A Predication Research On The Complete Characteristics Of Pumps Based On Improved PSO Artificial Neural Network

Pumps have been widely used in various sectors of the national economy. With thedevelopment of national economy, the safety of pump and its transport systems becomeincreasingly demanding. However, the study of the pump, in particular the full characteristicparameter of the pump, is far from to meet the actual needs. The whole parameters of pumpcan represent various operating conditions of pump–turbine, and it is especially importantfor the calculation and analysis of transition of hydraulics of pumps and its transport system.While, the calculation of hydraulic transition process directly affects the design of thetransport system of pumps and its safe operation after being built. Although many scholars inhome and abroad have done a great deal of work in getting the data of the full characteristicparameter of the pump and forecasting, until now the measured data is a few, and the accuracyof forecasting work remains to be improved. As a result, looking for a better model to predictthe full characteristic parameter of the pump has significance in reality.The main contents and results of the study are as follows:1. This paper describes in detail the relevant theories of the pump, analyzes theapplications of the curve of the full characteristic parameter of the pump, and summarizesmethods of data acquisition and prediction of the full characteristic parameter of the pump.2. In this paper, the neurons model, learning algorithms, classification and other commonmodels of the neural network are introduced, and the characteristics of several neuralnetworks are analyzed and compared. In addition, PSO algorithm theory and its developmentare introduced either. PSO which uses adaptive inertia weight are constructed to optimizeRBF neural network forecasting model.3. On the platform of MATLAB R2012b, using neural network toolbox and a GUI toolkitwhich provided by MATLAB and based on the proposed prediction model, this paperdevelops the prediction software of the curve parameters of the full characteristics of thepumps. On the basis of the existing data, this paper tries to predict the full characteristicparameter of the unknown pump, adopts a reasonable method to evaluate the predicting results, and by comparing it with other methods the advantages of this paper’s method arereflected.4. The prediction software, when using in the actual project, can optimize the calculationof hydraulic transition process. This paper puts forward some reasonable protective measuresto solve the problems that may occur in the transport system.