Research On Framework Design And Algorithm Of Distributed Elevator Fault Diagnosis System

According to the questions that the aging of elevators with wide distribution caused byfrequent failures, Analyzed the functional requirements of the distributed elevators faultdiagnosis system, Designed and elaborated the overall framework and the methods to designand implement for each functional module of the distributed elevator fault diagnosis systembased on intelligent fault diagnosis methods and GPRS network communication technology,emphasis on the Structure model of system, That the system structure of hypogyny terminaland the framework design for the remote fault diagnosis center(RFDC), hypogyny terminaladopted Embedded system with high performance-price ratio. It’s good to complete the dataacquisition and transmission of the elevator operating state by using a simplified sensorconfiguration and design optimization of elevators common faults reasoning strategy, Byanalyzing the elevator running state information to determine whether it is normal and tocomplete the identification of common faults,solving elevators fault trapped people.Then,emphasis on the key technologies of the distributed elevator fault diagnosis system to achieveand the solutions of key problems to achieve system in the process. Elaborated the remotedata transmission schemes based on GPRS DTU module; Introduced the establishment of WANconnection between hypogyny terminal and RFDC; Studied technical solutions andimplementation steps that RFDC received field data;Described the realization of diagnosticdata stored in the SQL2000; Designed the remote the host fault diagnosis software with VB;aiming at exceptional faults, The host intelligent fault diagnosis software integration withLS-SVM and Rand Forests by Matlab GUI. This system not only can deal with elevatorrunning monitoring and fault diagnosis well and improve the scientific decision-making, butalso provide technical support for elevator safety supervision model innovation.Aiming at the method for elevator fault diagnosis by neural network has lots ofdrawbacks, such as overfitting, slow convergence and requiring large amounts of data, In this paper, elevator fault diagnosis method based on data-driven is studied: First, in order toimprove the fault recognition rate and the degree of automation, fault classification modelsbased on LS-SVM optimized by IPSO and MPGA is proposed; Second, Aiming at theelevator fault sample are with noise and highly nonlinear, traditional method can’t reflect thisquality of non-linear; KPCA can be used for fault data dimension reduction and de-noising, RFowns excellent noise tolerance and high fault recognition rate, Fault diagnosis method based onKPCA-RF is proposed.To verify the feasibility of the two fault diagnosis methods by UCIdatasets, the above method is used for elevator exceptional fault diagnosis finally, Researchdemonstrates that these methods with excellent effectiveness and practicality are suitable foractual fault diagnosis.