Fault diagnosis method of large-scale complex electromechanical system based on extension neural network

Abstract

Owing to the complex fault information features in the operation of complex large-scale electromechanical systems, the existing semi-artificial state detection method is unable to obtain fast, accurate, or reliable diagnosis results, which causes significant loss to engineering construction projects. To solve this problem, a fault diagnosis method for a shield machine system based on an extension neural network is proposed. Using a matter-element model to describe the fault characteristic information, the upper and lower bounds of the classical domain of the characteristic index are used as the double weights of the neural network. Then, a fault diagnosis is made using the extension distance as the measuring tool, the algorithm structure and flow are constructed, the steps of the algorithm are given, and the effectiveness of the method is verified using tunnel boring machine as an example. The simulation results show that this method simplifies the network structure, improves the convergence speed, and obtains consistent diagnostic results. This research can provide a new method and idea for solving the problems of time, and manpower consumption, as well as the cost, of a fault diagnosis of a semi-automatic shield machine system.