Research on fault diagnosis strategy for air conditioning system based on MSMF-PD model

The energy consumption of the air conditioning system accounts for approximately 30 % of the total energy consumption of a building. Its fault diagnosis is of great significance for energy conservation and emission reduction. This study proposes a fault diagnosis method for air conditioning systems based on the multi-source modulation signal feature fusion and Probability distribution (MSMF-PD) model. This method adopts the DPCA signal demodulation technology to extract the vibration modulation signal characteristics of the scroll compressor in the horizontal, vertical and axial directions. It enhances the fault feature characterization ability through multi-source feature fusion technology and realizes fault classification in combination with the Bayesian probability distribution model. The experiment verified four typical faults: condenser fan failure, refrigerant leakage, excessive refrigerant and main shaft wear. The results show that multi-source feature fusion significantly improves the fault identification ability. When the input frequency band is set to 150 Hz, the model accuracy rate reaches 98.75 %. Compared with the DC, FSCB and FSCC models, the diagnostic accuracy of the MSMF-PD model has increased by 28 %, 8 % and 23 % respectively, demonstrating excellent diagnostic performance and robustness, and providing an effective technical solution for the fault diagnosis of air conditioning systems.