A multi-agent based fault diagnosis method for tractor electrical system

Through the multi-agent system combined with BDI model and fuzzy neural network, early detection and difference of tractor electrical system failures is achieved, and the problems of insufficient predictiveness and low diagnostic accuracy in traditional fault diagnosis methods are solved, and fault evaluation and maintenance are improved efficiency.

CN114995355BActive Publication Date: 2024-11-12MAHINDRA YUEDA YANCHENG TRACTOR +1
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202210732040.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-06-26
Publication Date
2024-11-12
Estimated Expiration
2042-06-26

AI Technical Summary

Technical Problem

The traditional fault diagnosis method of electrical system of tractors lacks predictability, resulting in an expansion of fault hazards, high diagnosis time and cost, and it is difficult to distinguish different faults and identify the timely change characteristics of fault sizes.

Method used

Adopt fault diagnosis methods based on multi-agents, including coordination of agents, status monitoring agents, system management agents, fault diagnosis agents and fault output agents, use BDI models and fuzzy neural networks to diagnose electrical system faults, and early detection Fault signals are responded in advance and maintained synchronization through multi-agent systems to improve diagnostic accuracy and timeliness.

Benefits of technology

It realizes early fault detection, reduces fault hazards, distinguishes different faults and distinguishes fault sizes, improves fault evaluation and maintenance efficiency, and significantly improves the accuracy and timeliness of fault diagnosis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114995355B_ABST
    Figure CN114995355B_ABST
Patent Text Reader

Abstract

The present invention discloses a tractor electrical system fault diagnosis method based on multi-agents, and the specific steps are as follows: Step 1: When the tractor starts running, the state monitoring agent performs online real-time monitoring; Step 2: When the state monitoring agent sends out a warning or alarm signal, a diagnosis request is sent to the fault diagnosis agent through the fault diagnosis user interface; Step 3: The fault phenomenon observed by the user is submitted to the state monitoring agent, and the data is collected through the sensor, and the current when the fault state is collected is collected. I and voltage U The collected electrical fault data are sent together to the fault diagnosis agent system; Step 4: Based on the collected electrical fault data, the system management agent quickly locates the corresponding electrical system subsystem, and then manages and coordinates the fault diagnosis task; Step 5: Locate the subsystem, accurately perform fault diagnosis, and obtain the original diagnosis data through the status monitoring agent to complete the fault diagnosis of the tractor electrical system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of fault diagnosis, and in particular to a multi-agent-based tractor electrical system fault diagnosis method. Background Art

[0002] When a traditional tractor electrical system fails, professional maintenance personnel are required to arrive at the site of the tractor failure and use professional instruments and equipment to repair and diagnose the tractor. The investigation found that when a tractor fails, it takes 70% of the total maintenance time to find the cause of the tractor failure, while it only takes 30% of the time to eliminate the fault. At the same time, traditional maintenance requires professional maintenance personnel, which leads to high maintenance costs. Therefore, it is particularly important to establish an intelligent fault diagnosis method for tractors. The document with the national patent publication number CN112631261A proposes a remote fault diagnosis system and method for high-power tractors. The fault diagnosis model is based on the BP neural network, and the particle swarm optimization algorithm is used to optimize the weights and thresholds of the BP neural network. At the same time, the multi-swarm co-evolution algorithm is used to further optimize the particle swarm algorithm, so that information is shared between multiple populations to prevent falling into the local minimum, enhance the global search ability of the particle swarm algorithm, and effectively improve the diagnostic accuracy of the fault diagnosis model. However, the patent does not take into account the predictive nature of tractor fault diagnosis, has a certain time lag, and often further expands the harm of tractor failure.

[0003] Use new content

[0004] The technical problem to be solved by the present invention is to provide a tractor electrical system fault diagnosis method based on multi-agent, which can not only detect fault signals at an early stage, respond to faults in advance, and reduce the harm of system faults, but also distinguish different faults and identify the size and time-varying characteristics of faults, which is conducive to the evaluation and maintenance of faults.

[0005] In order to solve the above technical problems, the present invention provides a tractor electrical system fault diagnosis method based on multi-agent, wherein the multi-agent includes a coordination agent, a state monitoring agent, a system management agent, a fault diagnosis agent and a fault output agent; when a fault occurs in the tractor electrical system, the system management agent initially receives the diagnosis request from the state monitoring agent, quickly locates the corresponding subsystem, and manages, allocates and coordinates the fault diagnosis task at the same time; the state detection agent is responsible for obtaining the necessary information from each monitoring node, sending the information to the fault diagnosis agent, providing original evidence for the specific fault diagnosis, and also collecting and processing the information of the voltage and current sensors on the tractor body to form an alarm signal; the fault diagnosis agent uses a fuzzy neural network based on the BDI model to complete the fault diagnosis of the electrical system; during the diagnosis, the system management agent uses a fuzzy neural network to complete the fault diagnosis of the electrical system; The user sends a diagnosis request to the fault diagnosis agent through the fault diagnosis user interface in the tractor cab, and submits the observed fault phenomenon and parameters such as current I and voltage U measured by sensors to the state monitoring agent to form the initial evidence for the multi-agent startup. The fault output agent is used to store the global database of information, record the information required by each agent and the decision conclusions generated by each agent, and provide them to other agents for sharing. At the same time, the diagnosis results are recorded. When the fault diagnosis agent needs data, it reads the data from the state monitoring agent. When the diagnosis is completed, the diagnosis results are sent to the fault output agent. The coordination agent is introduced so that the multi-agent system can still maintain synchronization when encountering communication interruption, reconnection and communication delay, so as to improve the flexibility, adaptability and robustness of the entire control system. The specific steps are as follows:

[0006] Step 1: When the tractor starts running, the condition monitoring agent performs online real-time monitoring of the tractor electrical system;

[0007] Step 2: When the condition monitoring agent issues a warning or alarm signal, the user sends a diagnosis request to the fault diagnosis agent through the fault diagnosis user interface in the tractor cockpit;

[0008] Step 3: The fault phenomenon observed by the user is submitted to the condition monitoring agent. At the same time, the condition monitoring agent collects data through sensors and sends the current I and voltage U collected during the fault state to the fault diagnosis agent system to form the original evidence for fault diagnosis.

[0009] Step 4: Based on the electrical fault data collected by the condition monitoring agent, the system management agent quickly locates the corresponding electrical system subsystem. Then, the system management agent manages and coordinates the fault diagnosis tasks of the electrical system subsystem and completes the information docking with the fault diagnosis agent.

[0010] Step 5: The fault diagnosis agent locates the subsystem through the system management agent, accurately performs fault diagnosis, and obtains the original diagnosis data through the condition monitoring agent to complete the fault diagnosis of the tractor electrical system. The specific electrical system diagnosis process is as follows:

[0011] Step 5.1: The fault diagnosis agent obtains the fault symptom current set I and voltage set U from the condition monitoring agent, and constructs the fault symptom sets I and U:

[0012] I={i1,i2,…,i m}

[0013] U={u1,u2,…,u m}

[0014] Among them, i s (s=1,2,…,m) and u s (s=1,2,…,m) indicates the current and voltage signs of fault occurrence;

[0015] Step 5.2: Obtain the fuzzy vector of the fault symptom current set I and voltage set U:

[0016]

[0017]

[0018] in, and They are the current and voltage signs of fault occurrence. s and u s The degree of membership;

[0019] Step 5.3: The fuzzy vectors of the fault symptom current set I and voltage set U Stored in the belief library;

[0020] Step 5.4: Obtain the fault cause set Y from the fault output agent, such as fuse blown, starter switch disconnected, line short circuit and short circuit, and construct the fault cause set Y:

[0021] Y={y1,y2,…,y n}

[0022] Among them, y t (t=1,2,…,n) represents the cause of the tractor electrical system failure;

[0023] Step 5.5: Determine the fuzzy vector of the fault cause set Y:

[0024]

[0025] in, Is the cause of the fault t The degree of membership;

[0026] Step 5.6: The fuzzy vector of the fault cause set Y Stored in the wish library,

[0027] Step 5.7: Establish a fuzzy matrix between the fault symptom vector and the fault cause vector. This matrix is ​​the connection weight matrix of the fuzzy neural network:

[0028]

[0029] Each connection weight in the matrix also represents a fuzzy relationship from the fault phenomenon to the fault cause, that is, c st The mapping value of the sth feature describing the fault symptom to the tth fault cause;

[0030] Step 5.8: Introduce fuzzy neural network into the desire reasoner, convert rules and reasoning into neural network mapping processing; transform the fuzzy vectors of fault symptom current set I and voltage set U in the belief library into As the input layer of the fuzzy neural network. According to the empirical formula p2 = 2p + 1, where p2 is the number of hidden layer neurons and p is the number of input neurons, 5 are selected. Too many or too few hidden layer neurons are not conducive to improving the accuracy of fault diagnosis. The fuzzy vector of the fault cause set is As the output layer of the fuzzy neural network;

[0031] Step 5.9: Establish a fault diagnosis model:

[0032]

[0033] in, is the characteristic function symbol, In the tractor electrical fault diagnosis model, take:

[0034]

[0035] Step 5.10: In the fault diagnosis of tractor electrical system fuzzy neural network, the fuzzy matrix is ​​obtained by the neural network through learning the fault diagnosis samples. Assume that there are X groups of learning samples, each group of learning samples contains the fault symptom fuzzy vector and and the fuzzy vector of the fault cause set For the above c st Assign initial value: Let c st =1, yes X=1;

[0036] Step 5.11: Given Input and Output Input each group of fault symptom and fault cause fuzzy vector pairs as the input and output mode of the fuzzy neural network. The input mode is Output mode is

[0037] Step 5.12: Calculate the theoretical actual output,

[0038]

[0039] in, Represents the actual output of the Xth learning sample for the tth component during training. and Characterizes the sth component of the input pattern, c st is the connection weight from the sth node in I and U to the tth node in Y;

[0040] Step 5.13: Adjust the weights and set but

[0041]

[0042] Where η is the proportional factor, satisfying 0<η≤1;

[0043] Step 5.14: Verify that there exists c for all s, t st (z+1)=c st (z), if it exists, proceed to step 20, otherwise return to step 17;

[0044] Step 5.15: Let X = X + 1, and repeat from step 16 until all groups of learning samples are finished. Through continuous learning in actual application, the corrected fuzzy matrix c in fuzzy fault diagnosis can be obtained. st , thereby improving the accuracy and reliability of system diagnosis;

[0045] Step 5.16: The wish reasoner continuously trains the fuzzy neural network based on the knowledge in the wish library, thereby improving the accuracy of the diagnosis results of the entire system. After repeated training and learning, the knowledge in the wish library is continuously updated;

[0046] Step 5.17: Finally, the desire reasoner sends the fault diagnosis result to the intention library and stores it;

[0047] Step 5.18: The fault diagnosis agent sends the diagnosis result to the fault output agent;

[0048] Step 6: Based on the SAE J1939 protocol, the fault output agent sends the electrical system fault diagnosis results to the user interface. At this point, the tractor electrical system fault diagnosis is completed.

[0049] Beneficial effects of the present invention:

[0050] The present invention is based on a multi-agent tractor electrical system fault diagnosis method. The multi-agent diagnosis system can detect fault signals at an early stage, respond to the fault in advance, and reduce the harm of system faults. The multi-agent diagnosis system has the ability to distinguish different faults and identify the size and time-varying characteristics of the fault, which is beneficial to the evaluation and maintenance of the fault. At the same time, through the training of fuzzy neural networks, the accuracy and timeliness of fault diagnosis are effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a structural diagram of a multi-agent-based tractor electrical system fault diagnosis method;

[0052] Figure 2 It is a fault diagnosis agent based on the BDI model;

[0053] Figure 3 A flowchart for tractor electrical system fault diagnosis;

[0054] Figure 4 Schematic diagram of fuzzy neural network structure. DETAILED DESCRIPTION

[0055] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0056] In the description of the present invention, it should be noted that the terms "upper", "lower", "inner", "outer", "front end", "rear end", "two ends", "one end", "the other end" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0057] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "provided with", "connected", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0058] A fault diagnosis method for tractor electrical system based on multi-agent. The electrical system of the tractor is mainly composed of power supply system, starting system, instrument alarm part, lighting signal system, auxiliary electrical part and electro-hydraulic control system. The power supply system of the tractor is mainly composed of generator, battery, regulator and charging device, which is the starting point for the normal operation of the electrical system; the starting system of the tractor is mainly composed of four parts: starter, ignition switch, relay and preheater; the instrument alarm part of the tractor is mainly composed of sensors and instruments, and its main function is to feedback the working status of related parts of the tractor during operation to the driver; the lighting signal system of the tractor is mainly composed of signal lights and lighting lights composed of hazard warning lights and high and low beam lights, which are the main power-consuming parts of the tractor electrical system; the auxiliary electrical system of the tractor is mainly responsible for providing a comfortable riding environment for the cockpit; the electro-hydraulic control system of the tractor greatly improves the power, economy and comfort of the tractor.

[0059] like Figure 1 As shown in the figure, the multi-agent includes a coordination agent, a state monitoring agent, a system management agent, a fault diagnosis agent and a fault output agent. The system management agent is the basis for the operation of the fault diagnosis of the tractor electrical system, and is mainly responsible for the task management, allocation and coordination of the fault diagnosis system. When the tractor electrical system sends an alarm signal or a fault occurs, the corresponding subsystem where the fault occurs can be quickly located, thereby greatly shortening the fault diagnosis time. The fault diagnosis agent is an agent that looks for the "cause" of the tractor electrical system. The fault diagnosis agent is based on the BDI model, such as Figure 2As shown. The fault diagnosis agent is mainly composed of a belief library, a desire library, an intention library, a belief generator and a desire reasoner. The belief library is used to store fault diagnosis indicators; the desire library is used to store fault causes; the intention library is used to determine the fault diagnosis results; the belief generator is used to generate fault diagnosis indicators; and the desire reasoner is used to generate fault mapping relationships. When the fault diagnosis agent is working, it first obtains the original data when the tractor electrical system is started from the state detection agent and sends it to the belief generator. The belief generator generates indicators for fault diagnosis and stores them in the belief library. The desire reasoner generates a fault diagnosis mapping relationship through the fault diagnosis indicators and fault causes in the belief library and the desire library. By using a trained fuzzy neural network, the final diagnosis result is sent to the intention library. After the intention library integrates the information, it sends the fault diagnosis result to the fault output agent. The fault output agent sends the diagnosis results in the fault diagnosis agent to the user interface based on the protocol SAE J1939. The introduction of coordination agents enables the multi-agent system to maintain synchronization when encountering communication interruptions, reconnections, and communication delays, thereby improving the flexibility, adaptability, and robustness of the entire control system. Information is shared between each agent. By storing the previous diagnosis results in the intention library of the fault diagnosis agent, and then sending the diagnosis results to the intention library after each fault diagnosis, the intention library is continuously improved.

[0060] When a fault occurs in the tractor electrical system, the system management agent initially receives the diagnosis request from the state monitoring agent and quickly locates the corresponding subsystem. At the same time, it manages, allocates and coordinates the fault diagnosis task. The state detection agent is responsible for obtaining the necessary information from each monitoring node and sending this information to the fault diagnosis agent to provide original evidence for the specific fault diagnosis. It also collects and processes the information of the voltage and current sensors on the tractor body to form an alarm signal. The fault diagnosis agent is based on the BDI model and uses fuzzy neural networks to complete the fault diagnosis of the electrical system. During the diagnosis, the user sends a diagnosis request to the fault diagnosis agent through the fault diagnosis user interface in the tractor cab, and submits the observed fault phenomenon and the parameters such as current I and voltage U measured by the sensor to the state monitoring agent to form the initial evidence for the multi-agent startup. The fault output agent is used to store the global database of information, record the information required by each agent and the decision conclusions generated by each agent, and provide it to other agents for sharing. At the same time, it records the diagnosis results. When the fault diagnosis agent needs data, it reads data from the state monitoring agent. When the diagnosis is completed, the diagnosis results are sent to the fault output agent, and the coordination agent is introduced so that the multi-agent system can still maintain synchronization when encountering communication interruption, reconnection and communication delay, so as to improve the flexibility, adaptability and robustness of the entire control system.

[0061] A multi-agent-based tractor electrical system fault diagnosis method, the specific steps are as follows:

[0062] Step 1: When the tractor starts running, the condition monitoring agent performs online real-time monitoring of the tractor electrical system;

[0063] Step 2: When the condition monitoring agent issues a warning or alarm signal, the user sends a diagnosis request to the fault diagnosis agent through the fault diagnosis user interface in the tractor cockpit;

[0064] Step 3: The fault phenomenon observed by the user is submitted to the condition monitoring agent. At the same time, the condition monitoring agent collects data through sensors and sends the current I and voltage U collected during the fault state to the fault diagnosis agent system to form the original evidence for fault diagnosis.

[0065] Step 4: Based on the electrical fault data collected by the condition monitoring agent, the system management agent quickly locates the corresponding electrical system subsystem. Then, the system management agent manages and coordinates the fault diagnosis tasks of the electrical system subsystem and completes the information docking with the fault diagnosis agent.

[0066] Step 5: The fault diagnosis agent locates the subsystem through the system management agent, accurately performs fault diagnosis, and obtains the original diagnosis data through the condition monitoring agent to complete the fault diagnosis of the tractor electrical system. The specific electrical system diagnosis process is as follows:

[0067] Step 5.1: The fault diagnosis agent obtains the fault symptom current set I and voltage set U from the condition monitoring agent, and constructs the fault symptom sets I and U:

[0068] I={i1,i2,…,i m}

[0069] U={u1,u2,…,u m}

[0070] Among them, i s (s = 1, 2, ..., m) and u s (s=1, 2, ..., m) indicates the current and voltage signs of fault occurrence;

[0071] Step 5.2: Obtain the fuzzy vector of the fault symptom current set I and voltage set U:

[0072]

[0073]

[0074] in, and They are the current and voltage signs of fault occurrence. s and u s The degree of membership;

[0075] Step 5.3: The fuzzy vectors of the fault symptom current set I and voltage set U Stored in the belief library;

[0076] Step 5.4: Obtain the fault cause set Y from the fault output agent, such as fuse blown, starter switch disconnected, line short circuit and short circuit, and construct the fault cause set Y:

[0077] Y={y1,y2,…,y n}

[0078] Among them, y t (t=1, 2, ..., n) represents the cause of the tractor electrical system failure;

[0079] Step 5.5: Determine the fuzzy vector of the fault cause set Y:

[0080]

[0081] in, Is the cause of the fault t The degree of membership;

[0082] Step 5.6: The fuzzy vector of the fault cause set Y Stored in the wish library,

[0083] Step 5.7: Establish a fuzzy matrix between the fault symptom vector and the fault cause vector. This matrix is ​​the connection weight matrix of the fuzzy neural network:

[0084]

[0085] Each connection weight in the matrix also represents a fuzzy relationship from the fault phenomenon to the fault cause, that is, c st The mapping value of the sth feature describing the fault symptom to the tth fault cause;

[0086] Step 5.8: Introduce fuzzy neural network into the desire reasoner, convert rules and reasoning into neural network mapping processing; transform the fuzzy vectors of fault symptom current set I and voltage set U in the belief library into As the input layer of the fuzzy neural network. According to the empirical formula p2 = 2p + 1, where p2 is the number of hidden layer neurons and p is the number of input neurons, 5 are selected. Too many or too few hidden layer neurons are not conducive to improving the accuracy of fault diagnosis. The fuzzy vector of the fault cause set is As the output layer of the fuzzy neural network;

[0087] Step 5.9: Establish a fault diagnosis model:

[0088]

[0089] in, is the characteristic function symbol, In the tractor electrical fault diagnosis model, take:

[0090]

[0091] Step 5.10: In the fault diagnosis of tractor electrical system fuzzy neural network, the fuzzy matrix is ​​obtained by the neural network through learning the fault diagnosis samples. Assume that there are X groups of learning samples, each group of learning samples contains the fault symptom fuzzy vector and and the fuzzy vector of the fault cause set For the above c st Assign initial value: Let c st =1, yes

[0092] Step 5.11: Given Input and Output Input each group of fault symptom and fault cause fuzzy vector pairs as the input and output mode of the fuzzy neural network. The input mode is Output mode is

[0093] Step 5.12: Calculate the theoretical actual output,

[0094]

[0095] in, Represents the actual output of the Xth learning sample for the tth component during training. and Characterizes the sth component of the input pattern, c st is the connection weight from the sth node in I and U to the tth node in Y;

[0096] Step 5.13: Adjust the weights and set but

[0097]

[0098] Where η is the proportional factor, satisfying 0<η≤1;

[0099] Step 5.14: Verify that there exists c for all s, t st (z+1)=c st(z), if it exists, proceed to step 20, otherwise return to step 17;

[0100] Step 5.15: Let X = X + 1, and repeat from step 16 until all groups of learning samples are finished. Through continuous learning in actual application, the corrected fuzzy matrix c in fuzzy fault diagnosis can be obtained. st , thereby improving the accuracy and reliability of system diagnosis;

[0101] Step 5.16: The wish reasoner continuously trains the fuzzy neural network based on the knowledge in the wish library, thereby improving the accuracy of the diagnosis results of the entire system. After repeated training and learning, the knowledge in the wish library is continuously updated;

[0102] Step 5.17: Finally, the desire reasoner sends the fault diagnosis result to the intention library and stores it;

[0103] Step 5.18: The fault diagnosis agent sends the diagnosis result to the fault output agent;

[0104] Step 6: Based on the SAE J1939 protocol, the fault output agent sends the electrical system fault diagnosis results to the user interface. At this point, the tractor electrical system fault diagnosis is completed.

[0105] The above-mentioned multi-agent-based tractor electrical system fault diagnosis method detects fault signals early through the multi-agent diagnosis system, responds to the fault in advance, and reduces the harm of system faults; the ability of the multi-agent diagnosis system to distinguish different faults and identify the size and time-varying characteristics of the fault is beneficial to the evaluation and repair of the fault; at the same time, through the training of fuzzy neural networks, the accuracy and timeliness of fault diagnosis are effectively improved.

[0106] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A multi-agent-based tractor electrical system fault diagnosis method, wherein the multi-agent includes a coordination agent, a state monitoring agent, a system management agent, a fault diagnosis agent, and a fault output agent; when a fault occurs in the tractor electrical system, the system management agent initially receives the diagnosis request from the state monitoring agent, quickly locates the corresponding subsystem, and manages, allocates, and coordinates the fault diagnosis task; the state detection agent is responsible for obtaining the necessary information from each monitoring node, sending the information to the fault diagnosis agent, providing original evidence for the specific fault diagnosis, and also collecting and processing the information of the voltage and current sensors on the tractor body to form an alarm signal; the fault diagnosis agent uses a fuzzy neural network based on the BDI model to complete the fault diagnosis of the electrical system. During the diagnosis, the user can The fault diagnosis user interface in the cab sends a diagnosis request to the fault diagnosis agent, and submits the observed fault phenomenon and parameters such as current I and voltage U measured by sensors to the condition monitoring agent to form the initial evidence for the multi-agent startup. The fault output agent is used to store the global database of information, record the information required by each agent and the decision conclusions generated by each agent, and provide them to other agents for sharing. At the same time, the diagnosis results are recorded. When the fault diagnosis agent needs data, it reads the data from the condition monitoring agent. When the diagnosis is completed, the diagnosis results are sent to the fault output agent. The coordination agent is introduced so that the multi-agent system can still maintain synchronization when encountering communication interruption, reconnection and communication delay, so as to improve the flexibility, adaptability and robustness of the entire control system. The specific steps are as follows: Step 1: When the tractor starts running, the condition monitoring agent monitors the tractor electrical system online in real time; Step 2: When the condition monitoring agent issues a warning or alarm signal, the user sends a diagnosis request to the fault diagnosis agent through the fault diagnosis user interface in the tractor cockpit; Step 3: The fault phenomenon observed by the user is submitted to the condition monitoring agent. At the same time, the condition monitoring agent collects data through sensors and sends the current I and voltage U collected during the fault state to the fault diagnosis agent system to form the original evidence for fault diagnosis. Step 4: Based on the electrical fault data collected by the condition monitoring agent, the system management agent quickly locates the corresponding electrical system subsystem. Then, the system management agent manages and coordinates the fault diagnosis tasks of the electrical system subsystem and completes the information docking with the fault diagnosis agent. Step 5: The fault diagnosis agent locates the subsystem through the system management agent, accurately performs fault diagnosis, and obtains the original diagnosis data through the condition monitoring agent to complete the fault diagnosis of the tractor electrical system. The specific electrical system diagnosis process is as follows: Step 5.1: The fault diagnosis agent obtains the fault symptom current set I and voltage set U from the condition monitoring agent, and constructs the fault symptom sets I and U: I={i1,i2,...,i m} U={u1,u2,...,u m} in, i s (s=1,2,…,m) and u s (s=1,2,…,m) indicates the current and voltage signs of fault occurrence; Step 5.2: Calculate the fuzzy vector of the fault symptom current set I and voltage set U: in, and They are the current and voltage signs of fault occurrence. s and u s The degree of membership; Step 5.3: The fuzzy vectors of the fault symptom current set I and voltage set U Stored in the belief library; Step 5.4: Obtain the fault cause set Y from the fault output agent, such as fuse blown, starter switch disconnected, line short circuit and short circuit, and construct the fault cause set Y: Y={y1,y2,...,y n } Among them, y t (t=1,2,…,n) represents the cause of the tractor electrical system failure; Step 5.5: Determine the fuzzy vector of the fault cause set Y: in, Is the cause of the fault t The degree of membership; Step 5.6: The fuzzy vector of the fault cause set Y Stored in the wish library, Step 5.7: Establish a fuzzy matrix between the fault symptom vector and the fault cause vector. This matrix is ​​the connection weight matrix of the fuzzy neural network: Each connection weight in the matrix also represents a fuzzy relationship from the fault phenomenon to the fault cause, that is, c st The mapping value of the sth feature describing the fault symptom to the tth fault cause; Step 5.8: Introduce fuzzy neural network into the desire reasoner, convert rules and reasoning into neural network mapping processing; transform the fuzzy vectors of fault symptom current set I and voltage set U in the belief library into As the input layer of the fuzzy neural network; according to the empirical formula p2 = 2p + 1, where p2 is the number of hidden layer neurons and p is the number of input neurons, 5 are selected. Too many or too few hidden layer neurons are not conducive to improving the accuracy of fault diagnosis. The fuzzy vector of the fault cause set is As the output layer of the fuzzy neural network; Step 5.9: Establish a fault diagnosis model: in, is the characteristic function symbol, In the tractor electrical fault diagnosis model, take: Step 5.10: In the fault diagnosis of tractor electrical system fuzzy neural network, the fuzzy matrix is ​​obtained by the neural network through learning the fault diagnosis samples. Assume that there are X groups of learning samples, each group of learning samples contains the fault symptom fuzzy vector and and the fuzzy vector of the fault cause set For the above c st Assign initial value: Let c st =1, yes X=1; Step 5.11: Given Input and Output Input each group of fault symptom and fault cause fuzzy vector pairs as the input and output mode of the fuzzy neural network respectively. The input mode is Output mode is Step 5.12: Calculate the theoretical actual output, in, Represents the actual output of the Xth learning sample to the tth component during training; and Characterizes the sth component of the input pattern, c st is the connection weight from the sth node in I and U to the tth node in Y; Step 5.13: Adjust the weights and set but Where η is the proportional factor, satisfying 0<η≤1; Step 5.14: Verify that there exists c for all s, t st (z+1)=c st (z), if it exists, proceed to step 20, otherwise return to step 17; Step 5.15: Let X = X + 1, and repeat from step 16 until all groups of learning samples are finished. Through continuous learning in actual application, the corrected fuzzy matrix c in fuzzy fault diagnosis can be obtained. st , thereby improving the accuracy and reliability of system diagnosis; Step 5.16: The wish reasoner continuously trains the fuzzy neural network based on the knowledge in the wish library, thereby improving the accuracy of the diagnosis results of the entire system. After repeated training and learning, the knowledge in the wish library is continuously updated; Step 5.17: Finally, the desire reasoner sends the fault diagnosis result to the intention library and stores it; Step 5.18: The fault diagnosis agent sends the diagnosis result to the fault output agent; Step 6: Based on the SAE J1939 protocol, the fault output agent sends the electrical system fault diagnosis results to the user interface. At this point, the tractor electrical system fault diagnosis is completed.

Citation Information

Cited By

  • Scalable electric power controls system using stateless services

    US12348356B2

  • Scalable electric power controls system using stateless services

    US20240243965A1