Research strategy of fault diagnosis of air conditioning chiller based on machine learning adaptive ELM

Faults in chillers in HVAC systems often lead to decreased cooling efficiency, increased operating noise, and higher building energy consumption. Accurate diagnosis of these faults is essential for identifying the fault location and guiding repair work, thereby extending the lifespan of air-conditioning equipment. In recent years, data-driven fault diagnosis methods for air-conditioning chillers have rapidly developed. However, these methods commonly face challenges such as slow model training speed and high reliance on model parameter tuning for diagnostic performance. This paper proposes a fault diagnosis method for air conditioning chillers using adaptive parameter optimization extreme learning machine (ELM). This method employs a particle swarm optimization algorithm to optimize the hyperparameters of the ELM model, enabling adaptive parameter optimization and enhancing the model’s automatic feature extraction capability. As a result, it improves the model’s generalization ability and robustness. Additionally, the impact of different activation functions on the performance of the proposed fault diagnosis model is discussed. The effectiveness of the proposed algorithm is tested on the ASHRAE-1043 project data, and the results demonstrate an accuracy exceeding 98%. Furthermore, the proposed method has been proven to have advantages in terms of performance and robustness compared to various other contrasting method.