Health status evaluation of gas driven compressor units based on GA-BP

A gas turbine health status evaluation method based on genetic algorithm optimization of BP neural network parameters is proposed to address the instability and singularity of data obtained from sensors in natural gas station combustion driven compressor units, and the low prediction accuracy caused by weak algorithm feature extraction ability. Based on the machine structure and working principle of gas turbines, a health indicator system is constructed and evaluated. Through genetic algorithm selection, crossover, and mutation operations, a new BP neural network parameter combination is generated and its performance is evaluated. Input the preprocessed gas turbine dataset into the BP neural network model, and use the BP algorithm to train the neural network corresponding to each parameter combination. Gradually adjust the parameters of the neural network to better fit the training data and complete state evaluation. According to the experimental results, it can be sure that the GA-BP model has the highest accuracy in health assessment of combustion driven compressor units, reaching 93%, which can effectively solve the problems of dataset instability and difficult health status assessment.