Enhanced adaptive Smith predictor control for underground coal gasification with long time delays

Background Underground coal gasification (UCG) represents an in situ conversion technology used to convert solid coals into gaseous fuel. However, the reaction process during UCG exhibits significant time delays. Such a limitation will reduce the response capability of the UCG system and is prone to induce overshoot and oscillation, thereby increasing the system’s control difficulty and reducing its stability and dynamic performance.

Method To address these issues, this study proposed an enhanced Smith predictor control algorithm based on a genetic algorithm and established a semi-physical experimental platform integrating a physical distributed control system (DCS) and MATLAB for modeling. Using the platform, experiments on the enhanced Smith predictor control algorithm were carried out under time delays ranging from 500 s to 2000 s and CO₂ volume fractions varying between 25% and 40%.

Results The results indicate that the time delay emerged as a key factor influencing the UCG control performance. Compared to proportional-integral-derivative (PID) control and Smith predictor control, the proposed enhanced Smith predictor control algorithm performed the best in dealing with long time delays. Specifically, under PID control, the time required for CO₂ concentration to stabilize increased by 30.77 times in the case of a prolonged time delay of 1500 s. Under model match (τ=1 500 s), Smith predictor control reduced the time from approximately 400 min to 38 min compared to PID control, representing a 90.5% improvement. However, in the case of 40% model mismatch, Smith predictor control increased the time by 4.37 times, suggesting limited robustness. In contrast, under identical conditions of model mismatch, the enhanced Smith predictor control algorithm reduced the time by 36%–59% and demonstrated smoother responses and a lower percentage overshoot than the Smith predictor control.

Conclusions The results of this study demonstrate the effectiveness of the proposed enhanced Smith predictor control algorithm, which exhibits high adaptability to UCG systems with long time delays under model uncertainty. This study provides a theoretical and practical engineering reference for UCG with long time delays.