1. Introduction With the increase in voltage levels and system capacity of power systems, the structure of power systems is becoming more complex, and the reliability requirements for the operation of power equipment are also higher. Under harsh environments and adverse operating conditions, the internal insulation materials of electrical equipment often age rapidly, causing sudden damage. Once insulation fails, the entire piece of equipment will be taken out of service, or even cause a large-scale power outage, which is a disaster in modern society. Therefore, insulation monitoring and fault diagnosis are of great significance for ensuring the normal operation of electrical equipment and are a routine task for power supply companies. Previously, the assessment of equipment insulation status was made by professionals through longitudinal and horizontal comparisons and analyses of test data, based on relevant regulations, guidelines, standards, and experience. Since insulation monitoring is a routine task, manual judgment is very labor-intensive and it is difficult to comprehensively examine historical data. Moreover, expert judgment is easily affected by geographical location, workload, external environment, and psychological state. Therefore, power supply companies have a very urgent need for electrical equipment insulation monitoring and fault diagnosis systems. With the development of computer and artificial intelligence technologies, it has become possible to analyze and monitor test data using databases and to use fuzzy logic for fault diagnosis. To meet future development needs, this paper develops an electrical equipment insulation monitoring and fault diagnosis system based on the Management Information System (MIS) project of the Fuzhou Power Supply Bureau. The system uses Windows NT as the network operating system and adopts a client/server architecture. The database is Oracle 7.3, and the client development tool is PowerBuilder 5.0. To ensure the system's openness, reusability, and maintainability, object-oriented design, analysis, and programming methods are used. Taking transformer fault diagnosis as an example, fuzzy theory is introduced based on the method of detecting dissolved gases in oil. This not only allows for a reasonable judgment of the nature of transformer faults but also provides suggestions and preventative measures. 2. Introduction of Fuzzy Theory Currently, various regulations used for electrical equipment fault diagnosis generally only provide a description of the judgment boundary, which is difficult to accurately reflect the objective laws between faults and their manifestations. This "description of the judgment boundary" is sometimes given in a definite numerical form, and sometimes described in highly ambiguous language. When using the characteristic gas method for transformer fault diagnosis, the gas content is described using vague terms such as "high," "relatively high," and "main component," which introduces uncertainty. When using the three-ratio method for judgment, due to the limited range of ratio codes, it is often impossible to judge the fault in actual work because the ratio range cannot be found. Fuzzy mathematics is precisely to quantitatively analyze human fuzzy thinking and fuzzy language, to find mathematical tools suitable for computers to imitate the human brain for fuzzy recognition and discrimination, and has the characteristics of multi-factor comprehensive analysis. This paper uses the threshold principle and the nearest principle to judge transformer overheating faults, discharge faults and core multi-point grounding faults. Its basic idea is: (1) Determine the fuzzy mode Let U be the set of all faults of all transformers, and u be any specific fault in the fault set. The fuzzy subset is determined by the mutual ratio of H2, CH4, C2H2, C2H4, C2H6, C1+C2 (total hydrocarbons) as parameters. (2) Construct the membership function of the fuzzy mode According to the statistical test results, the membership function of the fuzzy subset is approximately a normal distribution: μ(u) = e-k(u-a)2, where k and a can be obtained by analysis and experiment. After calculating the membership function value, the severity of each parameter belonging to the fault sample can be represented by numbers. (3) Fault identification and judgment 1) Judge overheating and discharge faults according to the threshold principle Let A1 be a transformer discharge fault and A2 be a transformer overheating fault, with membership functions μA1(u) and μA2(u) respectively. According to the statistical calculation results, a threshold λ∈[0,1] is specified. Then ① when max{μA1(u),μA2(u)}<λ, uA1 and uA2, it is considered that no fault has occurred; ② when min{μA1(u),μA2(u)}≥λ, u∈(A1⌒A2), it is considered that a discharge and overheating fault has occurred; ③ when λ is other values, u∈A1 or u∈A2, it is considered that a discharge or overheating fault has occurred. 2) Judging Transformer Core Multi-Point Grounding Faults Based on Proximity Principle: Let there be a state B with membership function μB(ui); a multi-point grounding fault A in the transformer core with membership function μA(ui). The probability of fault A existing in state B is described by proximity n(A,B). The larger n(A,B), the greater the probability of fault A existing in state B. When the proximity exceeds 0.5, a transformer core multi-point grounding fault is considered to exist; otherwise, the system is considered to be operating well. 3 System Analysis and Design 3.1 Network Architecture Design: To achieve rapid information transmission and high resource sharing, a monitoring information network must be established. The network structure is shown in Figure 1. The system operates on a wide area network consisting of the Fuzhou Power Bureau and the High Voltage Testing Team. They are directly connected by cables or via telephone lines through modems. The Oil Testing Team, Substation Department, and Production and Maintenance Department are located within the Power Bureau's local area network. The Production and Maintenance Department's internal Ethernet is based on a server architecture, while the High Voltage Testing Team is based on a client architecture, with each machine acting as both a server and a client. The Oil Testing Team and the Insulation Department are network clients. The system uses Windows NT Server 4.0 as the network operating system, TCP/IP as the network protocol, Oracle 7.3 as the network database, PowerBuilder as the client development tool, and Windows 95 as the operating system. Figure 1 shows the system network structure. 3.2 Operating Mode Design Considering the requirements of coverage, security, openness, and scalability, the traditional file management system development model is far from meeting the needs. Therefore, this system adopts a client/server model. In the electrical equipment fault diagnosis system based on the client/server architecture, all kinds of key data are uniformly stored on the server. The client only has general input, output, and fault diagnosis applications. Querying, browsing, and storing operations are all achieved by the client application sending requests to the server. In this structure, what is transmitted over the network is the client request and general results, rather than the entire data file as before, minimizing the amount of information transmitted over the network to improve system performance and data processing efficiency. This ensures that the system meets the needs of enterprise management information systems in terms of data sharing, network performance, security performance, data integrity and consistency, concurrent data operations, and data fault recovery. 3.3 Data Architecture Design The database design adopts a combination of centralized management and distributed application. The power bureau uses a centralized structure, with oil handling team data and test precautions stored on the production and maintenance department's server. The power bureau and high-voltage teams use a distributed structure, with local data accessed locally, requesting services from the corresponding nodes only when needed. This effectively reduces data transmission volume on the network and alleviates server pressure. Therefore, the database is divided into a production and maintenance department database and a high-voltage team database. Different users on the network have different access permissions to the data. Based on system functional requirements, the data in the system can be divided into the following categories according to usage: Type 1: Equipment nameplate data from the substation department, globally unified into one set of data, only the substation department has the right to modify, and grassroots test teams can only access it; Type 2: Electrical test data for various electrical equipment, dissolved gas analysis data in oil-filled equipment (including intermediate calculation results and diagnostic conclusions), and oil handling test data, the production and maintenance department can only query, not modify; the oil handling team has read and write rights to oil handling data, but only read rights to high-voltage team data, while the high-voltage team's data access permissions are the opposite; Type 3: Test precautions in the test procedures, insulation specialists can modify, and grassroots teams can only access them. For Type 1 data, database links and synonyms are established in the Production and Construction Department and High Voltage Testing Team databases to access the substation data. For Type 2 data, data tables are created in both the Production and Construction Department and High Voltage Testing Team databases, with the data stored at its origin. The Production and Construction Department database creates a database link and a snapshot of the data tables in the High Voltage Testing Team database. Insulation specialists query equipment and indicators with alarm status. For Type 3 data, the data is stored in the Production and Construction Department database, and the High Voltage Testing Team accesses the data by creating views. The specific relationships between tables, views, and snapshots are shown in Figure 2. Figure 2. Relationships between database objects in the system . 4. Software Functions and Implementation The main functions of the system include: calling equipment nameplate data from the substation server; completing the input, modification, query, and printing of electrical test data for various electrical equipment, dissolved gas analysis data in oil-filled equipment, and oil testing data; monitoring important test indicators and issuing alarms when they exceed warning values; graphical monitoring is also available; insulation specialists can query equipment with alarms by equipment and indicators; automatic diagnosis of transformer faults; and test data can be printed in different combinations. 4.1 Main Control Module The main control module, usually called the "main menu," is the main entry point to the system and the interface for human-machine interaction between the system and operators. When the system starts running, the main control module displays the various functions and prompts of the system to the user. Users simply need to follow the prompts in the menu and select the information they need step by step. The system implements password and access control for operations. Each operator has their own password, and unauthorized personnel are prohibited from using the system. Access control is implemented for operators. Operators are divided into three categories: system administrator, general operator, and insulation specialist. Because certain operations can only be performed by specific types of operators, the system, after identifying the operator's identity, will block some function menus that cannot be performed by that operator. The system administrator can modify the permissions of various operators. The main control module has six independent sub-modules, each of which can be installed as a separate system, facilitating the expansion of new functions. 4.2 High-Voltage Insulation Testing Subsystem This module includes test reports for 17 types of electrical equipment, such as transformers, instrument transformers, bushings, capacitors, and surge arresters. These reports are similar in format, consisting of different insulation preventative tests. Each test report has functions for inputting, modifying, querying, printing, and retrieving remote data. 4.3 Oil Testing Subsystem This module includes four test reports: insulating oil gas chromatography, insulating oil quality, insulating oil trace moisture, and SF6 gas humidity; and four analysis logs: gas chromatography, oil quality, trace moisture, and gas humidity. The four analysis logs mainly summarize the data in the test reports by equipment and sort them by time, which facilitates longitudinal historical data analysis and more intuitively reflects data change trends. The testing cycles for each piece of equipment are automatically divided according to the testing procedures and the previous test values, and the insulation changes of the equipment are tracked. If the test value differs significantly from the warning value with a large margin, a longer cycle is used. This is done to reduce the workload of management and testing personnel. 4.4 Warning Subsystem When the data from high-voltage tests and oil tests exceed the warning values ​​specified in the "Preventive Testing Procedures for Power Equipment," the background color of the data turns red to alert staff, shorten the testing cycle, and conduct follow-up analysis. In addition to the above methods of supervision, the development trend of each indicator can be monitored more intuitively using graphical methods, such as the curves of the change of gas content over time and the TD (overheating-discharge) graph. Insulation specialists can also query warnings in two ways: one is by equipment, which allows for any combination of substation, equipment type, and equipment name; the other is by indicator, such as viewing the warning status of H2, C1+C2 (total hydrocarbons), and dielectric loss factor. 4.5 When the fault diagnosis subsystem analyzes the chromatographic data of transformer oil and uses the chromatographic data for condition diagnosis, the system automatically retrieves the required data from the database. There are two possibilities: either two sets of data are retrieved from the database, or only one set is retrieved. The latter indicates that the database has only one record (e.g., a newly commissioned transformer, or a transformer undergoing its first oil chromatographic analysis). In this case, the diagnostic system is defined as being in a non-tracking state, while the former is defined as being in a tracking state. After acquiring the data, the system performs a series of calculations and diagnoses. If any gas content exceeds the warning value specified in the guidelines, it is treated differently depending on whether it is in a tracking state. In a non-tracking state, due to insufficient data, although fault diagnosis is performed, only a "possible conclusion" is given, and a prompt is made to conduct chromatographic tracking analysis in the near future. If the total hydrocarbon production rate also exceeds the warning value, two fuzzy mathematical methods are used to determine the nature of the fault and display the fault location, fault source temperature, and other conclusions. If the total hydrocarbon production rate does not exceed the warning value, a prompt is given that further tracking is needed in the near future. The diagnostic process is shown in Figure 3. Figure 3. Fault Diagnosis Flowchart. When the mouse hovers over certain controls in the window, a yellow-background, black-text tooltip will automatically appear near the control to indicate precautions. For example, for open transformers, when the CO content suddenly increases and exceeds 300 ppm, a warning about load and oil temperature will be given; when the CO/CO2 ratio exceeds the warning value, a corresponding warning will also be given, and so on. When both of the above situations exist simultaneously, a warning "Possible solid insulation fault" will appear. 4.6 Combined Printing Subsystem. The printing function in the high-voltage insulation test and oil test subsystems is for test reports that are being edited. In actual work, test personnel cannot retrieve and print one report at a time; they usually record data in large batches and then print them all at once. This module can perform batch printing according to different combinations of conditions. 4.7 System Maintenance Subsystem. This module includes high-voltage test personnel maintenance, oil test personnel maintenance, warning value maintenance, and user permission management. Different operators have different passwords and permissions to ensure system security. 5 System Features This system serves as an effective tool for power bureaus and grassroots work teams to preserve historical data on electrical equipment and evaluate equipment performance. It features the following characteristics: (1) Monitoring information network to achieve timely information transmission and high resource sharing. (2) Introducing fuzzy mathematics into fault diagnosis to diagnose the nature of transformer faults and provide corresponding preventative measures and auxiliary conclusions such as fault source location and fault point temperature. (3) Employing multiple methods to monitor insulation. For example, warnings are issued when important test indicators exceed warning values, and gas change trends are monitored graphically, with queries available for warning information. (4) The entire system interface is clear and aesthetically pleasing, and operation is simple. Users do not need extensive computer knowledge; they can complete all required tasks by following the on-screen prompts. It provides both keyboard and mouse operation, minimizing the amount of Chinese character input. (5) Multiple security measures are implemented. Whenever a page or window changes, a dialog box prompts for saving, preventing data loss. The system has strong confidentiality and good security. 6. Case Study A 110 kV SFSB6-31500/110 transformer at a substation was put into operation in November 1996. One month after commissioning, chromatographic analysis revealed a total hydrocarbon level of 184 ppm (exceeding the 150 ppm limit specified in the Ministry-issued guidelines), and further monitoring was conducted. On August 19, 1997, the total hydrocarbon level reached a staggering 604 ppm, and acetylene was 10 ppm (exceeding the warning limit of 5 ppm). Specific test data are shown in Table 1. Table 1: Chromatographic Analysis Data Before and After the Fault (Unit: ppm) To investigate the cause and eliminate potential hazards, the transformer was shut down on October 14, 1997. Testing showed that multiple electrical tests were normal. However, during a core inspection, after disconnecting the upper grounding bushing connection, the core still showed grounding. A capacitor discharge method was then used to test the grounding, revealing an iron wire and a large amount of metal shavings between the lower yoke and the bottom of the transformer, causing multiple grounding points and overheating. After treatment, the transformer was put back into operation, and subsequent chromatographic analyses were normal. The system was used to analyze the chromatographic data from "1997-08-19". The diagnostic results (see Figure 4) were consistent with the experimental results, verifying the accuracy and practicality of the system. Figure 4: Fault Diagnosis Window of the System . Conclusion This system not only transmits data promptly and quickly, achieving data sharing and eliminating data redundancy, but also improves data accuracy, ensuring data consistency and security. It changes the original management model, greatly reducing the workload of testing personnel and insulation specialists, improving office efficiency and work quality. Through the comprehensive analysis of the alarm system and fault diagnosis system, it provides a basis for testing personnel to take corresponding measures, ensuring the safe operation of electrical equipment. Especially in fault diagnosis, the timely and accurate analysis and diagnosis of latent faults in transformers of several units has brought certain economic and social benefits to the enterprises, while also verifying the accuracy and practicality of the system. The application of this system makes insulation monitoring and management more scientific and modern.