The Mine Hoist Fault Diagnosis Based On FTA And SVM

In the production of mine, Mine hoist is the core device in mining equipmentsystem, mainly used for lifting materials, equipments and persons. The brakingcontrol system in mine hoist system mechanism play a crucial role to ensure the safetyin the operation of lifting equipment, production of mines and the staff’s life. It directlyinfluences the safety and reliability of the mining industry and economic benefits.Therefore, the research work in braking control system of mine hoist has the importanttheoretical and practical significance.At present, the theory and method in intelligent fault diagnosis has made a lot ofachievements. As the introduction of computing intelligence technology to the field offault diagnosis, the machine learning and pattern recognition problem had solvedeffectively. But now the fault diagnosis system has many problems, it can be roughlyto the completeness and complex degree of knowledge base; the poor Adaptive andlearning ability; the poor Robustness; lack of universal and low degree of promotion.Based on these problems, intelligent fault diagnosis technology needs to be furtherresearch to make its continued development; we can fusion multiple sensor andvarieties information of in diagnostic methods, integration multi-level diagnosis todecrease the deficiency of a single method.This paper propose a kind of algorithm to fault diagnosis on brake system of minehoist combined by fault tree analysis (FTA) and decision tree based on geneticalgorithm support vector machine (GADT-SVM). And set up a remote fault diagnosissystem on mine hoist by the virtual private network (VPN).In this paper, a fault tree was set up to brake system of mine hoist by using thequalitative analysis method of FTA. The fault tree is embodied to mine machine hoistsafety circuits designed by PLC program. When the fault signal is sent to the node inPLC program corresponded to the fault tree, the PLC control system will issue awarning signal, we can get the failure analysis report from the CPU gained by the fault diagnosis algorithm and thus for maintenance.In the face of fault diagnosis method on mine hoist, proposes a new algorithmmulti-SVM, genetic algorithm support vector machine (GADT-SVM). The decisiontree was structured by this algorithm according to separation degree of sample betweendifferent categories and genetic algorithm. The construction of the decision tree usingthe thought of hierarchy clustering, in this way, it can effectively avoid the not pointsarea such as "refused to classification area" and "overlap classification". Thisalgorithm was used to the fault-recognition simulation experiment0n brake system ofmine hoist. A higher fault recognition rate was gained and the feasibility of thealgorithm was verified in this experiment.In addition, A safe, reliable and efficient remote fault diagnosis system was set upby VPN technology in this paper. This system was tested on the operation platform ofmine hoist for TongChuan group. A VPN gateway routing of SinforM5100-AC was setas a server. When completing the setting of VPN network, connected it to thediagnosis computer. Each mining area was connected to the VPN network by fiber. Inthis way, the remote monitoring to the run of mine hoist was finished.