Prediction-based rapid force control of a single-acting pneumatic cylinder under hysteresis nonlinearity

Hysteresis characteristics widely affects the performance and reliability of pneumatic systems across various industrial applications. Addressing this challenge can significantly enhance system efficiency and precision. This paper aims to develop a rapid and accurate method for controlling the actuating force of a Single-Acting Pneumatic Cylinder (SAPC), considering hysteresis characteristic. To achieve these objectives, a Neural-Network-Prediction-based Proportional-Integral-Differential (NNP-PID) control strategy is introduced for the rapid prediction and precise control of the actuating force. Control experiments were conducted to elucidate the rapid control mechanism of the proposed NNP-PID strategy and assess its performance. Experimental results indicate that the developed neural network prediction model operates with a computational cost of 1.22 ms on an 8-bit microcontroller, thus meeting real-time control requirements. Compared to a conventional Proportional-Integral-Differential (PID) controller, the NNP-PID controller reduced control overshoot, rise time, settling time, and steady-state error by approximately 17.5 %, 65.9 %, 19.8 %, and 46.4 %, respectively.