Search

In view of unstable problem existed in fault diagnosis result for rolling bearing of shearer based on K-means clustering algorithm, a new fault diagnosis method of rolling bearing of shearer based on TDKM-RBF neural network was proposed. The method adopts Tree Distribution algorithm to determine initial clustering center of the K-means clustering algorithm, so as to eliminate volatility of K-means clustering results. The method uses K-means algorithm to determine the parameters of the RBF neural network, then the trained neural network was used for fault diagnosis. The simulation results show that the method has quick clustering process,higher steability, and obviously improves accuracy of fault diagnosis for rolling bearing of shearer.

In order to improve accuracy of fault diagnosis of mine hoist, a new fault diagnosis method of mine hoist based on optimizing fuzzy Petri networks was proposed. Firstly, momentum BP network was used to optimize weights, thresholds and reliability parameters of the fuzzy Petri networks, and then the optimal fuzzy Petri networks model was used for fault diagnosis of mine hoist. The test results show that the method has fast convergence rate of parameter and high accuracy of fault diagnosis of mine hoist, which has a certain practical value.

A new method of location and classification of power quality disturbance based on wavelet packet and PNN was proposed according to essential characteristics of transient power quality disturbance. The disturbance signals were sampled and decomposed by using wavelet packet to extract wavelet packet reconstructed coefficient and to locate signal saltation point, then the energy of each band was calculated and normalized, energy feature vectors were constructed as input sample of PNN for network training and testing, and finally classification of different disturbance signal was achieved. Matlab simulation results show that the method can quickly and accurately locate and classify disturbance signal．

To improve the ability of network cabinet to deal with early warning and failure, a microenvironment intelligent monitoring system of network cabinet was designed. The system installs embedded terminal in the cabinet, collects power and non-power information and transmits to Web server via Ethernet. When collecting information was abnormal or cabinet fault occurs, the terminal can alarm intelligently and send SMS tips to remind abnormal situation. The experimental results show that the system can quickly and accurately locate abnormal and failure, and its remote control function is stable and reliable.

In view of problem of low computation efficiency and poor generality of existing evaluation method of coal mine power grid, an improved BFS-the minimal path evaluation method was proposed. The method uses the breadth-first search method to quickly solve division optimization model of distribution network island, and then adopts the minimal path evaluation method to evaluate reliability of power supply according to the operation scope of island. The experimental results show that the method improves evaluation speed of weak links of power grid, and evaluation results are safe, reliable, accurate and effective.