The Complete Characteristics Predictions Of Pumps And Softwere Development Based On Improved GARBFNN

For reliable and efficient pump function, it forecast its performance in a lot of research work, but because many factors affect the pump performance, making the relationship between performance parameters showed a typical non-linear relationship, it is difficult to use traditional methods to establish a unified mathematical model to be expressed. The pump characteristics prediction is such that using the geometry parameters of pump's components such as impeller, volute, diffuser and the pattern of pump interior flow to evaluate pump's performance. It can not only shorten the cycle of product research and development but also reduce the cost of the pump design and manufacture. The pump characteristics prediction has been a very important subject in pump field all the time, many experts and researchers have done a lot of study on the subject. But the result is still not accurate enough to satisfy the need of engineerings application, and there is still no practical software of the pump characteristics prediction in China. Therefore, it is necessary to have a serious study on the characteristics prediction of water pumps, which is both of academic meaning and social economy benefits. The main research contents of this paper are below:(1) The present situations and trend on characteristics prediction of water pumps are summarized. discuss the traditional way of pump performance prediction of the advantages and disadvantages, points out the pump comprehensive new approach to predict the performance.(2) In this paper, the characteristics of various pumps, pump parameters taken into account the distribution of the characteristics of the relationship between parameters and other factors related to the genetic algorithm neural network theory, based on radial basis function,to realize the pump full-featured forecast.(3) Parameters of a variety of water pump has been simulated for accurate computation and predicting its performance by using the mathematical analysis software-MATLAB,based on the ancient analysis of pump performance predictions.In order to predict the global performance of the water pump,RBF neural networks is selected,which is popularly used in academic and industrial applications,as the research platform.Finally, this paper summarizes the work done, and prospect the future content to be done.