Event-triggered fuzzy neural-network PID control for nonlinear gas-blending processes

To address low control accuracy and frequent controller updates in gas blending, a fuzzy neural network PID control method based on an event-triggered mechanism (ET-FNN-PID) is proposed. Key operational variables correlated with gas concentration are selected as model inputs based on real conditions and expert knowledge. A data-driven model is then constructed using a Takagi-Sugeno (TS) fuzzy neural network. The FNN-PID controller dynamically adjusts PID parameters through the fuzzy neural network, with online parameter updates via gradient descent. An event-triggered condition using a fixed threshold is introduced to reduce unnecessary controller updates and mechanical wear. Simulation results using real gas data show that the ET-FNN-PID controller effectively captures nonlinear system behavior and achieves precise gas concentration control while significantly reducing update frequency compared to traditional time-triggered and standard FNN-PID controllers. This approach enhances control performance and contributes to energy efficiency and emission reduction.