🤖 AI Summary
This work addresses the challenge that traditional PID tuning relies heavily on model identification and fails to capture the empirical expertise of engineers who iteratively adjust parameters based on observed system responses. To bridge this gap, the authors propose a novel tuning framework that integrates control-domain knowledge with the reasoning capabilities of large and small language models. The approach formalizes the engineer’s tuning process into an executable task by leveraging closed-loop response characteristics, diagnostic cues, tuning preferences, and IMC-based examples to guide parameter generation and refinement. Physical constraints and reinforcement learning are incorporated to enhance performance, with the method employing supervised fine-tuning (SFT) and a physics-informed group relative policy optimization (PI-GRPO). Evaluated on 200 FOPDT/SOPDT processes, the cloud-based large model achieves a success rate of 75–89%, while a locally deployed Qwen3-0.6B model, after optimization, attains a first-recommendation success rate of 94.0%.
📝 Abstract
PID tuning for chemical processes commonly relies on identified process models, whereas plant engineers often retune loops iteratively by observing responses, diagnosing deficiencies, adjusting gains, and validating the result. This work formalizes this engineer-like workflow in a language-model-assisted PID tuning framework applicable to both large and small language models (LLMs/SLMs). Hosted LLMs receive closed-loop response features, control-engineering diagnoses, tuning preferences, and internal model control (IMC)-based demonstrations to generate and iteratively correct PID gains under common acceptance criteria. For local deployment, Qwen3-0.6B is adapted through supervised fine-tuning (SFT) with simulation-verified IMC targets and physics-informed group relative policy optimization (PI-GRPO) with non-compensable stability and performance rewards. On 100 first-order plus dead time (FOPDT) and 100 second-order plus dead time (SOPDT) test cases, hosted LLMs (DeepSeek-V4-Flash and Qwen3.7-Plus) achieve final success rates of 75-89% and 77-79%, respectively. As for Qwen3-0.6B, supervised fine-tuning raises first-recommendation success to 86.5%, and PI-GRPO further increases it to 94.0%, primarily improving first-attempt reliability and stability margins.