引力搜索算法优化ENN模型的天然气管道球阀冲蚀深度预测

Optimized ENN model based on gravity search algorithm for predicting erosion depth of ball valve used in natural gas pipeline

天然气管道球阀在长期使用中易出现冲蚀现象,准确预测球阀的冲蚀深度对于管道的安全运行具有重要的实际意义。针对传统Elman神经网络(Elman Neural Network, ENN)模型的不足,提出了一种基于引力搜索算法(Gravitational Search Algorithm, GSA)的优化Elman神经网络模型并预测了不同影响因素下球阀的冲蚀深度,探讨了种群规模和隐含层节点个数对优化模型预测精度的影响。结果表明:传统模型预测所得的平均相对误差和均方误差分别为14.382%和0.042 5,优化模型预测所得的平均相对误差和均方误差分别为3.850%和0.003 9,因此,优化模型的预测精度大幅度高于传统模型;随着隐含层节点个数的增加,优化模型的预测精度先升高后降低;种群规模越大并不意味着优化模型的预测精度越高,合理的种群规模可使优化模型达到较好的预测精度;当种群规模和隐含层节点个数不同时,优化模型的预测精度始终高于传统模型,因此所提优化模型具有可靠性,可用于天然气管道球阀冲蚀深度的预测。

Many factors affect the erosion depth of natural gas pipeline ball valves.To more accurately predict the erosion depth of ball valves,a prediction model was established using machine learning algorithms.Considering the shortcomings of the traditional Elman neural network model(easy to fall into local minima and weak generalization ability)and the many advantages of the gravity search algorithm during the prediction process,an optimized Elman Neural Network(ENN)model is established by introducing the gravity search algorithm and the erosion depths of natural gas pipeline ball valves are predicted.The differences in prediction results between the optimized model and the traditional model are compared and analyzed,and the influences of population size and the number of hidden layer nodes on the prediction accuracy of the optimized model are explored.The calculation results of the example show that the average relative error and mean square error predicted by the traditional model are 14.382%and 0.0425,respectively.The average relative error and mean square error predicted by the optimized model are 3.850%and 0.0039,respectively.Therefore,the prediction accuracy of the optimized model is significantly higher than that of the traditional model.In the case of a population size is 30 and the number of iterations is 50,as the number of hidden layer nodes increases,the prediction accuracy of the optimized model first increases and then decreases.When the number of hidden layer nodes is 10,the prediction accuracy of the optimized model is the highest.In the case of the hidden layer nodes is 10 and the number of iterations is 50,as the population size increases,the prediction accuracy of the optimized model shows a trend of first increasing,then decreasing,and then increasing again.When the population size is 30,the prediction accuracy of the optimized model is the highest.Therefore,larger population size and the number of hidden layer nodes do not necessarily mean higher prediction accuracy of the optimized model.Overall,when the population size and the number of hidden layer nodes are different,the prediction accuracy of the optimized model is always higher than that of the traditional model,and it has good reliability.