Application Of Cuckoo Search Neural Network In The Pumping Units’ Fault Diagnosis

In the development of economy, petroleum is enjoying a rising position. As the most important part of the oil field, pumping units’ fault diagnosis is more and more significant. At present, the technology of fault diagnosis has tended to automatic and intelligent. Therefore,research on intelligent fault diagnosis method for pumping unit has become a problem of oilfield enterprises which needs to be settled urgently. Taking problem of the pumping units’ fault diagnosis, a new type optimization algorithm cuckoo search(CS) was improved, and was applied to neural network training, the optimized neural network was applied to pumping units’ diagnosis, the concrete content is as follows:First,the step size of algorithm was updated randomly which leads to the results of lack of adaptability. Therefore, a self-adaptive step was introduced in CS to cope with this problem.The step size updated adaptively depending on the variation tendency of fitness function value so as to improve the search ability of algorithm. The self-adaptive step of the SCS(self-adaptive cuckoo search) algorithm was introduced in detail. The convergence of the SCS algorithm was proved, through the benchmark function test, and compared with traditional CS algorithm performance.Second, SCS algorithm was applied to train BP(Back Propagation) neural network. The nest in the CS is weights and thresholds in the neural network;the fitness function of nest is system target error in conventional BP algorithm. Slow training speed and premature result were improved by the SCS algorithm which has strong search ability. Experiment simulation was realized with training sample, and was compared with the traditional BP algorithm.Finally, the trained neural network was used in pumping units’ fault diagnosis.Processing to the current data from pumping units and extracting the feature vector based on wavelet packet decomposition. The neural network based on SCS algorithm was built, the diagnosis was carried out on the test sample, and the diagnosis result was compared with BP method of diagnosis. The result shows that CS-BP can develop the accuracy of fault diagnosis,which can meet the needs of fault diagnosis.