R/S analysis method is used to analyze gas well production,and two different expressions of historical data time series and stationary time series are given.Then the output of gas well is predicted by using BP network method.In order to verify the effectiveness of this method,BP network is compared with BP neural network and Bayesian algorithm.The results show that the prediction result obtained by Bayes algorithm is more accurate and realistic than the other two methods.For the prediction results of the three methods,the average error between the prediction results and the actual value using the Bayesian algorithm is 0.67%and 0.33%respectively,while the average error between the prediction results and the actual value using the traditional BP neural network is 1.71%and 0.96%respectively.
R/S analysisBP network methodgas well productionBayesian algorithm
NETL
NSTL
万方数据
