APPLICATION OF HYBRID NEURAL – PID CONTROL SYSTEM TO CONTROL WATER LEVEL IN TANK USING PLC S7-400

Abstract

Stability is a critical requirement in industrial control systems. A major challenge lies in designing a highperformance controller capable of meeting the stringent accuracy demands of technological processes. This paper proposes the development of a hybrid Neural–PID controller, which integrates the stability advantages of the conventional PID controller with the adaptive learning capabilities of a neural network, to regulate the water level in a single-tank system. The neural network utilized is a three-layer feedforward architecture, trained using a supervised learning approach. Experimental results indicate that the hybrid Neural-PID controller outperforms the traditional PID controller, particularly under nonlinear operating conditions and scenarios involving significant system variations. During setpoint changes, the Neural-PID controller achieves approximately a 50% reduction in overshoot compared to the conventional PID controller. Moreover, the control error remains low at around 0.5%, and the settling time is significantly faster. Even under substantial system disturbances, the Neural-PID controller maintains effective and robust regulation. Furthermore, this study demonstrates that simple neural networks can be directly implemented on the S7-400 Programmable Logic Controller (PLC) using the Structured Control Language (SCL), thereby paving the way for new applications of artificial intelligence techniques in industrial automation systems.