Modeling And Fault Diagnosis Methods Of Hydraulic AGC System In Cold Rolling Mill

Gauge control is a pivotal factor which determines the quality of strip steel products. Hence, the hydraulic automatic gauge control(AGC) system of cold rolling mill becomes an important part of the rolling process. The AGC system has multiple closed-loop structure which makes states with insignificant abnormal symptoms propagate and diffuse easily to induce intractable multiple faults, thus bringing about severe device failure. In addition, the system is affected by exoteric disturbances and uncertain factors, leading to the occurrence of hidden faults. All the faults might destroy the normal condition of the rolling mill, even resulting in device hidden danger such as rolling vibration and fearful quality accident such as strip steel sliding and off tracking, strip breakage and steel pile-up. To improve reliability and security of the AGC system, compound fault diagnosis is the imperious demand for rolling process, and fault diagnosis under the circumstance of uncertain information is a difficult problem ugently to be solved. Aiming at those issues, the main researches conducted in this thesis and the achieved contributions are summarized as below.1. The research status of problems about fault diagnosis and uncertainty has been summarized. In allusion to the fault diagnosis problem and uncertainty problem in AGC system, fault diagnosis methods of hydraulic AGC system, compound fault diagnosis methods, research methods of uncertainty in AGC system and fault diagnosis methods of uncertain systems are summarized respectively, which lay a foundation for proposing and researching the fault diagnosis methods in the thesis.2. The dynamic modeling and simulation of hydraulic AGC system in cold rolling mill have been carried out. The analysis of the regulation principle of hydraulic AGC system as well as its equivalent model is provided. The simulation test validates that the exit-gauge tallies with the roller gap setting on both conditions of stable rolling and unstable one, demonstating the favorable performance of the constructed model.3. An improved multi-state based multi-signal model method has been proposed and the verification of the method in AGC system has been implemented. The fault diagnois strategy of the multi-signal model is described and the improvement to the multi-signal model is conducted. Moreover, fault simulation and test point optimal design are executed to make preparations for fault diagnosis. Finally, the diagnosis for incipent faults and big faults are developed respectively, testifying the effectiveness of the multi-state based multi-signal model method.4. A compound fault diagnosis method based on multi-signal model and blind source separation has been proposed and the verification of the method in AGC system has been successfully implemented. The principle and application procedure of the proposed compound fault diagnosis method are focused on. Furthermore, the diagnosticability analysis as well as its procedure is elaborated. Finally, the diagnosis result in AGC system shows that the proposed method could separate the fault ambiguity groups effectively, improving the isolation rate of compound faults.5. The research of multi-signal model in uncertain system as well as its application in AGC system has been explored. The probabilistic reasoning method based on probability theory, the uncertainty reasoning method based on D-S evidence theory, the uncertainty reasoning method based on fuzzy theory, and the rule extraction method based on rough set are discussed to improve the multi-signal model in uncertain system to acquire more accurate dependency matrix, which could deal with uncertainty problems of the evidence uncertainty, unknown prior probability, test uncertainty and incomplete model respectively.