Adaptive learning controlling algorithm for A hydraulic servosystem

A hydraulic servosystem is commonly used in heavy industries in which high power is required. The most widely utilized valve is the solenoid ON/OFF valve because of its simplicity and low cost. However, it has poor controlling performance, which increases the nonlinear behavior occurring in hydraulic systems. This paper devotes on the development and implementation of an adaptive learning algorithm as a real-time controlling algorithm for a hydraulic servosystem utilizing a solenoid valve. The proposed adaptive learning algorithm is a special Artificial Neural Network (ANN) model. ANN is a computer program that provides human brain learning capability to computers for a specific task. In this approach, ANN predicts the controlling action for certain operating parameters learned through a training process. The operating parameters and the result controlling action datasets are collected from experimental operations and then provided to the ANN to learn during the training process. The real-time operating results of the hydraulic servosystem using the adaptive learning algorithm show a compliance of the actuator, which is a linear hydraulic cylinder, to the desired displacements. The controlling through learning concept has the potential to overcome the drawbacks of traditional controllers through the ability to adapt to any changes in the dynamic behavior of a hydraulic servosystem.