the input / output data and the control model 22 divided into three.
Specifically, actual machine input data of input / output data is input to the divided first control model 22a of the control model 22, and a torque command output from the first control model 22a is used as first model output data. Ask.
Further, the first model output data is input to the second control model 22b obtained by dividing the control model 22, and the motor model position response output from the second control model 22b is obtained as the second model output data.
Further, the second model output data is input to the third control model 22c obtained by dividing the control model 22, and the mechanical model position response output from the third control model 22c is obtained as the third model output data.
These torque command, motor model position response, and mechanical model position response can be obtained by (Expression 1), (Expression 3) and the following (Expression 4), (Expression 5), and (Expression 6).

・ Θfbm = Σ {dt * (Trefm−K * (θfbm−Xfbm)) / Jm} (Formula 4)
here,
K: Reducer spring constant Xfbm: Mechanical model position response
Jm: Rotor inertia of motor

Xfbm = Σ {dt * K * (θfbm−Xfbm) / JL} (Formula 5)
here,
・ Xfbm: Mechanical model speed response
JL: Mechanism inertia (including reduction gear)

Xfbm = Σ (dt * · Xfbm) (Expression 6)

In the failure determination unit 23, the failure part is specified by taking the difference value between the first to third model output data and the first to third actual machine output data included in the input / output data.
The first data comparison unit 231 in the failure determination unit 23 obtains the absolute value of the difference Δtref between the first actual machine output data and the first model output data, and compares it with a preset torque command difference threshold value. Thus, it is determined whether or not there is a failure part in the first control object 12a.
Further, the second data comparison unit 232 obtains an absolute value of the difference Δθfb between the second actual machine output data and the second model output data, and compares it with a preset motor position response difference threshold value. Then, it is determined whether or not there is a failure part in the second control object 12b.
Further, the third data comparison unit 233 obtains the absolute value of the difference ΔXfb between the third actual machine output data and the third model output data, and compares it with a preset mechanical position response difference threshold value. Then, it is determined whether or not the third controlled object 12c has a failure part.

Next, FIG. 12 will be described by showing the state quantities of each control object 12 and each control model 22.
As shown in FIG. 12A, the first actual machine output data (actual machine torque command) input to the first data comparison unit 231 and the first model output data (model torque command) substantially coincide with each other. Therefore, as shown in FIG. 12B, the absolute value of the difference ΔTref is smaller than the torque command difference threshold value.
Further, as shown in FIG. 12C, the second actual machine output data (actual machine motor position response) and the second model output data (motor model position response) input to the second data comparison unit 232 are Since they are almost the same, the absolute value of the difference Δθfb becomes smaller than the motor position response difference threshold as shown in FIG.
On the other hand, as shown in FIG. 12 (e), the third actual machine output data (actual machine mechanical position response) and the third model output data (mechanical model position response) input to the third data comparison unit 233 are reduced gears. As shown in FIG. 12 (f), the absolute value of the difference ΔXfb becomes larger than the mechanical position response difference threshold value, and the speed reducer is identified as the failure part.
In this way, since only a simple waveform comparison by matching the outputs of the actual machine and the model is required, no special know-how or complicated analysis program is required, and the failure part can be easily and automatically identified in a short time.

In this embodiment, failure diagnosis is performed from input / output data for each operation. However, by accumulating data for a longer period, a failure due to secular change can be predicted. An example thereof will be described with reference to FIGS.
The rough processing flow is divided into five steps as shown in the processing flow of FIG.
For example, when it is predicted that the reducer 124 has run out of grease, as shown in the fourth step of FIG. 14, the friction value of the actual reducer 124 is measured every day and transmitted to the failure diagnosis unit 20. A difference value between the friction value of the speed reducer model 224 of the control model 22 and the friction value of the actual speed reducer 124 is stored in the history storage unit 24. As shown in the fifth step, when the absolute value of the difference between the difference value measured on the next day and the difference value stored one day ago becomes larger than a preset difference threshold value, When it is determined that the decelerator 124 is almost out of grease, a failure occurs to a worker who operates the control device 10 on the user side and a worker who operates the failure diagnosis unit 20 on the manufacturer side by a display unit described later. Informs and requests grease at the same time.

  In addition, the example in which the failure diagnosis unit 20 is installed in the remote place of the control device 10 on the user side has been described so far, but the failure diagnosis unit 20 is built in the control device 10 on the user side or arranged in the vicinity thereof. It is good also as a structure to be performed. By disposing the fault diagnosis unit 20 on the user side, a user operator can perform a fault diagnosis without sending the fault diagnosis unit 20 to a remote fault diagnosis unit 20 if the fault diagnosis is simple. Information such as the division method and threshold values of the control object 12 and the control model 22 when specifying the faulty part can be acquired from the manufacturer's database as necessary, thereby reducing the communication time.

Further, in the above-described example, the third data comparison unit 233 identifies one failure site in the seventh step of the failure site identification process of FIG. 4. For example, the first data comparison unit 231 and the first data comparison unit 231 In some cases, two data comparison units 232 may simultaneously detect a failure. In this case, according to the previous failure rate and the manufacturer's know-how accumulated in the manufacturer's database, the failure location is specified in advance for those with a high frequency of failure or those that are seriously dangerous due to the occurrence of failure. Priorities are set and stored in the data storage unit 13. Then, the failure part can be specified in a short time by executing the comparison process of the module having a higher priority first.

A second specific embodiment of the present invention will be described with reference to FIG.
This is an example in which the input / output data, the output data of the control model 22, the difference value between the actual machine output data and the model output data, and the threshold value are displayed on the display screen on the operation pendant 15 in FIG. These values may display real time data, but the stored data may be displayed later in a format according to the user's request.
FIG. 15A shows an example in which the waveform of the difference (FIG. 12F) between the actual output data and the model output data of the third controlled object (reduction gear) in the first specific example described above is displayed. Indicates. By displaying these waveforms, the operator who operates the control device 10 on the user side can intuitively confirm the identified failure part, the threshold value used for failure detection, and the like.
In addition, on the display screen on the operation pendant 15, the process of identifying the failure site and the identified failure site are displayed on the screen in characters or graphics. Figure 15 (b), the show an example of displaying the site of the reduction gear 124 which is a specific failure area in the above-mentioned embodiment 1 in figure, also the cause characters.
By displaying the failure part and the cause on the screen in this manner, the operator operating the control device 10 on the user side can intuitively recognize the failure part, and can immediately deal with the arrangement of the failed part. The time required for recovery from a failure can be shortened.

Further, when an operator who operates the failure diagnosis unit 20 on the manufacturer side performs an operation of adjusting the threshold value for failure determination, a display screen connected to the failure diagnosis unit 20 installed on the manufacturer side ( By performing the same display on (not shown), the adjustment operation can be performed intuitively.
Further, the operator operating the failure diagnosis unit 20 on the maker side goes back to the diagnosis process while displaying the difference value and the threshold value as shown in FIG. The threshold value can be readjusted.

Further, the identified failure part is also stored in the history storage unit 24 in the failure diagnosis unit 20. By displaying the previous failure history (for example, the failure part and the number of failures) on the display screen of the control device 10 or the failure diagnosis unit 20, it can be used as a reference when setting priorities for a plurality of failure part candidates, Priorities can be set without the know-how of skilled workers.

Since the present invention can determine the difference between the actual machine and the model by inquiring the output from the control model and the input / output data of the control target including the motors such as robots, servos, and NC devices installed in remote places, It can also be applied to the use of remote parameter identification and adjustment.

First basic configuration diagram of the present invention The flowchart showing the 1st basic process of this invention Second basic configuration diagram of the present invention Flowchart showing the second basic processing of the present invention The block diagram showing the 2nd basic composition of the present invention The block diagram showing the 2nd basic composition of the present invention Outline drawing showing a first specific embodiment of the present invention First block diagram representing a first specific embodiment of the present invention Second block diagram representing the first specific embodiment of the present invention Third block diagram showing the first specific embodiment of the present invention. FIG. 4 is a fourth block diagram showing the first specific embodiment of the present invention. The figure which shows the controlled variable and the state quantity of a control model in the 1st specific Example of this invention. The 5th block diagram showing the 1st concrete example of the present invention. The flowchart showing the processing in the first specific embodiment of the present invention. The figure showing the 2nd specific Example of this invention Diagram showing conventional method

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Control apparatus 11 Motor control part 12 Control object 12a, 12b, 12c 1st control object, 2nd control object, 3rd control object 121 Motor control part 122 Motor 123 1st detector 124 Decelerator 125 Mechanism 126 1st 2 detector 13 data storage unit 14 communication path 15 operation pendant 20 fault diagnosis unit 22 control model 22a, 22b, 22c, first control model, second control model, third control model 221 motor control model 222 motor model 224 Decelerator model 225 Mechanical model 23 Failure determination unit 231 First data comparison unit 232 Second data comparison unit 233 Third data comparison unit 24 History storage unit 30 Robot 31 Welding jig

Claims (12)

A control device including a command generation unit that generates input data to a control target including a motor, and a data storage unit that stores the input data and output data of the control target;
A fault diagnosing unit having a control model that simulates the control target, and comprising a fault determination unit that diagnoses a fault of the control target from the control model, the input data, and the output data, and identifies a fault site;
A failure diagnosis system comprising a communication path connecting the control device and the failure diagnosis unit. The failure diagnosis system according to claim 1, wherein the input data is a command to the control target, and the output data is a response of the control target to the command.   The data storage unit transmits the input data and the output data to the failure diagnosis unit, and the failure determination unit compares the output data with the output data of the control model for the input data, The failure diagnosis system according to claim 1 or 2, wherein failure diagnosis is performed. The failure determination unit, when diagnosing the control object as a failure, divides the control object and the control model into a plurality of modules according to their hardware or software configuration, and the data storage unit is divided. Each input data to the control target and each output data of the divided control target is transmitted to the failure diagnosis unit,
The failure determination unit identifies the failure part to be controlled by comparing each output data with each output data of the divided control model for each input data. The failure diagnosis system according to any one of claims 1 to 3. The fault diagnosis system according to claim 1, wherein the control device or the fault diagnosis unit includes a display screen that displays the input data, the output data, and the output data of the control model. The fault diagnosis system according to claim 1, wherein the control device or the fault diagnosis unit includes a process of specifying a faulty part and a display screen for displaying the specified faulty part in characters or graphics.   The failure diagnosis unit according to claim 5 or 6, wherein the failure diagnosis unit includes a history storage unit, and the history storage unit records the specified failure part and displays the failure history on the display screen. system. The failure diagnosis unit, when a plurality of candidates are listed when specifying the control target failure site,
8. The fault diagnosis system according to claim 7, wherein a fault site is specified based on a fault history stored in the history storage unit or a preset order. The history storage unit stores a difference value between the output data and the output data of the control model for each predetermined period, and compares the time difference of the difference value with a preset threshold value, The failure diagnosis system according to claim 7 or 8, wherein a failure to be controlled is predicted and a warning is displayed on the display screen. In a control system failure diagnosis method for controlling a control object including a motor,
Storing input data to the controlled object and output data of the controlled object;
Input the input data to a control model simulating the control object,
Comparing the output data with the output data of the control model to diagnose a failure of the controlled object. In a control system failure diagnosis method for controlling a control object including a motor,
Storing input data to the controlled object and output data of the controlled object;
Input the input data to a control model simulating the control object,
Comparing the output data with the output data of the control model to diagnose a failure of the controlled object,
When a failure is diagnosed, the control object and the control model are divided into a plurality of modules according to their hardware or software configuration,
Storing each input data to the divided control object and each output data of the divided control object;
Input each input data to the divided control model,
A failure diagnosis method characterized by comparing each output data of the divided control object with each output data of the divided control model to identify a failure part of the control object. In a control system failure diagnosis method for controlling a control object including a motor,
The difference value between the output data and the output data of the control model is stored at regular intervals, and the failure of the control target is predicted by comparing the time difference of the difference value with a preset threshold value. 13. The fault diagnosis method according to claim 11 or 12, wherein:

Images