Papers by Author: Jia Tang Cheng

Abstract: In order to improve the accuracy of fault diagnosis of power transformer, in this paper, a method is proposed that optimize the weight of BP neural network by adaptive mutation particle swarm optimization (AMPSO). According to the characteristic of transformer fault, the optimized neural network is used to diagnose fault of the power transformer. Individual particles action is amended by this algorithm and local minima problems of the standard PSO and BP network are overcooked. The experimental results show that, the method can classify transformer faults, and effectively improve the fault recognition rate.

Abstract: In order to improve the prediction accuracy of gas emission, propose a prediction method of evidence theory combining with neural network. According to the experimental data of gas emission, three different particle swarm optimization-neural network models are used for the initial prediction. And use the BP and RBF network to get the credibility of model by analyzing forecasting errors and its influence factors. Then the evidence theory is used to obtain the weights of combination model, realize the gas emission combination forecasting. Examples results show that the error of evidence theory is less than error of the neural network and equal weight method, and it is suitable for gas emission prediction of coal mine.

Abstract: The equivalent salt deposit density (ESDD) of insulator in power system is the main basis of defining pollution classes and mapping pollution areas. However, The meteorological factors on it is complex, using traditional methods is difficult to establish accurate mathematical model to express the relationship, In this paper, the gray theory and neural network model to reflect the changing trend of data series on the apparent effect, Gray neural network model used to predict the insulators ESDD under certain meteorological factors, and to design a neural network compensator correction on the predicted results. The simulation results show that the model has higher prediction accuracy, better than a simple gray neural network model, and have certain theoretical value and practical application value.

Abstract: In order to improve the accuracy of fault diagnosis of asynchronous motor, neural network model combined with ant colony algorithm is presented. Taken the mean square error as objective function, then the weights and threshold values are optimized through multiple generation computation of ant colony, and the fault diagnosis is accomplished via the optimized neural network. The simulation results show that the algorithm is to overcome the slow convergence, easy to fall into local minimum problem of BP network, and achieve good diagnosis.

Abstract: In order to improve the prediction accuracy and prediction speed of coal mine gas emission, ant colony algorithm combining with neural network is used for prediction models design. Choose an important factor influencing gas emission, establish of its neural network prediction model. Select the network mean square error as the objective function, through the ant colony algorithm iteration achieve optimal BP network weights, and use the optimized BP network for gas emission prediction. Simulation results show that the method has high fitting prediction accuracy.

Abstract: Aiming at the problems Expert PID parameter tuning for time-consuming, and the results are not necessarily the best. In this paper, genetic algorithm is introduced to the parameter optimization, finally get a set of optimal PID parameter values. In comparison with simulated experiments, the results show that the performance of the Designed to optimize the performance of optimization expert PID controller is better than conventional controller, can achieve good dynamic performance.