Development and Comparative Study of Neuronal Control with Remote Monitoring for a Level Plant

Industrial automation through artificial intelligence-based control is not yet applied with the desired frequency, primarily due to a lack of awareness of its benefits. Within the industry, the implementation of neural control and remote monitoring in the manufacturing process has demonstrated significant progress, thanks to optimized response times and the greater precision that characterizes these systems. This article focuses on innovative research in the field of tank filling, addressing challenges related to liquid storage, loading, and dispensing, while comparing the performance between a fuzzy system (adapted to the operator and plant’s needs) and a neuro-fuzzy system capable of learning from its predecessor’s mistakes. The plant with a fuzzy control system performs well when emptying and filling the tank; however, the sensors and actuators meet the operator’s needs but are not always accurate, and the execution time is not optimal, often varying. The study introduces a pioneering solution by applying an artificial intelligence approach that combines a neuro-fuzzy control system with real-time monitoring capabilities. The system optimizes resource utilization, achieving up to a 32.9% reduction in electrical energy consumption, and provides precise performance estimates under efficient control, with an error rate as low as 1%. The method includes an automation software tool (TIA PORTAL) and the implementation of sensors and actuators. Additionally, the programming platform (MATLAB) was crucial to harness the power of supervised learning in artificial intelligence, enabling optimal control of timing, stabilization, and result prediction. The neural network training is based on data obtained from the fuzzy control method. Neuro-fuzzy control systems with remote monitoring capabilities have proven to be highly effective in optimizing the tank filling process, achieving predefined set points with greater precision and resource efficiency. The results indicate that this research is of significant interest and carries crucial implications for defining further research directions and encouraging others to experiment with artificial intelligence-based models in other processes.