Abstract: Based on an in-depth study of the working principle of air-circulating refrigeration equipment, fault tree analysis and expert systems are introduced into the fault diagnosis of this equipment, and a fault diagnosis expert system for this equipment is designed. The theoretical basis and design method of the system are given in detail, and the design of the knowledge base and inference engine is described in detail. This system can quickly and accurately identify the fault location, improving maintenance efficiency. Keywords: air circulating refrigeration equipment; fault tree; knowledge base; inference engine Abstract: Based on in-depth research of the working principle of air circulating refrigeration equipment, fault tree and expert system are introduced to fault diagnosis of air circulating refrigeration equipment. A fault diagnosis expert system of that equipment is successfully designed. This paper introduces the basic scheme and designing theory and gives a detailed description of the design of the knowledge base and inference engine. The system can find the fault part quickly and accurately, thus improving the efficiency of engineering maintenance. Keywords: air circulating refrigeration equipment; fault tree; knowledge base; inference engine 1 Introduction Air circulating refrigeration equipment is used to provide dry and clean cold air, hot air, and ventilation at a given temperature and humidity to the aircraft equipment compartment when the aircraft engine is off and the ground is powered on for inspection and maintenance of aircraft electrical and electronic equipment, in order to control the working environment conditions of the aircraft electrical and electronic equipment. The equipment uses air compression and expansion technology to achieve cooling and heating, which is similar to the principle of the onboard environmental control system of some new aircraft in recent years. The difference is that the onboard environmental control system uses compressed air from the aircraft engine as bleed air, while the equipment uses compressed air from an air compressor as the air power source [1]. Due to the complex structure and harsh working environment of the equipment, there are many reasons for equipment failure. On-site analysis, judgment and handling of failures often depend on the maintenance personnel's grasp of the equipment failure mechanism and experience, which has a certain degree of subjectivity. This inevitably leads to mistakes due to insufficient experience, delaying maintenance time. Establishing an expert system for fault diagnosis of air circulation refrigeration equipment can greatly reduce the blindness of repair, improve economy and safety, and has important economic benefits and application prospects. This paper combines fault tree analysis and expert system and introduces them into the fault diagnosis of air circulation refrigeration equipment. It not only gives full play to the characteristics of rapid and effective diagnosis of expert system, but also uses fault tree to solve the bottleneck problem of acquiring diagnostic knowledge, thereby ensuring the integrity of diagnostic knowledge. 2 Composition of the expert system for fault diagnosis of air circulation refrigeration equipment 2.1 Advantages of the fault tree-based expert system for fault diagnosis [2-6] (1) All normal working modes and fault modes of the system can be determined based on the minimum path set and minimum cut set of the fault tree. (2) The probability of each fault mode can be calculated based on the probability of occurrence of the bottom event. The fault modes are sorted according to the probability of occurrence, and the influence of each fault mode causing the system fault can be determined. (3) It has strong logic and is not easy to miss the cause of the fault. Starting from the top event of the fault tree, through a rigorous step-by-step analysis, all the causes that can cause the fault can be found and maintenance opinions can be provided. (4) The cause of the fault is clear at a glance. All factors that cause the fault can be obtained from the fault tree, which can avoid blindness when troubleshooting and improve efficiency. (5) Utilizing the characteristics of relational databases, the knowledge base can be quickly modified and browsed, which is convenient for experts to check and update knowledge; the modular design makes it easy to separate the knowledge base, inference engine and application program, which helps to maintain the entire expert system. 2.2 Composition of the Fault Diagnosis Expert System The fault diagnosis expert system for air-circulating refrigeration equipment includes a human-machine interface, knowledge base, inference engine, knowledge acquisition subsystem, interpretation subsystem, and comprehensive database. Figure 1 shows the structure of this expert system. [align=center] Figure 1 Structure of the Expert System[/align] The fault diagnosis expert system for air-circulating refrigeration equipment uses expert knowledge about the equipment to solve problems that require expert intervention in practice. The design of the knowledge base and inference engine is particularly important in the system's construction. 3 Fault Diagnosis Expert System Knowledge Base The knowledge base mainly stores domain knowledge provided by domain experts and is an important component of the entire expert system. The quantity and quality of the knowledge it possesses are important factors in measuring the performance and problem-solving capabilities of the expert system. 3.1 Knowledge Acquisition Possessing knowledge is a key characteristic that distinguishes expert systems from other computer software systems, and the quality and quantity of knowledge are crucial factors determining the performance of the expert system. How to enable the expert system to acquire high-quality knowledge is precisely the problem that knowledge acquisition aims to solve. Knowledge acquisition employs a fault tree construction method. Fault trees possess a standardized knowledge structure. Utilizing fault tree knowledge to generate an expert system knowledge base not only significantly reduces the difficulty of knowledge acquisition but also simplifies the generated knowledge base as much as possible by solving for minimal cut sets, reducing redundancy and facilitating reasoning. Before constructing the system's knowledge base, a fault tree is first built from the rich and complex fault diagnosis knowledge. Then, the relationships between various fault phenomena and causes are analyzed and reflected in the rules. In fault tree construction, the air circulation refrigeration equipment fault is used as the root node, and 24 common faults, such as abnormal noise and malfunctioning electric valves, are used as secondary nodes to form the first fault tree, i.e., the main tree (Figure 2). The corresponding event list is shown in Table 1. Then, 24 sub-fault trees are constructed using these 24 common faults as root nodes. In the ordering of the leaf nodes (corresponding to basic events) in these 24 sub-fault trees, events that are easier to check are placed first to align with the thinking of experienced workers and maximize matching efficiency. The corresponding event list is shown in Table 1. [align=center]Figure 2 Schematic diagram of the overall fault tree[/align] Table 1 Event codes of the overall fault tree The fault diagnosis procedure first locates the sub-fault tree corresponding to the fault phenomenon, and then submits the highest-ranked event to the maintenance personnel for judgment. If the judgment result is a basic event, it cannot be expanded further, and the diagnosis stops; if it is not a basic event, the judgment continues according to the prompts. During the diagnosis process, the fault tree provides corresponding test guidelines or diagnostic judgment data, and finally provides maintenance strategies and corresponding fault tree diagnostic trajectories. 3.2 Knowledge Representation Knowledge representation is the process of symbolizing and formalizing knowledge. In an expert system, the choice of knowledge representation mode is not only related to the effective storage of knowledge, but also directly affects the traditional knowledge acquisition capability and the efficiency of knowledge application. Therefore, knowledge representation is one of the most fundamental problems in knowledge engineering. Since the fault diagnosis knowledge base of this expert system is built based on the fault tree analysis method, knowledge generally has the characteristics of being empirical and causal. The conclusion knowledge in the knowledge base has a hierarchical relationship unique to fault trees. That is to say, a conclusion knowledge in the knowledge base may be the antecedent of another conclusion knowledge, which is very suitable for production rules to organize the knowledge base. Therefore, this system uses production rules to represent fault diagnosis knowledge, represented as IF E THEN H (CF(H, E)). Here, E is the premise of the knowledge, which can be a single condition or a composite condition connected by AND or OR; H is the conclusion, which can be a single conclusion or multiple conclusions. CF(H, E) is the credibility of this knowledge, called the credibility factor. CF(H, E) takes values ​​in the range [-1, 1], indicating the degree of support for the conclusion to be true when the evidence corresponding to the premise E is true. It is usually given directly by domain experts. The larger the value, the more truthful the corresponding knowledge. When the value of CF is 1, it indicates that the corresponding knowledge is true; when the value of CF is -1, it indicates that the corresponding knowledge is false. 3.3 Knowledge Base Management Knowledge base management involves organizing, managing, and maintaining knowledge. Based on actual usage, new knowledge is continuously added, useless knowledge is deleted, and erroneous knowledge is modified, gradually improving the quality of the knowledge base and the level of the system. This mainly includes the management and maintenance of knowledge, such as adding, deleting, editing, retrieving, and checking consistency and integrity, to maintain the consistency and integrity of the knowledge base. In practical applications, administrators input the domain knowledge to be added through the expert system's knowledge addition function, including fault phenomena, fault causes, and troubleshooting methods. The system then converts the input knowledge into the knowledge base's default knowledge rule format and performs verification. Knowledge verification mainly includes redundancy verification and contradiction verification. Redundancy verification checks whether the premises or conclusions of the newly input knowledge are the same as those of existing knowledge in the knowledge base. Contradiction verification checks whether there are cases where the premises of the input domain knowledge are the same as those of existing knowledge rules in the knowledge base, but the conclusions are different or opposite. After verification, correct knowledge is stored in the knowledge base; otherwise, the system displays an error dialog box. The specific implementation of database query and deletion functions will be described in detail in the system design. 4. Reasoning Process and Implementation There are three reasoning methods: forward reasoning, backward reasoning, and mixed reasoning. Forward reasoning is the most commonly used method, and this system uses forward reasoning. Forward reasoning, also known as data-driven strategy or forward inference, is based on the following idea: starting with existing facts about the problem, knowledge is used forward. When a match is found with existing facts, that knowledge is considered usable. Then, through conflict resolution, an enabling rule is selected from the available unmatched knowledge. The use of the enabling rule changes the context, leading to new rule matching. This process is repeated until a problem state has no usable knowledge or the required solution has been found. The reasoning process is shown in Figure 3: [align=center] Figure 3 Reasoning Flowchart[/align] The concept of an unmatched rule is: rule a is any rule in the rule base. If the observed fault symptom condition is contained in the condition part of rule a, then rule a is called an unmatched rule. The concept of a matching rule is: rule a is any rule in the rule base. If the observed fault symptom condition is completely equal to the content contained in the condition part of the rule, then rule a is called a matching rule. 5. Conclusion This paper mainly introduces the establishment of a knowledge base and the design of an inference engine for a fault diagnosis expert system for air-circulating refrigeration equipment. Based on an in-depth study of the system structure and working principle of air-circulating refrigeration equipment, fault trees were introduced into the knowledge base structure, and a fully functional fault diagnosis expert system for aircraft ground air conditioning vehicles was successfully constructed. This expert system can partially replace expert guidance for on-site maintenance, which is of great significance for extending the life of air-circulating refrigeration equipment and aircraft, reducing maintenance and repair costs, and ensuring the reliability and safety of experiments. The innovation of this paper lies in constructing a fault diagnosis expert system for air-circulating refrigeration equipment specifically for its characteristics. Introducing fault tree analysis into the expert system effectively solves the bottleneck problem of knowledge acquisition, ensures the completeness of diagnostic knowledge, and also leverages the rapid and effective fault diagnosis characteristics of expert systems. The expert system built using fault trees has high reliability, comprehensive fault analysis, and fully utilizes the long-term maintenance experience of experts on air-circulating refrigeration equipment. In particular, the fault tree itself is a kind of visual technical data, which is a very effective intuitive teaching material and maintenance guide for maintenance personnel, and it is consistent with the thinking of human experts, making it easy to understand and master. References: [1] Li Jun. Design of intelligent fault diagnosis system for aircraft ground air conditioning vehicle [D]. [Master's Thesis] Dalian: Dalian University of Technology. 2005: 1-3 [2] Chen Wei, Wu Zhiliang. Structural study of fault diagnosis expert system combined with fault tree technology [J]. Navigation Technology, 2005 (6): 43-45 [3] Cai Zongping, Tang Zhengping, Min Haibo. Application of expert system of fault tree analysis in fault diagnosis [J]. Microcomputer Information, 2006, 8-1: 135-137. [4] Wang Dongmei, Wang Li, Zhang Tao. Design of railway locomotive fault diagnosis expert system [J]. Microcomputer Information, 2006, 10-1: 221-223. [5] Jin Xing, Hong Yanji. Intelligent fault diagnosis system based on fault tree [J]. Journal of Astronautics, 2001, 22-3: 111-113. [6] Chen Chaoyang, Zhang Daisheng, Ren Peihong, Xu Huadong. Automobile fault diagnosis expert system based on fault tree analysis [J]. Transactions of the Chinese Society for Agricultural Machinery, 2003, 34-5: 131-133.