| 1. Significance of StudyThe fuel injection system is part of the heart of diesel engine, whose technical performance has the direct influence to the work process of diesel engine. In order to make the fuel injection system of diesel engine always maintain good technical performance, enhance economic efficiency and social benefits, according to the situation of the system we repair the diesel engine. According results tested by advanced fault diagnosis technology, constitute repairing plans. This repairing methods can avoid the shortcomings of regular maintenance, namely when in need of repair, failure to repair, which could not play performance and easily caused the accident. When does not require repair ahead of repairing, resulting in unnecessary waste of human and material resources, and even bring the loss caused by mistaken handle. Repair depending on situation of the system that can improve repairing efficiency, reduce repairing time and reduce repairing cost. To achieve the repair depending on the situation, there must be advanced fault diagnosis equipment. By analyzing the status quo of diesel engine fault diagnosis both at home and abroad, the expert system of fault diagnosis for diesel fuel injection system based on ANFIS is researched and designed.2. Universal Design of SystemThe expert system of fault diagnosis for the fuel injection system mainly included Inference Engine, Knowledge Database, Database, Knowledge Acquisition System, Explaining System, Data Acquisition Disposal System, Man-machine Interface etc. The inference engine included fuzzy recognition module and ANFIS module. ANFIS (Adaptive-Network-based Fuzzy Inference System) is named adaptive network based on fuzzy inference system. ANFIS is the result of combining neural network and fuzzy inference, a multi-input and single-output structure, it can be divided into five: fuzzy segmentation layer, rules inference layer, fuzzy layer, betake fuzzy layer and output layer. ANFIS is applied in the experts system, which is a bright spot of intelligent diagnostic areas. ANFIS has the high diagnostic accuracy and the strong ability of learning. Knowledge Database is used to store knowledge of certain fields and the experience of experts, it provide the necessary knowledge for the inference engine. Database has been referred to as working memory or dynamic database, it can be said that the temporary data storage places of Inference Engine. Knowledge Acquisition System is a program, which deposited the knowledge in the field and experience of experts to Knowledge Database by the coding. Interpretation System can interpret for user, but also reflect wrong of the system, providing a convenient for maintenance of the system. Data Acquisition Disposal System is mainly responsible for collecting fuel pressure signals, corresponding signal processing, feature extraction parameters, and store the characteristic parameters in the Database. Man-machine Interface consist the application of procedures and the corresponding hardware, carrying out input and output. The figure 1 shows the total structure of the System. 3. Process of Fault DiagnosisThe characteristic of fault diagnosis process is that the diesel fuel injection system is separately diagnosed by fuzzy recognition module and ANFIS module. Data Acquisition Disposal System dispose the signal collected by the sensor and gain two characteristic parameters, which are the geometry character parameter and the function character parameter. When two characteristic parameters enter into the Database, the fuzzy recognition module read the appropriate rules from the Knowledge Database to match the geometry parameter, which is the initial diagnosis of the cause of the malfunction. Afterward ANFIS model read corresponding rule in accordance with the conclusions of the preliminary diagnosis from the knowledge Database to train ANFIS model. Inputting the function characteristic parameters into ANFIS model trained to validation the conclusion of initial diagnosis. 0.5 is the bound point of output numerical value, when more than 0.5, the initial diagnosis conclusions are correct, when less than 0.5, the initial diagnosis conclusions are wrong. Inference Engine is mainly responsible for control the process of the whole solved problem. Then apply appropriate control strategy to infer conclusions, which is complex. Therefore, the inference engine is the core of the design work about the expert system. In the process of ANFIS reasoning, which access to the new rules by self-learning, and the new rules are deposited into knowledgebase.4. Platform of System DesignWindows XP is the software of operating system in the computer, MATLAB is a programming language, which is the application of object-oriented programming method, the fault diagnosis expert system be developed. MATLAB has a friendly interface and a very rapid way for application development in the Windows XP environment. There are four modules (Fault Diagnosis, Knowledge Database Management, Fault Advice and Assistance etc). In addition, MATLAB toolbox has a GUI (graphical user interface) with ANFIS (Adaptive Neuro-Fuzzy Inference System), which used to build Sugeno-ANFIS. 5. Test Result AnalysisThe Fault Diagnosis Expert System for Diesel Fuel Injection System Based on ANFIS is used to diagnosis the CY1105 diesel engine fuel injection system , test result show that the examination results and the actual situation is consistency. The system has two-step method of diagnosis, and diagnostic precision is very good. Making full use of the pressure information of high-pressure fuel tubing, from which collect two characteristic parameters. It can avoid the wrong diagnosis by the characteristics of a single parameter. The self-learning is characteristics of ANFIS, which enrich rules of the Knowledge Database, and improve the diagnostic capacity for complex issues. Experimental gripper sensor trapped injector fuel-pressure signal in the high-pressure tubing with convenient and quick. Along with the continuous deepening of study, the system will become more perfect and maturity. |
PDF Full Text Request