Study On Fault Diagnosis Technology For Hydraulic Section Of Rolling Mill Agc System

Along with the improvement of the automation level of rolling mill, fault diagnosis for hydraulic AGC systems becomes one of the key techniques of modem rolling mills. It is highly important to develop such techniques for improving the productivity and the maintenance level of plant. Based on the deep research of the development on fault diagnosis for hydraulic systems, the implement methods of fuzzy neural networks expert system for fault diagnosis of hydraulic AGC systems of strip rolling mills are studied thoroughly. The research is financed by National Ninth Five-Year Scientific and Technical Key Project and National PhD foundation. Based on studying the composing of the actual hydraulic AGC system, and analyzing the characteristic of process and the functions of maintenance, the practicable strategy called 揻ault diagnosis of system function-based?is proposed. From the view of function-based diagnosis, the features and detecting points of process condition signals that can be used to define the fault of hydraulic AGC system are researched in detail. The investigation results form the theoretic references for drawing out the condition signals of system, and will be helpful for the reconstructing of actual system. By analyzing the principle of hydraulic systems, the distribution parameter model of hydraulic AGC system is established, and the simulations of the dynamic properties of the system is carried out. The research is much important for obtaining the model-based knowledge of fault diagnosis expert system of hydraulic AGC system. With the study on the characteristics of system status message, a new concept is proposed, which is to pick-up the status features used for system fault diagnosis with the minixnwn signals collection. The signal processing methodology is also discussed, especially the most significant methods for fault diagnosis of hydraulic AGC system, such as AR(Auto Regressive) model for recognizing system parameters , wavelet transforming for detecting cliscontinuities and breakdown points , and wavelet transforming for signals de-noising. Based on the deep research on the techniques such as expert system, fuzzy reasoning and neural networks, the method for constructing fault diagnosis expert system is studied, and the cooperation reasoning model and the fuzzy reasoning based on multi dimension with multi-scale problems are given out. The multi-layers fuzzy neural netwotks expert system for fault diagnosis of hydraulic AGC system is set up by using the hierarchy ?11 ? Abstract decomposing strategy. Finally, combined with the technique of database, the application software of frizzy neural networks expert system for the fault diagnosis of hydraulic AGC system is developed independently with the Microsoft Visual C++6.O. The analysis and experimental results indicate that the established modularization and multi-layers fault diagnosis expert system for hydraulic AGC system is not only simple but also specific for fault description. When tested, expert system is able to diagnose the fault of hydraulic system in real time. Besides, the expert system is much flexible and applicable to the actual system. All of these properties make it available for further researching on the fault diagnosis techniques of hydraulic AGC system.