Prediction of Output Force of Pneumatic System Using BP Neural Network

Abstract

The dynamics of complex pneumatic systems are highly non-linear. The traditional modeling method has many flaws, such as a large number of influencing parameters, difficulty solving expressions, and an easy accumulation of errors. The models obtained by the traditional method are often inaccurate and difficult to be adopted for accurate control. In this paper, the pneumatic system of a hybrid electric-pneumatic actuator was analyzed theoretically and tested experimentally. A BP neural network model for the pneumatic system inflation and deflation was developed by selecting appropriate state parameters, and a prediction algorithm was proposed to predict the output force of the system for a period of time during the process of inflation (deflation). The results showed that it was feasible to adopt this method to predict the output force of the pneumatic system containing the common switch valve. The prediction of output force variation is good within the range of experimental data. It provides an effective method for the dynamic modeling of complex pneumatic systems. The prediction results within the sample range were better than those outside the sample range. In the future, the training sample data can be increased to improve the prediction accuracy of the model.