Based on the cross-entropy loss function and stochastic gradient descent algorithm, a weight regression fault diagnosis model was established for seven common faults in a chiller. The weighted regression model was slightly more complex than the pure linear regression model; however, the fault diagnosis performance was clearly better, and the minimum performance was improved by 40.50% under different feature sets. When comparing the effects of feature sets from various sources in this model and introducing a new feature set, the accuracy reached 89.83%. Notably, the diagnostic accuracy for local faults exceeded 98%. The explicit model for chiller fault diagnosis is summarized, and by examining the parameter weights in the visual diagnosis model, it was determined that the oil supply pressure, oil supply temperature, and degree of subcooling were the most crucial parameters for diagnosing three types of system faults. Conversely, the refrigerant pressure in the condenser, temperature difference in the condenser, and water flow parameters between the evaporator and condenser were identified as the most important parameters for diagnosing four local faults.

The optimization of regenerator geometries to enhance the performance of active electrocaloric regenerators (AERs) has attracted significant attention. In this study, corrugated and tapered structures were applied to a parallel-plate AER, and the effects of electrical field parameters on device performance were compared. The results indicated that the tapered structure better balanced the flow resistance and heat transfer efficiency, thus achieving the best refrigeration performance under the same operating conditions, followed by the corrugated AER structure. Short or long device cycle periods resulted in poor refrigeration performance. The electrical field should be switched when the transferred heat reaches 63%-66% of its maximum value. For the same cycle period, each device exhibited an optimal polarization duration (0.2 s), during which the refrigeration capacities of the parallel-plate, corrugated, and tapered AERs were 4.16 W, 4.35 W, and 4.71 W, respectively, with corresponding coefficients of performance of 1.76, 2.04, and 3.17. As the electrical field intensity increases, the refrigeration capacity of the device increases exponentially. The greater the field intensity, the greater the improvement in the refrigeration capacity of the gradually shrinking AER. When the field intensity increased from 50 MV/m to 225 MV/m, the refrigeration capacity of the tapered AER increased from 0.68 W to 10.06 W, an improvement of approximately 13.79 times.

Refrigerant leakage is a frequent and costly fault that deteriorates the normal operation of a chiller; however, it is difficult to measure directly. This study proposes a data mining-and key-feature-based approach for the soft measurement of refrigerant leakage. Random forest importance ranking and distance correlation coefficients were used to select the characteristic features, and a support vector regression (SVR) soft measurement model was established to measure leakage quantitatively. The proposed model was validated through a leakage experiment conducted on a screw chiller with a rated cooling capacity of 1440 kW and a refrigerant charge of 330 kg. The results showed that the SVR soft measurement model established on the three selected key features achieved significantly improved performance. The model had a root mean square error (RMSE) of 0.844 kg and a mean absolute error (MAE) of 0.734 kg, outperforming the other three feature subsets.