A Phased Hybrid Optimization Method for Temperature Compensation in Ultrasonic Flowmeter Measurement

The accuracy of ultrasonic flowmeters is significantly affected by temperature variations in complex operating environments, leading to measurement deviations and limiting their practical performance. Traditional compensation approaches often struggle to effectively handle the nonlinear and dynamic effects of temperature variations, leading to suboptimal performance under varying flow conditions. Thus, a temperature compensation method based on a Phased Hybrid Optimization Method (PHOM) is proposed in this paper, which integrates Support Vector Regression (SVR) with metaheuristic optimization algorithms. The PHOM employs a two-stage optimization strategy using Grey Wolf Optimizer (GWO) for global search and Crow Search Algorithm (CSA) for local fine-tuning of SVR hyperparameters. Additionally, a sliding window mechanism is designed to identify flow regions dynamically, enabling adaptive compensation across flow ranges. An online incremental learning mechanism is also introduced to enhance long-term adaptability. Simulation experimental validation demonstrates that the proposed method improves compensation performance and achieves higher robustness and accuracy than traditional BP neural network methods.