Method for state monitoring and fault diagnosis of loading and unloading machine

Abstract

The invention belongs to the technical field of nuclear power equipment detection, and particularly relates to a loading and unloading machine state monitoring and fault diagnosis method. Comprising the following steps: step 1, design function analysis; 2, building a hardware detection system; 3, establishing a software detection system; 4, testing software and hardware; 5, generating a data set; and step 6, fault diagnosis. The method has the beneficial effects that software and hardware logic detection of the loading and unloading machine system is realized, and a data set is established. On-line monitoring of the operation state of the loading and unloading machine equipment is achieved, a fault diagnosis model based on the rough set theory is used for evaluating the operation state of the system in real time and diagnosing and positioning faults of the loading and unloading machine, and maintenance suggestions are given.

Description

Technical Field

[0001] The invention belongs to the technical field of nuclear power equipment detection, and in particular relates to a method for state monitoring and fault diagnosis of a loader and unloader. Background Art

[0002] The loader is one of the key equipment in a nuclear power plant, also known as a refueling machine, which is used to transfer and load and unload fuel assemblies. The healthy operation of the loader and unloader itself and the rapid diagnosis after a fault are indispensable for the safe and reliable loading and unloading of nuclear power plants. Through the analysis and statistics of past events of the loader, it is found that there are many types of faults on the loader site. Most of the faults or events cannot be predicted by the program before they occur, and after the fault occurs, it is often manually judged based on experience. Comprehensive judgment of the fault status and cause; there may also be various control logic defects in the design process of the loader system. Such defects often only occur under very specific conditions. The probability of occurrence during routine drills and loading and unloading processes is extremely low, but once it occurs, it is a high-level safety fault.

[0003] Therefore, effective monitoring means and fault diagnosis methods can greatly reduce the situation where the loading and unloading machine is delayed due to downtime due to faults and the overhaul process is delayed. Summary of the invention

[0004] The purpose of the present invention is to provide a method for monitoring the state and diagnosing faults of a loader and unloader, which can improve the operating stability of the loader and unloader and improve the efficiency of fault diagnosis. The method can realize state monitoring, fault diagnosis and positioning of the loader and unloader system and provide maintenance suggestions.

[0005] The technical solution of the present invention is as follows: A method for state monitoring and fault diagnosis of a loading and unloading machine, comprising:

[0006] Step 1: Design functional analysis;

[0007] Step 2: Build a hardware detection system;

[0008] Step 3: Build a software detection system;

[0009] Step 4: Software and hardware testing;

[0010] Step 5: Dataset generation;

[0011] Step 6: Troubleshooting.

[0012] The step 1 comprises:

[0013] Step 11: Analyze the Original Design

[0014] Design function analysis includes electrical function analysis, software function analysis, operation process and process flow analysis. Electrical function analysis includes analysis of the electrical design function of the loader and unloader, including drive, signal logic, safety, fault signal, interlock signal, and bypass operation;

[0015] Software function analysis includes analysis of software functions for management of loading and unloading machine operation process, sensor signal processing and storage, motor drive workflow and feedback processing, redundant signal processing, interlocking protection function, abnormal status information prompt, protection / interlocking threshold setting, and abnormal status judgment;

[0016] The operation process and process flow analysis includes analyzing the operation process and process flow of the loading and unloading machine, obtaining the components, operating parameters, keywords, and keywords or parameters of the interlocking relationship between components in the original design function, and forming a process flow analysis record table;

[0017] Step 12: Analyze and summarize past events

[0018] Collect statistics on past events of loading and unloading machines, summarize the causes of the events, event phenomena, final handling methods of the events, and the magnitude of the impact of the events;

[0019] Step 13: Create a fault sample database

[0020] Fault sample database based on past event analysis results.

[0021] The step 2 includes: the hardware detection system realizes the hardware detection of the loader and unloader by deploying internal and external detection systems, wherein the internal detection system refers to the detection of the sensors and actuators of the loader and unloader itself; the external detection system is the detection by deploying external sensors and collectors to obtain more status information of the loader and unloader.

[0022] The step 3 includes: the software detection system communicates with the originally designed PLC hardware by building a set of in-loop detection software, and uses the in-loop test system to simulate the signal feedback of the actuator and sensor to achieve a full functional test of the controller.

[0023] The step 4 comprises:

[0024] Step 41: Hardware Detection Content

[0025] The hardware detection content includes the controller operation data, driver operation data, encoder feedback data, limit switch, handwheel, brake wear, balance beam detection and other signals of the loader and unloader, load sensor signals, gripper engagement and release sensors, panel operation area signals, communication signals between various devices, motor vibration signals, and equipment noise signals;

[0026] Step 42: Hardware Detection Method

[0027] Step 43: Software Detection Content

[0028] The software detection object is the control program of the loader and unloader, in which the software modules used to control and operate the loader and unloader are the logic control system, the safety protection system and the communication system with other equipment, including the transfer device and the spent fuel pool crane;

[0029] Step 44: Software Detection Method

[0030] The detection method of the software detection system is to build an in-loop detection platform, use in-loop detection equipment to replace the real controlled object, receive control instructions and simulate status information to return to the controller under test. At the same time, the software detection system is connected to the controller under test to read the response information of the built-in software program of the controller under test in real time, forming an in-loop test closed loop. The detection contents involved in software detection include normal function detection of the program, fault injection detection, and exhaustive detection.

[0031] The step 42 comprises:

[0032] Step 421: Internal testing method

[0033] The test hardware of the internal test system consists of PLC, high-frequency data acquisition board, communication interface conversion module, and PC. The test host computer hardware is connected to the field controller through the bus;

[0034] The host computer software and PLC program of the internal test run on the PC and PLC system of the internal test system respectively. The PLC program is used to run the driver program and convert the sensor data into units and store them in the state variable table. The host computer software obtains the state variable table of the PLC at a certain frequency.

[0035] Step 422: External testing method

[0036] The external measurement system has added sensors, and its acquisition hardware and host computer software use independent equipment and programs. The new sensors in the external measurement system are aimed at the main components of the loader and unloader, namely the trolley motor, the trolley motor, the main lifting motor, and the main lifting wire rope, to monitor the vibration and noise data of each motor and the health status of the main lifting wire rope.

[0037] The step 44 comprises:

[0038] Step 441: Normal function detection

[0039] Build an in-loop detection platform, run the object to be tested, and the detection system drives the simulation equipment to return the simulation status information to the controller to be tested according to the test case requirements, read the response of the built-in software program of the controller to be tested in real time, compare the output signal with the test case database, and realize normal function detection;

[0040] Step 442: Fault injection detection

[0041] The detection system drives the simulation device to return the simulation status information to the controller under test according to the fault injection detection requirements, reads the response of the built-in software program of the controller under test in real time, compares the output signal with the fault injection database, and realizes fault detection;

[0042] Step 443: Exhaustive detection

[0043] The detection system drives the simulation device to return the simulation status information to the controller under test in accordance with the exhaustive full logic detection requirements, reads the response of the built-in software program of the controller under test in real time, and compares the output signal with the exhaustive database to achieve exhaustive detection.

[0044] The step 5 comprises:

[0045] Step 51: Data preprocessing and discretization of continuous attributes

[0046] The purpose of data preprocessing is to remove invalid data. When using rough set theory to process decision tables, the data in the table is required to be discrete, specifically in the form of character integers and enumerations. The attributes are discretized. In the decision table, the values ​​of conditional attributes and decision attributes are divided into two types:

[0047] Continuous attributes: the value range of the attribute is a continuous interval, and the number of values ​​is infinite, such as the signals collected by various sensors in equipment operation monitoring;

[0048] Discrete attributes: the range of attribute values ​​is a finite set, and the number of values ​​is finite;

[0049] When using rough set theory to process decision tables, the data in the table must be discrete, specifically in the form of character integers and enumerations. The attributes are discretized, and the discretization of continuous attributes is to determine the breakpoints in the middle. In the fault diagnosis process, the characteristic frequencies of vibration and noise signals are continuous attributes, and the equal distance and equal frequency methods are used; other signals are state quantity signals, and Boolean logic is used;

[0050] Step 52: Knowledge acquisition and diagnostic model

[0051] In actual diagnosis, the training sample set comes from empirical knowledge or original data. First, the original data is described, and the condition attribute value is quantified. If it is a continuous attribute, it is discretized, and the discretized attribute is assigned to form a decision table. The decision table is simplified, and the rules are evaluated using the simplification rate, confidence and coverage to obtain a rule set and store it in the system database; these rules become the knowledge source of the fault diagnosis system, and based on the obtained minimum condition attribute set, the actual test sample is constructed, and the fault type is directly obtained through rule reasoning, and the judgment result is output;

[0052] Step 3: Establish mapping through BP neural network algorithm

[0053] BP network is a multi-layer feedforward neural network. The transfer function of its neurons is an S-type function. The output is a continuous quantity between 0 and 1, realizing any nonlinear mapping from input to output. The transfer function of the neurons in the middle layer is an S-type tangent function, and the transfer function of the neurons in the output layer is an S-type logarithmic function.

[0054] The step 6 comprises:

[0055] Step 61: Data Collection

[0056] Step 611: Vibration and noise data

[0057] Use frequency domain analysis method to extract characteristic frequency parameters and convert continuous signals into discrete characteristic value signals;

[0058] Step 612: Control Signal

[0059] Control signals generated by the loader control program;

[0060] Step 613: Status signal

[0061] Motor position, speed, absolute encoder signal, limit switch signal, interlock signal;

[0062] Step 614: Operation Signal

[0063] The operator's operation input signals on the touch screen and panel, including handle input signals, speed selection signals, and fuel selection signals;

[0064] Step 62: Database Contents

[0065] Step 621: Real-time operation data

[0066] This part of data comes from the data generated by the hardware detection system on the verification platform or the on-site detection data;

[0067] Step 622: Fault Data

[0068] This part of data comes from the detection data generated when the hardware detection platform runs abnormally, the detection data obtained by using fault injection data during software detection, and the fault data expanded and corrected by rough set algorithm diagnosis;

[0069] Step 623: Software detection data set

[0070] The entire data set generated during software testing;

[0071] Step 624: Past event data

[0072] Keywords and data combinations obtained through analysis of past events;

[0073] Step 63: Data Analysis

[0074] Firstly, the real-time data set and the fault mode set are determined, and a decision table consisting of a sample library is established. The data completion, attribute reduction, value reduction and rule extraction methods in the rough set theory are used to extract rules, and the acquired knowledge rules are stored in the knowledge base.

[0075] For real-time data sets, the real-time data is input into the inference engine, and the diagnosis result can be obtained through reasoning according to a certain reasoning mechanism. After the diagnosis sample is verified, it is used as a case and input into the original sample database for storage according to certain principles, and incremental learning is carried out, and this cycle repeats.

[0076] Step 64: Processing Decisions

[0077] If the diagnosis result is correct, operation and maintenance advice should be given according to the fault type, and finally a diagnosis report should be output. For fault symptom data, if the program predicts that a fault is about to occur, a fault warning prompt will be issued.

[0078] The beneficial effects of the present invention are: realizing the software and hardware logic detection of the loader system and establishing a data set. It realizes the online monitoring of the operation status of the loader equipment, and uses the fault diagnosis model based on the rough set theory to evaluate the system operation status in real time, diagnose and locate the loader fault, and give maintenance suggestions. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 A system diagram of a method for state monitoring and fault diagnosis of a loading and unloading machine provided by the present invention;

[0080] Figure 2 Implement a flow chart for hardware detection;

[0081] Figure 3 Implement a flow chart for software detection. DETAILED DESCRIPTION

[0082] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0083] A method for state monitoring and fault diagnosis of a loader and unloader. The method introduces rough set theory and combines it with a BP neural network deep learning method to perform data mining on the loader and unloader operation data set. The method can objectively describe and process uncertainties without requiring prior information outside the data set, thereby greatly improving the overall fault diagnosis efficiency of large and complex equipment such as loader and unloader.

[0084] A method for state monitoring and fault diagnosis of a loader and unloader is proposed. The hardware and software of the loader and unloader are tested, and the test results are formed into a database. A special rule base is developed for the data set using rough set theory. Combined with on-site operation state monitoring, the fault of the loader and unloader is diagnosed and located, and maintenance suggestions are given. The specific method is as follows.

[0085] Build a data acquisition system to read the information of the loader and unloader's sensors, controllers, drivers, instruments, host computers, etc. online; deploy an external sensor system to monitor the loader and unloader's vibration, noise and other status-related information. According to the loader and unloader system's process, interlocking and other logic, exhaustively test all possible signal combinations to ensure the logical completeness of the acquired data set, extract the process parameters, environmental parameters, and status parameters (controller, driver, encoder) of the loader and unloader's operation process, and establish a database.

[0086] The database is used to acquire knowledge using rough sets, and the redundant information is simplified and removed according to the rough set theory to form a special rule base. The BP neural network algorithm is then used for nonlinear mapping to improve the success rate of fault diagnosis and prediction. Ultimately, the purpose of fault prediction and diagnosis can be achieved based on the actual state monitoring on site.

[0087] like Figure 1 As shown, a method for state monitoring and fault diagnosis of a loading and unloading machine includes the following: design function analysis, building a hardware detection system, building a software detection system, software and hardware testing, data set generation and fault diagnosis.

[0088] Step 1: Design Function Analysis

[0089] Step 11: Analyze the Original Design

[0090] Design function analysis mainly includes electrical function analysis, software function analysis, and operation process and process flow analysis. Electrical function combing: Analyze the electrical design functions of the loader and unloader, including drive, signal logic, safety, fault signal, interlocking signal, bypass operation, etc.; Software function combing: Analyze the management of the loader and unloader operation process, sensor signal processing and storage, motor drive workflow and feedback processing, redundant signal processing, interlocking protection function, abnormal status information prompt, protection / interlocking threshold setting, abnormal status judgment and other functions in the software functions; Operation process and process flow combing: Comb the operation process and process flow of the loader and unloader to obtain the components, operating parameters, keywords, interlocking relationships between components and other keywords or parameters in the original design function to form a process flow analysis record sheet;

[0091] Step 12: Analyze and summarize past events

[0092] Collect statistics on past events of loading and unloading machines, summarize the causes of the events, event phenomena, final handling methods of the events, and the magnitude of the impact of the events;

[0093] Step 13: Create a fault sample database

[0094] Fault sample database based on past event analysis results.

[0095] Step 2: Build a hardware detection system

[0096] The hardware detection system needs to implement hardware detection of the loader and unloader by deploying internal and external detection systems. The internal detection system refers to the detection of the sensors and actuators of the loader and unloader itself; the external detection system is the detection by deploying external sensors and collectors to obtain more status information of the loader and unloader. The hardware detection implementation process is as follows Figure 2 shown.

[0097] Step 3: Build a software detection system

[0098] The software detection system needs to build a set of in-loop detection software to communicate with the original PLC hardware design, and use the in-loop test system to simulate the signal feedback of the actuator and sensor to achieve a full functional test of the controller. The software detection implementation process is as follows: Figure 3 shown.

[0099] Step 4: Software and hardware testing

[0100] Step 41: Hardware Detection Content

[0101] The hardware detection objects are the controller operation data, driver operation data, encoder feedback data, limit switch, handwheel, brake wear, balance beam detection and other signals of the loader and unloader, load sensor signals, gripper engagement and release sensors, panel operation area signals, communication signals between various devices, motor vibration signals, equipment noise signals, etc.

[0102] Step 42: Hardware Detection Method

[0103] Step 421: Internal testing method

[0104] The hardware systems included in the original design include: motor, motor resolver, motor bus communication port, encoder, limit switch, brake controller, load sensor, counterweight limit sensor, front and rear and left and right anti-collision photoelectric switches for large and small vehicles, large and small vehicles and main lifting position display, etc.

[0105] The test hardware of the internal test system consists of PLC, high-frequency data acquisition board, communication interface conversion module, PC and other hardware. The test host computer hardware is connected to the field controller through a bus.

[0106] The host computer software and PLC program of the system internal test run on the PC and PLC system of the internal test system respectively. The PLC program is used to run the driver program, convert the sensor data into units, and store them in the state variable table; the host computer software obtains the PLC state variable table at a certain frequency, visualizes and stores the data in real time, and performs signal processing such as FFT (fast Fourier transform) and filtering on the real-time signal when necessary.

[0107] Step 422: External testing method

[0108] For parameters that are not included in the original design and require parameter testing, an external testing system is used for testing: necessary sensors are added, and at the same time, these sensors are required to be easily and quickly removed after the calibration is completed without affecting the on-site construction process.

[0109] In addition to adding new sensors, the external test system's acquisition hardware and host computer software use independent equipment and programs, and do not use the originally designed host computer hardware and software.

[0110] The newly added sensors of the external measurement system are aimed at the main components of the loader and unloader, namely the trolley motor, the trolley motor, the main lifting motor, and the main lifting wire rope. They monitor the vibration and noise data of each motor and the health status of the main lifting wire rope.

[0111] Step 43: Software Detection Content

[0112] The software detection object is the loader control program, among which the software modules used to control and operate the loader actions are the logic control system, the safety protection system and the communication system with other equipment (transfer device, spent fuel pool crane), so the above three systems are selected as the main detection objects.

[0113] Step 44: Software Detection Method

[0114] The detection method of the software detection system is to build an in-loop detection platform, use in-loop detection equipment to replace the real controlled object, receive control instructions and simulate status information and return it to the controller under test (software carrier). At the same time, the software detection system is connected to the controller under test for communication, and reads the response information of the built-in software program of the controller under test in real time to form an in-loop test closed loop.

[0115] The testing contents involved in software testing include normal function testing of the program, fault injection testing, and exhaustive testing.

[0116] Step 441: Normal function detection

[0117] Build an in-loop detection platform, run the object to be tested, and the detection system drives the simulation equipment to return the simulation status information to the controller to be tested according to the test case requirements, read the response of the built-in software program of the controller to be tested in real time, and compare the output signal with the test case database to achieve normal function detection.

[0118] Step 442: Fault injection detection

[0119] The detection system drives the simulation device to return the simulation status information to the controller under test according to the fault injection detection requirements, reads the response of the built-in software program of the controller under test in real time, and compares the output signal with the fault injection database to achieve fault detection.

[0120] Step 443: Exhaustive detection

[0121] Similarly, the detection system drives the simulation device in accordance with the exhaustive full logic detection requirements, returns the simulation status information to the controller under test, reads the response of the built-in software program of the controller under test in real time, and compares the output signal with the exhaustive database, thereby achieving exhaustive detection.

[0122] Step 5: Dataset Generation

[0123] After obtaining a large amount of data sets through the hardware detection system and the software detection system, the rough set method can be used to establish a rule base.

[0124] The main steps of knowledge acquisition based on rough sets can be summarized as: sample data collection, data preprocessing and continuous attribute discretization, decision table simplification, knowledge rule acquisition, and evaluation and classification of acquisition rules.

[0125] Step 51: Data preprocessing and discretization of continuous attributes

[0126] The purpose of data preprocessing is to remove invalid data. When using rough set theory to process decision tables, the data in the table must be discrete, specifically in the form of character integers and enumerations. Therefore, it is necessary to discretize the attributes. In the decision table, there are two types of conditional attributes and decision attributes:

[0127] 1) Continuous attributes: The value range of the attribute is a continuous interval, and the number of possible values ​​is infinite, such as the signals collected by various sensors in equipment operation monitoring.

[0128] 2) Discrete attributes: the range of the attribute is a finite set, and the number of possible values ​​is limited, such as the stability of the vibration signal of the equipment (stable, unstable), etc. In practice, if the range of the attribute is a finite set, but the number of possible values ​​is large, this attribute will also be treated as continuous.

[0129] When using rough set theory to process decision tables, the data in the table must be discrete, specifically in the form of character integers and enumerations. Therefore, the attributes need to be discretized. Discretization of continuous attributes means determining the breakpoints in the middle, and the breakpoints must be reasonably selected to ensure the effectiveness of the model. Considering that the characteristic frequencies of vibration and noise signals are continuous attributes in the fault diagnosis process, the equal distance and equal frequency methods are used; other signals are mostly state quantity signals (limit switches, etc.), and Boolean logic is used.

[0130] Step 52: Knowledge acquisition and diagnostic model

[0131] In actual diagnosis, the training sample set comes from empirical knowledge or raw data. First, the raw data is described and the condition attribute values ​​are quantified. If it is a continuous attribute, it is discretized and the discretized attribute is assigned to form a decision table. The decision table is simplified. The rules are evaluated using the simplification rate, confidence and coverage to obtain a rule set and store it in the system database.

[0132] These rules become the knowledge source of the fault diagnosis system. Based on the obtained minimum condition attribute set, the actual test sample is constructed, the fault type is directly obtained through rule reasoning, and the judgment result is output.

[0133] Step 3: Establish mapping through BP neural network algorithm

[0134] BP network is a multi-layer feedforward neural network, whose neuron transfer function is an S-type function, and the output is a continuous quantity between 0 and 1, which can realize any nonlinear mapping from input to output. According to the general design scheme, the transfer function of the middle layer neuron is an S-type tangent function, and the transfer function of the output layer neuron is an S-type logarithmic function.

[0135] There are two important guiding principles for determining the structure of the BP network model:

[0136] 1) For general pattern recognition problems, three-layer networks can solve them well.

[0137] 2) In a three-layer network, the number of neurons in the hidden layer n2 and the number of neurons in the input layer n1 can have the following approximate relationship: n2 = 2n1 + 1. In the three-layer network, n2 is the number of neurons in the hidden layer, and n1 is the number of neurons in the input layer.

[0138] The input and output units are established through the neural network, and if necessary, they have specific weights. First, through computer training, the weights are obtained through training, and a specific mathematical model is established. Then the trained mathematical model is used to achieve diagnosis.

[0139] Step 6: Troubleshooting

[0140] Step 61: Data Collection

[0141] Step 611: Vibration and noise data

[0142] The frequency domain analysis method is used to extract characteristic frequency parameters and convert the continuous signal into a discrete eigenvalue signal.

[0143] Step 612: Control Signal

[0144] Control signals generated by the loader control program, such as the target position and speed of the trolleys and gripper control signals.

[0145] Step 613: Status signal

[0146] Motor position, speed, absolute encoder signal, limit switch signal, interlock signal, etc.

[0147] Step 614: Operation Signal

[0148] The operator's operation input signals on the touch screen and panel, such as handle input signals, speed selection signals, fuel selection signals, etc.

[0149] Step 62: Database Contents

[0150] Step 621: Real-time operation data

[0151] This part of data comes from the data generated by the hardware detection system's detection of the verification platform or the on-site detection data, and includes all the content of the above data collection.

[0152] Step 622: Fault Data

[0153] This part of data comes from the detection data generated when the hardware detection platform runs abnormally, the detection data obtained by using fault injection data during software detection, and the fault data expanded and corrected by rough set algorithm diagnosis.

[0154] Step 623: Software detection data set

[0155] The entire data set generated during software testing.

[0156] Step 624: Past event data

[0157] A combination of keywords and data obtained through analysis of past events.

[0158] Step 63: Data Analysis

[0159] Firstly, the real-time data set and fault mode set are determined, and a decision table consisting of a sample library is established. The data completion, attribute reduction, value reduction and rule extraction methods in the rough set theory are used to extract rules, and the acquired knowledge rules are stored in the knowledge base.

[0160] For real-time data sets, the real-time data is input into the inference engine, and the diagnosis result can be obtained through inference according to a certain inference mechanism. After the diagnosis sample is verified, it can be used as a case and input into the original sample database for storage according to certain principles for incremental learning. In this way, as the analysis samples continue to accumulate, the knowledge base continues to accumulate and update, and the diagnostic accuracy of the expert system will gradually improve.

[0161] Step 64: Processing Decisions

[0162] If the diagnosis result is correct, operation and maintenance advice should be given according to the fault type, and a diagnosis report will be output. For fault symptom data, if the program predicts that a fault is about to occur, a fault warning prompt will be issued.

Claims

1. A method for monitoring the state and diagnosing faults of a loading and unloading machine, characterized in that: These include: Step 1: Design functional analysis; Step 2: Build a hardware detection system; Step 3: Build a software detection system; Step 4: Software and hardware testing; Step 5: Dataset generation; Step 6: Troubleshooting.

2. A method for state monitoring and fault diagnosis of a loading and unloading machine as claimed in claim 1, characterized in that: The step 1 comprises: Step 11: Analyze the Original Design Design function analysis includes electrical function analysis, software function analysis, operation process and process flow analysis. Electrical function analysis includes analysis of the electrical design function of the loader and unloader, including drive, signal logic, safety, fault signal, interlock signal, and bypass operation; Software function analysis includes analysis of software functions for management of loading and unloading machine operation process, sensor signal processing and storage, motor drive workflow and feedback processing, redundant signal processing, interlocking protection function, abnormal status information prompt, protection / interlocking threshold setting, and abnormal status judgment; The operation process and process flow analysis includes analyzing the operation process and process flow of the loading and unloading machine, obtaining the components, operating parameters, keywords, and keywords or parameters of the interlocking relationship between components in the original design function, and forming a process flow analysis record table; Step 12: Analyze and summarize past events Collect statistics on past events of loading and unloading machines, summarize the causes of the events, event phenomena, final handling methods of the events, and the magnitude of the events’ impact; Step 13: Create a fault sample database Fault sample database based on past event analysis results.

3. A method for state monitoring and fault diagnosis of a loading and unloading machine as claimed in claim 1, characterized in that: The step 2 includes: the hardware detection system realizes the hardware detection of the loader and unloader by deploying internal and external detection systems, wherein the internal detection system refers to the detection of the sensors and actuators of the loader and unloader itself; the external detection system is the detection by deploying external sensors and collectors to obtain more status information of the loader and unloader.

4. A method for state monitoring and fault diagnosis of a loading and unloading machine as claimed in claim 1, characterized in that: The step 3 includes: the software detection system communicates with the originally designed PLC hardware by building a set of in-loop detection software, and uses the in-loop test system to simulate the signal feedback of the actuator and sensor to achieve a full functional test of the controller.

5. A method for state monitoring and fault diagnosis of a loading and unloading machine as claimed in claim 1, characterized in that: The step 4 comprises: Step 41: Hardware Detection Content The hardware detection content includes the controller operation data, driver operation data, encoder feedback data, limit switch, handwheel, brake wear, balance beam detection and other signals of the loader and unloader, load sensor signals, gripper engagement and release sensors, panel operation area signals, communication signals between various devices, motor vibration signals, and equipment noise signals; Step 42: Hardware Detection Method Step 43: Software Detection Content The software detection object is the control program of the loader and unloader, in which the software modules used to control and operate the loader and unloader are the logic control system, the safety protection system and the communication system with other equipment, including the transfer device and the spent fuel pool crane; Step 44: Software Detection Method The detection method of the software detection system is to build an in-loop detection platform, use in-loop detection equipment to replace the real controlled object, receive control instructions and simulate status information to return to the controller under test. At the same time, the software detection system is connected to the controller under test to read the response information of the built-in software program of the controller under test in real time, forming an in-loop test closed loop. The detection contents involved in software detection include normal function detection of the program, fault injection detection, and exhaustive detection.

6. A method for state monitoring and fault diagnosis of a loading and unloading machine as claimed in claim 5, characterized in that: The step 42 comprises: Step 421: Internal testing method The test hardware of the internal test system consists of PLC, high-frequency data acquisition board, communication interface conversion module, and PC. The test host computer hardware is connected to the field controller through the bus; The host computer software and PLC program of the internal test run on the PC and PLC system of the internal test system respectively. The PLC program is used to run the driver program and convert the sensor data into units and store them in the state variable table. The host computer software obtains the state variable table of the PLC at a certain frequency. Step 422: External testing method The external measurement system has added sensors, and its acquisition hardware and host computer software use independent equipment and programs. The new sensors in the external measurement system are aimed at the main components of the loader and unloader, namely the trolley motor, the trolley motor, the main lifting motor, and the main lifting wire rope, to monitor the vibration and noise data of each motor and the health status of the main lifting wire rope.

7. A method for state monitoring and fault diagnosis of a loading and unloading machine as claimed in claim 5, characterized in that: The step 44 comprises: Step 441: Normal function detection Build an in-loop detection platform, run the object to be tested, and the detection system drives the simulation equipment to return the simulation status information to the controller to be tested according to the test case requirements, read the response of the built-in software program of the controller to be tested in real time, compare the output signal with the test case database, and realize normal function detection; Step 442: Fault injection detection The detection system drives the simulation device to return the simulation status information to the controller under test according to the fault injection detection requirements, reads the response of the built-in software program of the controller under test in real time, compares the output signal with the fault injection database, and realizes fault detection; Step 443: Exhaustive detection The detection system drives the simulation device to return the simulation status information to the controller under test in accordance with the exhaustive full logic detection requirements, reads the response of the built-in software program of the controller under test in real time, and compares the output signal with the exhaustive database to achieve exhaustive detection.

8. A method for state monitoring and fault diagnosis of a loading and unloading machine as claimed in claim 1, characterized in that: The step 5 comprises: Step 51: Data preprocessing and discretization of continuous attributes The purpose of data preprocessing is to remove invalid data. When using rough set theory to process decision tables, the data in the table is required to be discrete, specifically in the form of character integers and enumerations. The attributes are discretized. In the decision table, the values ​​of conditional attributes and decision attributes are divided into two types: Continuous attributes: the value range of the attribute is a continuous interval, and the number of values ​​is infinite, such as the signals collected by various sensors in equipment operation monitoring; Discrete attributes: the range of attribute values ​​is a finite set, and the number of values ​​is finite; When using rough set theory to process decision tables, the data in the table must be discrete, specifically in the form of character integers and enumerations. The attributes are discretized, and the discretization of continuous attributes is to determine the breakpoints in the middle. In the fault diagnosis process, the characteristic frequencies of vibration and noise signals are continuous attributes, and the equal distance and equal frequency methods are used; other signals are state quantity signals, and Boolean logic is used; Step 52: Knowledge acquisition and diagnostic model In actual diagnosis, the training sample set comes from empirical knowledge or original data. First, the original data is described, and the condition attribute value is quantified. If it is a continuous attribute, it is discretized, and the discretized attribute is assigned to form a decision table. The decision table is simplified, and the rules are evaluated using the simplification rate, confidence and coverage to obtain a rule set and store it in the system database; these rules become the knowledge source of the fault diagnosis system, and based on the obtained minimum condition attribute set, the actual test sample is constructed, and the fault type is directly obtained through rule reasoning, and the judgment result is output; Step 3: Establish mapping through BP neural network algorithm BP network is a multi-layer feedforward neural network. The transfer function of its neurons is an S-type function. The output is a continuous quantity between 0 and 1, realizing any nonlinear mapping from input to output. The transfer function of the neurons in the middle layer is an S-type tangent function, and the transfer function of the neurons in the output layer is an S-type logarithmic function.

9. A method for state monitoring and fault diagnosis of a loading and unloading machine as claimed in claim 1, characterized in that: The step 6 comprises: Step 61: Data Collection Step 611: Vibration and noise data Use frequency domain analysis method to extract characteristic frequency parameters and convert continuous signals into discrete characteristic value signals; Step 612: Control Signal Control signals generated by the loader control program; Step 613: Status signal Motor position, speed, absolute encoder signal, limit switch signal, interlock signal; Step 614: Operation Signal The operator's operation input signals on the touch screen and panel, including handle input signals, speed selection signals, and fuel selection signals; Step 62: Database Contents Step 621: Real-time operation data This part of data comes from the data generated by the hardware detection system on the verification platform or the on-site detection data; Step 622: Fault Data This part of data comes from the detection data generated when the hardware detection platform runs abnormally, the detection data obtained by using fault injection data during software detection, and the fault data expanded and corrected by rough set algorithm diagnosis; Step 623: Software detection data set The entire data set generated during software testing; Step 624: Past event data Keywords and data combinations obtained through analysis of past events; Step 63: Data Analysis Firstly, the real-time data set and the fault mode set are determined, and a decision table consisting of a sample library is established. The data completion, attribute reduction, value reduction and rule extraction methods in the rough set theory are used to extract rules, and the acquired knowledge rules are stored in the knowledge base. For real-time data sets, the real-time data is input into the inference engine, and the diagnosis result can be obtained through reasoning according to a certain reasoning mechanism. After the diagnosis sample is verified, it is used as a case and input into the original sample database for storage according to certain principles, and incremental learning is carried out, and this cycle repeats. Step 64: Processing Decisions If the diagnosis result is correct, operation and maintenance advice should be given according to the fault type, and finally a diagnosis report should be output. For fault symptom data, if the program predicts that a fault is about to occur, a fault warning prompt will be issued.

Images