TW201925942A - Intelligent diagnosis system and method - Google Patents

Abstract

Disclosed are an intelligent diagnosis system and method. The system comprises a main system, a subsystem, and a diagnosis prediction card. The main system comprises a master control card of the main system, a main hub card, and multiple main data cards. The main hub card is connected to a sensor corresponding to the main system, and is used for receiving measurement data of the sensor and sending the measurement data to the main data cards of the main system for computing. The subsystem comprises a master control card of the subsystem, a sub hub card, and multiple sub data cards. The sub hub card is connected to a sensor corresponding to the subsystem, and is used for receiving measurement data of the sensor and sending the measurement data to the sub data cards of the subsystem for computing. The diagnosis prediction card is connected to the master control card of the main system and the master control card of the subsystem, and is used for regularly obtaining intermediate operating data of the main data cards or the sub data cards, and performing fault prediction according to the received intermediate operating data, and feeding back the prediction result to the master control card of the main system or the master control card of the subsystem. Real-time memory operations can be performed on data measured by the sensors, and time-consuming frequency-IO operations on batch data can be avoided.

Description

Intelligent diagnosis system and method

The invention relates to the technical field of equipment fault diagnosis, in particular to an intelligent diagnosis system and method.

With the advancement of science and technology and the development of computer science, various devices have begun to evolve toward integration, complexity, and intelligence, making machines or devices closer to or meeting the operating habits and functional requirements of natural people. However, similarly, the intelligence of machines and equipment is based on the complexity of machine design, manufacturing, and operation. The operation of intelligent equipment is often accompanied by various problems. The appearance of problems is often "lagging", that is, once the machine shows problems that people can find, the machine is already "disease-ridden". At this time, the problem positioning, repair, and maintenance are relatively cumbersome, usually relying on the experience of the engineer Troubleshoot, unable to quickly locate the problem, quickly restore the normal operation of the equipment, and unable to complete the inheritance of technology.

The common methods used by the industry to solve the above problems are to add various sensing devices to detect the temperature, humidity, pressure, current and other factors in the device in real time. Since the occurrence of faults has the characteristics of "randomness" and "accidentality", it is necessary Store the detected data in real time, which greatly increases the storage time and storage space, that is, the space complexity and time complexity will increase linearly, or even increase exponentially; when a fault occurs, it often only needs The data in the first few seconds or the first few minutes before the fault occurs, a large amount of data is garbage or redundant data, which not only takes up a lot of hardware space, increases the hardware cost, but also greatly increases because of the high frequency I / O operations of the computer The running time affects the overall performance of the device.

In addition, the detection and processing of faults in the industry are generally after the fault occurs, the shutdown process is not adopted, and some fault prediction mechanisms are not adopted to avoid the occurrence of simple or repeated faults. Although the fault problem can be solved eventually, it will cause MTTR (Mean Time To Repair, the average time before recovery) is large, which affects the production schedule and reduces the productivity of the equipment.

In view of the problems existing in the prior art, the present invention provides an intelligent diagnosis system and method that can realize fault pre-judgment and fault pre-processing, speed up production progress, and improve equipment yield.

In order to solve the above technical problems, the present invention provides an intelligent diagnosis system, including:

The main system includes a main system main control board, a main hub board, and a plurality of main data boards. The main hub board is connected to a sensor corresponding to the main system, and is used to receive the sensor ’s Measuring data and sending to each of the main data boards of the main system for calculation;

Sub-system, including sub-system master control board, slave hub board and multiple slave data boards, the slave hub board is connected to the sensor corresponding to the subsystem, and used to receive the measurement of the sensor Data, and sent to each of the slave data boards of the subsystem for calculation;

The diagnosis prediction board is connected to the main system main control board and the sub-system main control board, and is used to periodically obtain the intermediate operation data of the master data board or the slave data board card and receive Of the intermediate operation data to perform fault prediction, and feed back the prediction results to the main system main control board or the sub-system main control board.

Further, the main system further includes a data bus and a control bus connected to the main control board of the main system, the main hub board and the main data board; the subsystem also includes A data bus and a control bus connected to the system main control board, the slave hub board and the slave data board.

Further, the main control board of the main system and the main control board of the sub-system use PowerPC boards.

Further, the diagnosis and prediction board uses a host computer or a PowerPC board.

Further, the diagnostic prediction board includes a fault prediction module, a database, and a fault receiving and processing module.

Further, the diagnosis prediction board is also connected to the master hub board and the slave hub board through an HSSL optical fiber transmission bus and a serial connection bus.

Further, the main system and the subsystem also include a fault diagnosis board, the fault diagnosis board is connected to the data bus and the control bus, and the fault diagnosis board and the diagnosis prediction board Through the HSSL optical fiber transmission bus and serial connection bus connection.

The invention also provides a diagnosis method using the intelligent diagnosis system as described above, which includes the following steps:

S1: The master hub board and the slave hub board acquire the detection data of the corresponding sensors in real time, and send the detection data to the master data board and the slave data board for calculation;

S2: periodically obtain the intermediate operation data of the master data board or the slave data board, and transmit it to the diagnosis prediction board;

S3: The diagnosis prediction board performs fault prediction according to the received intermediate operation data, and feeds back the prediction result to the main system main control board or the sub-system main control board.

Further, in step S2, the intermediate operation data is periodically acquired through the master hub board and the slave hub board, and the intermediate operation data is transmitted to the diagnosis prediction board.

Further, in the step S2, the intermediate operation data is periodically acquired through the fault diagnosis board to perform fault prediction, and the prediction information and the intermediate operation data are transmitted to the diagnosis prediction board.

Further, in the step S2, the fault prediction includes the following steps:

S21: The fault diagnosis board periodically obtains the intermediate operation data according to the time parameter and the data of interest in the configuration file, and puts it in the memory buffer;

S22: The fault diagnosis board monitors the acquired intermediate operation data in real time to determine whether the data value is within the safety range set by the configuration file;

S23: When the data value exceeds the safe range by between 0 and m%, report a warning message to the fault prediction board, and feed back to the main system main control board or the sub-system main control board in real time Drive components in the corresponding adjustments to avoid the deterioration of operating conditions;

S24: When the data value exceeds the safe range by more than m%, report the fault information to the fault prediction board for processing, and at the same time directly feed back to the driver in the main system main control board or the sub-system main control board The component is initialized, where m is a real number and is set by the configuration file.

Further, in step S24, when the data value exceeds the safe range by more than m%, the fault diagnosis board encodes the fault type, and triggers the serial interrupt to notify the diagnosis prediction board.

Further, the step S3 further includes that when the diagnostic prediction board receives the fault information, it first sends an instruction to suspend all sub-system actions, and handles the fault, judges the fault mechanism, whether to run the action and retry, if allowed , Then send a "retry" command to the participating sub-systems, retry the action, if the retry also fails, report to the server; if not, send a "system error" message to the server, wait Manual intervention.

Further, the step S3 includes the following steps:

S31: The diagnostic prediction board is configured according to the sampling time and the data of interest in the configuration file;

S32: The diagnostic prediction board samples the data of the fault diagnosis board every n servo cycles and stores it in the database, where n is a natural number, which is set by the configuration file;

S33: The fault prediction module in the diagnostic prediction board integrates the sampled data with the historical data in the database, fits the data change curve, and finds the corresponding rules in the database to obtain the fault prediction information;

S34: The diagnosis prediction board feeds back the fault prediction information to the corresponding sub-system through the main hub board in the main system, and the sub-system main control board in the sub-system makes corresponding adjustments and operations.

Further, in the step S33, the rules are all saved in the form of a fault tree.

Further, in step S33, if no rule is found, the fault prediction module performs fault training and stores it as a new rule.

Further, in step S33, the data change curve is fitted by the method of least squares or average trend.

Further, it also includes step S4, injecting a fault into the diagnosis prediction board or the main system main control board or the sub-system main control board to check the diagnosis effect of the intelligent diagnosis system.

Compared with the prior art, the intelligent diagnosis system and method provided by the present invention have the following advantages:

(1) Perform real-time memory operation on the data detected by the sensor, avoiding the time-consuming operation of frequency I / O of batch data;

(2) Adopt configurable data acquisition mode of interest to avoid the processing of a large amount of redundant data;

(3) Perform "online" processing and analysis of the operation intermediate data of the data board in the device, and the failure can be predicted;

(4) Intelligent fault judgment and online processing, avoiding equipment shutdown and waiting for human intervention, speeding up production progress and improving equipment yield;

(5) The "interrupt trigger" and "pause" modes are adopted to avoid the enlargement of the fault and easy to locate the fault;

(6) Adopt distributed fault diagnosis and processing model, which can quickly and timely deal with the faults of main system and sub-system;

(7) Adopt "fault training" and "rule processing" coexistence to quickly, real-time and accurately predict the occurrence of faults and deal with faults.

(8) Simulate the occurrence of faults through fault injection to test the fault handling capability of the system and improve system reliability.

The present invention will be described in detail below with reference to the drawings.

Example 1

As shown in FIG. 1, the present invention provides an intelligent diagnosis system, including:

The main system 100 includes a system main control board 1, a master hub board (Master Hub Board, MHB) 2 and a plurality of data boards (Data Board) 3, as well as the system main control board 1, a main hub board 2 Data bus 4 and control bus 5 connected to the data board 3. The main hub board 2 is connected to a sensor corresponding to the main system 100, and is used to receive the measurement data of the sensor and send it to each data board 3 of the main system 100 for calculation.

The sub-system 200 includes a system main control board 1, a slave hub board 6 and a plurality of data boards 3, and a system main control board 1, a slave hub board 6 and a data board 3 Connected data bus 4 and control bus 5, the slave hub board 6 is connected to the sensor corresponding to the sub-system 200, and is used to receive the measurement data of the sensor and send each data to the sub-system 200 Board 3 performs the calculation.

Among them, the system main control board 1 uses PowerPC (Performance Optimization With Enhanced RISC-Performance Computing, reduced instruction set RISC architecture central processing unit) board, also known as PPC board, which is mainly responsible for receiving commands to diagnose and predict the board 300. The command will be interpreted and sent to other boards in the system. Specifically, the system main control board 1 receives the commands issued by the diagnostic prediction board 300 through the Ethernet bus 7 (such as initialization, machine parameter issuance, operation firmware allocation, etc.), and after interpreting the commands, the data bus 4 Send the command to each board in the system, for example, the system main control board 1 in the main system 100 is used to send the command to the main hub board 2 and the data board in the main system 100 through the data bus 4 Card 3, the system main control board 1 in the sub-system 200 is used to send commands to the slave hub board 6 and the data board 3 in the sub-system 200 through the data bus 4. The data bus 4 can use SRIO, SDB, MDB, PCIe, etc .; during the operation of the system, some other auxiliary information (such as Trace, etc.) will be sent to the PPC board through the control bus 5 to complete the processing or storage of this information The control bus 5 is a VME64x or VPX bus, or Ethernet. At the same time, the driver of the system runs on the PPC board, and the parameters of the data board 3 are distributed and controlled through the control bus 5.

The data board 3 is mainly used for the realization of the control algorithm and the calculation and processing of data in the control process. After the system is initialized, the data board 3 is always in a ready state, waiting for the sensor to detect the arrival of data; when data arrives, the data is quickly moved to the RAM of the data board to complete the task at the fastest speed In the second calculation, the calculation result is broadcasted through the data bus 4 according to the sequence agreed in advance. All the boards in the system can obtain the calculation result from the data bus 4 and store it, realizing the data interaction between the boards; data The intermediate operation data in the process of processing can be written into the external storage of the data board 3 or DPRAM according to the configuration requirements, so as to be provided to the main hub board 2 or the slave hub board 6 to grab, to avoid directly discarding after the execution of the operation is completed data.

A diagnosis prediction board (Master Diagnosis Trigger Board, MDT) 300, which uses a host computer or a PowerPC board, is connected to the system main control board 1 through an Ethernet bus 7, and the diagnosis prediction board 300 is also connected to the main hub The board 2 and the slave hub board 6 are connected by an HSSL optical fiber transmission bus 8 and a serial connection bus 9. The serial bus 5 may be RS232, RS485, USB, IEEE1394, etc. As shown in FIG. 2, the diagnostic prediction board 300 includes a fault prediction module 10, a database 11 and a fault receiving and processing module 12, wherein the fault prediction module 10 Use "rule processing" and "fault training" to predict faults and improve the fault rules of database 11 in real time; database 11 is mainly used to store processing rules, and the rules are all saved in the form of "fault tree", that is, a The data trend corresponds to a fault type; the fault receiving and processing module 12 is mainly responsible for fault processing. Specifically, the master hub board 2 or slave hub board 6 periodically grabs the data written by the data board 3 to the external memory or DPRAM, and every n servo cycles, n is a natural number, set by the configuration file , Upload the operation data of this servo cycle to the diagnosis prediction board 300 through the HSSL optical fiber transmission bus 8, the diagnosis prediction board 300 is stored in the database 11 after receiving; the failure prediction module 10 will grab the operation data and database 11 of this time Comprehensively process the historical data in the system, fit the data change curve, and find the corresponding rules in the database 11 to obtain the fault prediction information, and feed it back to the corresponding sub-system 200 through the main hub board 2 in the main system 100. The system main control board 1 in the system 200 makes corresponding adjustments and operations; if no rules are found, "fault training" is performed and stored as new rules.

There are two ways to fit the parameter data:

Least square method: due to the current, voltage, temperature and other parameter values, it should be kept as stable as possible during the operation of the equipment, and its growth trend is slow, which can be set as a linear regression linear equation (For different parameter data, the set function terms are different, such as speed and acceleration, etc., which are multi-order polynomials), according to ,Correct Partial derivation can get the values of a and b, so as to determine the relationship function. In the above equation, t is the time sampling value, y is the sampling value of the current, voltage, temperature and other parameter values, a and b are the linearization coefficients of the linear regression equation to be fitted, and the difference between the true value and the fitting value Value (that is, the error value), i represents the sampling point.

MA method: that is to find the average trend, the function can be set as Through the average value between two points, the formula is gradually determined and repaired, and the failure trend is predicted by the formula.

This embodiment also provides a diagnosis method of the above intelligent diagnosis system, including the following steps:

S1: The master hub board 2 and the slave hub board 6 acquire the detection data of the sensor in real time, and send the detection data to the data board 3 for calculation. The data board 3 runs in the middle of the data processing process Data can be written to its external memory or DPRAM according to configuration requirements.

S2: The main hub board 2 and the slave hub board 6 periodically obtain the intermediate operation data of the data board 3 and transmit it to the diagnosis prediction board 300; specifically, the master hub board 2 or the slave The hub board 6 periodically grabs the data written by the data board 3 to the external memory or DPRAM, and every n servo cycles, n is a natural number, set by the configuration file, and uploaded this time through the HSSL fiber transmission bus 8 The operation data of the servo cycle is sent to the diagnosis prediction board 300, and the diagnosis prediction board 300 is stored in the database 11 after being received.

S3: The diagnosis prediction board 300 performs fault prediction according to the received intermediate operation data, and feeds back the prediction result to the system main control board 1. Specifically, the fault prediction module 10 will comprehensively process the running data of this time and the historical data in the database 11, fit the data change curve, and find the corresponding rules in the database 11, to obtain the fault prediction information, and pass the main system The main hub board 2 in 100 feeds back to the corresponding sub-system 200, and the system main control board 1 in the sub-system 200 makes corresponding adjustments and operations; if no rules are found, then "fault training" is performed and stored as New rules.

S4: Inject faults into the diagnosis prediction board 300 or the system main control board 1 to check the diagnosis effect of the system. That is, fault injection can be injected by the diagnosis and prediction board 300, and the diagnosis and prediction board 300 packages the fault information to the main system 100 and the sub-system 200, and processes the corresponding faults of the main system 100 and the sub-system 200 in real time; or for the main system 100 or the system main control board 1 of the sub-system 200 is injected separately, and the diagnosis and prediction board 300 receives and processes the faults of the main system 100 and the sub-system 200.

Example 2

As shown in FIG. 3, unlike Embodiment 1, the main system 100 and the sub-system 200 in this embodiment further include a fault diagnosis board (Slave Diagnosis Trigger Board, SDT) 13, and the fault diagnosis board 13 Connected to the data bus 4 and the control bus 5, the fault diagnosis board 13 and the diagnosis prediction board 300 are connected via an HSSL optical fiber transmission bus 8 and a serial connection bus 9.

The functions of the fault diagnosis board 13 mainly include the following aspects:

1. The configurability of time and data can be configured according to the time parameters of the configuration file DTS.cfg and the data of interest. If it is a motion subsystem, the data can be taken from the motor voltage or current data; if it is a lighting subsystem, The data can be taken from laser light intensity, laser dose and other parameters; if it is an environmental subsystem, the data can be taken from temperature, pressure, humidity and other parameters. Of course, it is not limited to the above parameters, the specific parameters are defined by the actual scene or the engineer.

2. Go to the DPRAM or external memory of each data board 3 at a certain time to grab the middle operation data, and put it into the memory buffer developed by the fault diagnosis board 13 for each data board 3; The number of servo cycles is the unit. The initialization time is read from the configuration file DTS.cfg. Similarly, this parameter can be set in real time by the user interface; because the probability of failure is often highest during initialization and machine startup, therefore, the time at this time The interval is as small as possible, and can be set to 1 servo cycle; when the equipment is stable, it can be adjusted in real time according to demand or actual conditions;

3. Real-time monitoring of the intermediate operating data to determine whether the data value is in the safe range, which is set by the configuration file DTS.cfg; as shown in Figure 4, the safe range can be set to three states: health, fault and sub Health; health status corresponds to data within a safe range; sub-health status corresponds to data within m% of the safety threshold; failure status corresponds to data above m% of the safety threshold, where m is a real number, set by the configuration file ;

4. When the monitored parameter value exceeds the safe range and the range is between 0 and m%, that is, the system is in a sub-health state, the fault diagnosis board 13 reports a warning message to the fault prediction board through the serial connection bus 9 300, and real-time feedback to the drive components in the main control board 1 of the system to make corresponding adjustments to avoid the deterioration of operating conditions;

5. When the monitored parameter value exceeds the safe range amplitude ≧ m%, the system is now defined as a fault state, and the fault diagnosis board 13 first encodes the fault type and reports the fault information through the serial connection bus 9 Handle the fault prediction board 300, and directly feed back to the drive components in the system main control board 1 to perform the initialization operation to avoid being in a fault waiting state, which is convenient for the upper layer to send "Retry" or other requests, where m is a real number, configured by File settings.

Correspondingly, after the fault receiving and processing module 12 in the diagnosis and prediction board 300 receives the fault information of the system, it first sends an event to suspend all the actions of the sub-system 200, and processes the fault to determine the fault mechanism and whether to run the action Retry, if allowed, send a "Retry" command to the participating subsystem 200 to retry the action; if not, send a "system error" to the server, waiting for human intervention.

The diagnosis method of the intelligent diagnosis system described in this embodiment includes the following steps:

S1: The main hub board 2 and the slave hub board 6 acquire the detection data of the sensor in real time, and send the detection data to the data board 3 for calculation; the data board 3 runs in the middle of the data processing process Data can be written to its external memory or DPRAM according to configuration requirements.

S2: The fault diagnosis board 13 periodically acquires the intermediate operation data of the data board 3, performs fault prediction, and transmits the prediction information and the intermediate operation data to the diagnosis prediction board 300. The failure prediction includes the following steps:

S21: The fault diagnosis board 13 periodically grabs the intermediate operation data of the data board 3 according to the time parameters and interested data in the configuration file DTS.cfg, of course, it can also be actively obtained and put into the memory buffer ; It should be noted that if it is a motion subsystem, the data of interest can be taken from the voltage or current data of the motor; if it is a lighting subsystem, the data can be taken from laser light intensity, laser dose and other parameters; if it is an environmental subsystem, it is Interest data can take parameters such as temperature, pressure and humidity. The above parameters are taken as examples and are not limited to the above parameters. The specific parameters are defined by the actual scene or the engineer.

S22: The fault diagnosis board 300 monitors the acquired intermediate operation data in real time to determine whether the data value is within the security range set by the configuration file; the security range is set by the configuration file DTS.cfg; the security range can be set to three states : Health, failure and sub-health; health status corresponds to data within a safe range; sub-health status corresponds to data within m% of the safety threshold; failure status corresponds to data exceeding m% of the safety threshold, where m is Real number, set by configuration file;

S23: When the data value exceeds the safe range by between 0 and m%, the fault diagnosis board 13 reports a warning message to the fault prediction board 300 through the serial connection bus 9 and feeds back to the system main control board in real time The drive components in 1 should be adjusted accordingly to avoid the deterioration of operating conditions;

S24: When the data value exceeds the safe range by more than m%, the system is defined as a fault state. The fault diagnosis board 13 first encodes the fault type and triggers it through a serial interrupt to notify the diagnosis prediction board 300 At the same time, it directly feeds back to the drive component in the main control board 1 of the system to perform the initialization operation, where m is a real number, which is set by the configuration file.

S3: The diagnosis prediction board 300 performs fault prediction according to the received intermediate operation data, and feeds back the prediction result to the system main control board 1. It includes the following steps:

S31: The diagnostic prediction board 300 is configured according to the sampling time and the data of interest in the configuration file;

S32: The diagnostic prediction board 300 is sampled once in n servo cycles (can be actively obtained, or passively received) and the data of the fault diagnosis board 13 is stored in the database 11, where n is a natural number, which is set by the configuration file;

S33: The fault prediction module 10 in the diagnostic prediction board 300 performs comprehensive processing on the sampled data and the historical data in the database 11, fits the data change curve, and looks for the corresponding rules in the database 11 to obtain the fault prediction information;

S34: The diagnostic prediction board 300 feeds back the fault prediction information to the corresponding sub-system 200 through the main hub board 1 in the main system 100, and the system main control board 1 in the sub-system 200 makes corresponding adjustments and operations .

In addition, when the diagnosis prediction board 300 receives the failure information of the failure diagnosis board 13, it first sends an instruction to suspend all the actions of the sub-system 200, and handles the failure, judges the failure mechanism, whether to run the operation and retry, if allowed, Then send a "retry" command to the sub-system participating in the action, retry the action, if the retry also fails, it will be reported to the server 400, waiting for manual intervention; if not allowed, send a "system error" to the server End 400, waiting for human intervention.

Compared with Embodiment 1, in the technical solution provided in this embodiment, a fault diagnosis board 13 is added, and a distributed fault diagnosis and processing model is adopted to realize the pre-detection and online processing of faults in the main system 100 and the sub-system 200, further improving For efficiency.

In summary, compared with the prior art, the intelligent diagnosis system and method provided by the present invention have the following advantages:

(1) Perform real-time memory operation on the data detected by the sensor, avoiding the time-consuming operation of frequency I / O of batch data;

(2) Adopt configurable data acquisition mode of interest to avoid the processing of a large amount of redundant data;

(3) Perform "online" processing and analysis of the operation intermediate data of the data board 3 in the device, and the failure can be predicted;

(4) Intelligent fault judgment and online processing, avoiding equipment shutdown and waiting for human intervention, saving time and improving efficiency;

(5) The "interrupt trigger" and "pause" modes are adopted to avoid the enlargement of the fault and easy to locate the fault;

(6) The distributed fault diagnosis and processing model can be used to quickly and timely deal with the faults of the main system 100 and the sub-system 200;

(7) Adopt "fault training" and "rule processing" coexistence to quickly, real-time and accurately predict the occurrence of faults and deal with faults.

(8) Simulate the occurrence of faults through fault injection to test the fault handling capability of the system and improve system reliability.

Although the description describes the embodiments of the present invention, these embodiments are only used as a reminder and should not limit the protection scope of the present invention. Various omissions, substitutions, and changes without departing from the gist of the present invention should be included in the protection scope of the present invention.

100‧‧‧Main system

200‧‧‧Subsystem

300‧‧‧Diagnosis prediction board

400‧‧‧Server

1‧‧‧System main control board

2‧‧‧Main hub board

3‧‧‧Data board

4‧‧‧Data bus

5‧‧‧Control bus

6‧‧‧From the hub board

7‧‧‧Ethernet bus

8‧‧‧HSSL optical fiber transmission bus

9‧‧‧ serial connection bus

10‧‧‧Fault prediction module

11‧‧‧ database

12‧‧‧Fault receiving and processing module

13‧‧‧Fault diagnosis board

1 is a schematic structural diagram of an intelligent diagnosis system in Embodiment 1 of the present invention; FIG. 2 is a structural schematic diagram of a diagnostic prediction board in Embodiment 1 of the present invention; FIG. 3 is a schematic structural diagram of the diagnostic prediction Board in embodiment 2 of the present invention; FIG. 4 is a schematic diagram of judgment of the three states by the fault diagnosis board in Embodiment 2 of the present invention.

Claims (18)

An intelligent diagnosis system is characterized by comprising: a main system, including a main system main control board, a main hub board and a plurality of main data boards, the main hub board being connected to a sensor corresponding to the main system For receiving the measurement data of the sensor and sending to each of the master data boards of the master system for calculation; subsystem, including the master control board of the subsystem, slave hub board and multiple slaves A data board, the slave hub board is connected to a sensor corresponding to the subsystem, and is used to receive the measurement data of the sensor and send it to each slave data board of the subsystem for calculation Diagnosis and prediction board, connected to the main system main control board and the sub-system main control board, used to periodically obtain the intermediate operation data of the master data board or the slave data board, and Perform failure prediction based on the received intermediate operation data, and feed back the prediction result to the main system main control board or the sub-system main control board. The intelligent diagnosis system according to claim 1, wherein the main system further includes a data bus and a control bus connected to the main system main control board, the main hub board and the main data board; The subsystem also includes a data bus and a control bus connected to the master control board of the subsystem, the slave hub board, and the slave data board. The intelligent diagnosis system according to claim 1, wherein the main system main control board and the sub-system main control board use PowerPC boards. The intelligent diagnosis system according to claim 1, wherein the diagnosis prediction board is a host computer or a PowerPC board. The intelligent diagnosis system according to claim 1, wherein the diagnosis prediction board includes a failure prediction module, a database, and a failure receiving and processing module. The intelligent diagnosis system according to claim 1, wherein the diagnosis prediction board is further connected with the master hub board and the slave hub board through an HSSL optical fiber transmission bus and a serial connection bus. The intelligent diagnosis system according to claim 2, wherein the main system and the subsystem further include a fault diagnosis board, the fault diagnosis board is connected to the data bus and the control bus, and the fault diagnosis board The card and the diagnosis prediction board are connected by an HSSL optical fiber transmission bus and a serial connection bus. An intelligent diagnosis method, which is characterized by adopting the intelligent diagnosis system as described in claim 1 and including the following steps: S1: The master hub board and slave hub boards acquire the detection data of the corresponding sensors in real time, and the The detection data is sent to the master data board and the slave data board for calculation; S2: periodically obtain the intermediate operation data of the master data board or the slave data board, and transmit it to the diagnosis prediction board; S3 : The diagnosis prediction board performs fault prediction according to the received intermediate operation data, and feeds back the prediction result to the main system main control board or the sub-system main control board. The intelligent diagnosis method according to claim 8, wherein in the step S2, the intermediate operation data is periodically obtained through the master and slave hub boards, and the intermediate operation data is transmitted to the Diagnose prediction board. The intelligent diagnosis method according to claim 8, wherein in the step S2, the intermediate operation data is periodically acquired through a failure diagnosis board, a failure prediction is performed, and the prediction information and the intermediate operation data are transmitted to the diagnosis prediction Board. The intelligent diagnosis method according to claim 10, wherein in the step S2, the fault prediction includes the following steps: S21: The fault diagnosis board periodically obtains the Intermediate operation data, and put it in the memory buffer; S22: The fault diagnosis board monitors the obtained intermediate operation data in real time to determine whether the data value is within the safe range set by the configuration file; S23: When the data value exceeds the safe range When the amplitude is between 0 and m%, report a warning message to the fault prediction board, and feedback it in real time to the drive components in the main system main control board or the sub-system main control board, and perform the corresponding Adjust to avoid deterioration of operating conditions; S24: When the data value exceeds the safe range by more than m%, report the fault information to the fault prediction board for processing, and at the same time directly feed back to the main system main control board or the branch The drive components in the main control board of the system are initialized, where m is a real number, which is set by the configuration file. The intelligent diagnosis method according to claim 11, wherein in step S24, when the data value exceeds the safe range by more than m%, the fault diagnosis board encodes the fault type and triggers it through a serial interrupt, Notify the diagnosis prediction board. The intelligent diagnosis method according to claim 12, wherein the step S3 further includes: after the diagnosis prediction board receives the failure information, first sends an instruction to suspend all sub-system actions, and handles the failure to determine the failure mechanism , Whether to run the action retry, if allowed, send the "retry" command to the participating subsystems, retry the action, if the retry also fails, then report to the server; if not, send " "System error" message to the server, waiting for human intervention. The intelligent diagnosis method according to claim 10, wherein the step S3 includes the following steps: S31: the diagnosis prediction board is configured according to the sampling time in the configuration file and the data of interest; S32: the diagnosis prediction board The data of the fault diagnosis board is sampled every n servo cycles and stored in the database, where n is a natural number, which is set by the configuration file; S33: The fault prediction module in the diagnosis prediction board compares the sampled data with this time Comprehensively process the historical data in the database, fit the data change curve, and find the corresponding rules in the database to obtain the fault prediction information; S34: The diagnostic prediction board feeds the fault prediction information back through the main hub board in the main system For the corresponding sub-system, the main control board of the sub-system performs corresponding adjustment and operation. The intelligent diagnosis method according to claim 12, wherein in the step S33, the rules are all saved in the form of a fault tree. The intelligent diagnosis method according to claim 12, wherein in step S33, if no rule is found, the fault prediction module performs fault training and stores it as a new rule. The intelligent diagnosis method according to claim 12, wherein in step S33, the data change curve is fitted by the method of least squares or average trend. The intelligent diagnosis method according to claim 8, further comprising step S4, injecting a fault into the diagnosis prediction board or the main system main control board or the sub-system main control board to verify The diagnosis effect of the intelligent diagnosis system is described.