Analysis of Compressor Sound Propagation Characteristics Driven by Simulation and Actual Data

To collect high-quality acoustic signals for compressor, the sound propagation characteristics is analyzed by building simulation model and supplementing actual data. The simulation model is established by calculating the dynamic loads on the blades, the vibrational state of the compressor, and the sound propagation condition outside of the compressor. To address the discrepancy between simulated and actual condition, the actual data is employed to modifying the model based on the auxiliary classifier generative adversarial network (ACGAN), which can address the discrepancy between simulated and actual measured signals. Based on the high-quality generated signal, the sound propagation characteristics is analyzed for compressor, and it is significant for structural health monitoring of compressor. The efficacy of the proposed approach is validated through compressor experiments, which can achieve a high cosine similarity of over 0.6, underscoring the robustness and effectiveness of the proposed model and method.