Neural Network and D-S Evidence Theory Based Condition Monitoring and Fault Diagnosis of Drilling

Abstract

This paper focuses on drilling fault diagnosis with the technology of information fusion based on neural network and Dempster-Shafer evidence theory. Neural network is used to process the drilling engineering data monitored from drilling on-site. The primary diagnosis results of drilling faults can be obtained by comparing the outputs of the neural network. And also the outputs of neural network are utilized to construct a basic probability assignment function (mass function) to assign a value of mass function for each type of drilling faults. The final fault diagnosis results will be achieved by using Dempster-Shafer evidence theory on decision level through further reasoning primary diagnosis results of neural network. The proposed method can in time discover the engineering data whether abnormal so that can diagnose and classify them, and will improve the accuracy of the drilling fault diagnosis.