Research On Control System And Fault Diagnosis Based On Data-driven For Belt Conveyor

As the advancement of science and technology, it has been a tendency for long-distance great-capacity belt conveyer driven by multi-motor. The key technologies of large belt conveyor are controlled soft start, multi-motor power balance and fault diagnosis. It is crucial to tackle the three problems for the safety-production in coal. The core of controlled soft start and power balance is the choice of driving device and control strategy. So, this paper applies ac frequency conversation timing system as actuating device to drive belt conveyor through contrastive analysis. On the one hand,soft-start wish the belt to move according to the “S”curve;On the other hand, the power balance is based on multi-motor speed synchronized. we should deal with the speed control problems of belt firstly in order to achieve soft start and power balance. Because ac asynchronous motor is a high-order, strong coupling, nonlinear multivariable system,moreover, it is very difficult to realize accurate identification of the dynamic model, the control effect of traditional control methods is not ideal. Model free adaptive control(MFAC)arithmetic is not dependent on the arith model of the controlled system. The advantages of MFAC are less adjustable parameters, good control effect. This paper researches the method and apply it in the control process of belt conveyor, the main studies are proposed:To solve the problem of the controlled soft start, first, the best start-up speed curve provided; then the controller of MFAC is designed to make the belt start smoothly according to the ideal starting curve.For the multi-motor power balance problem, this paper has done some research and analysis on the multiple solutions of power balance at home and abroad, the author proposes to conduct load distribution according to the torque.The author determines the master machine to use the vector control of speed closed loop and the slave machine to use the torque control under host speed cascade,master and slave machine is torque closed-loop control under the case of the given torque distribution.The main circuit adopts the speed controller based on MFAC to replace traditional controller to leave host motor to finish the vector control of speed closed loop.Because the belt conveyor has many kinds of sensors which collect large amountof information in the process of operation monitoring. We adopt data driven fault diagnosis method–Principle Component Analysis(PCA) to optimize the number of sensors based on the contribution of each sensor loacation to fault classification.Secondly, the Radial Basis Function Neural Network(RBFNN) used to classify and identify feature vectors. Last, utilizing the Dempster-Shafer(D-S) theory to fuse and deal to realize the fault isolation.