The Equipment's Remaining Life Prediction Based On GMDH

Mechanical Equipments are the core instruments and the main resources for enterprises producing. Many industries, such as manufacturing industry, electricity industry, energy industry and so on, all have their own mechanical equipments. The running status of mechanical equipments is extremely important for the development of enterprises. If these mechanical equipments occur mistakes or malfunctions, which will not only make production delaying, but also result in the production costs increased because of these equipment maintaining expenses, and which directly weaken the competitiveness of enterprise. Therefore, enterprises should be able to troubleshoot mechanical equipment effectively in advance to enhance their competitiveness. Troubleshooting mechanical equipment not only include monitoring and diagnosing equipment running states, but also forecasting the development trend of their future running condition. In the premise that mechanical equipments is on the normal, stable working conditions, according to the running states of mechanical equipments, how to predict the service time of mechanical equipments, and as much as possible to extend their service time ?Many industries have paid attention to solve above problems, and naturally these problems also become hot research topics of information industry.This paper focuses on the issue of forecasting the remaining life of mechanical equipments in the troubleshooting mechanical equipment. Firstly, this paper introduces several common methods for predicting the service time of mechanical equipments, and then analyzes and compares these methods, according to synthesize the advantages of these methods, a new method, which is different from traditional methods, is proposed to forecast the remaining life of mechanical equipments, which is based on nonlinear complex systems and the GMDH theory. This method achieves the automated model creating and validating processes, and based on the phase space reconstruction theory, it reconstruct the prime driving power system, which effectively resolve the problem that can't construct the system model caused by not enough state parameters in the course of the remaining life prediction using GMDH methods. Finally, through practical examples of an enterprise project, we verify the effectiveness of the methods.