Research On Gas Sensor Fault Diagnosis Based On Rough Set And Evidence Theory

Gas sensor is the critical component of safety inspection system for coal mine. The performance is essential. Once the gas sensor false alarmed, it would cause huge economic losses and casualties, as a result, the study of fault diagnosis technology is of great realistic significance to it.This thesis is aimed at terminating above-mentioned problem, researches the fault diagnosis method of gas sensor based on rough set theory and D-S evidence theory. It researches the advantage and disadvantage of the rough set theory and D-S evidence theory, and makes the analysis to the feasibilities of these two theories. And in the paper, the methods used in gas sensors are conducted. First, it analyzes the output and the fault type of the gas sensors, and sets up a model of fault diagnosis. Second, it makes discrete process for the data, and creates the decision table. An algorithm based on the discerning matrix is used to reduce the decision table. Next, it makes the reduced data in decision table applied to the evidence theory, and synthesizes evidence by decision rules based on basic probability assignment. Finally, at the testing stage, the paper applies above-theories to diagnosing the fault. It realizes information fusion process of rough sets and evidence theory. The simulation results show that this method can make fault diagnosis of gas sensors effectively, and the speed and accuracy of fault diagnosis are improved effectively. So it has good prospects in the field of fault diagnosis.