Research On Energy Saving Operation And Leakage Pattern Recognition Of Heat Pump System Based On Data Mining

Energy-saving and efficient operation of HVAC equipment is a long-term research goal.The heat pump system converts low-grade heat sources into high-grade heat energy,and is widely used in engineering practice.From design and production to long-term operation,a large amount of data will be generated in each stage.This paper applies data mining to the large amount of data accumulated by the heat pump system on the heat pump drying and heat pump water heater systems,extracts the energy-saving and efficient operating state rules for the heat pump drying system,and analyzes the best operation program under the premise of meeting the drying index;for the heat pump water heater system,The characteristics of fault characterization and the realization of fault mode recognition of heat pump air conditioning equipment have theoretical and practical significance for the efficient and safe operation of heat pump systems.First,based on the heat pump system in the heat pump drying and heat pump water heater system fault diagnosis background,from the unsupervised and supervised data mining algorithms to elaborate and establish a data mining-based heat pump system energy-saving operation and fault diagnosis and identification of the overall framework.Secondly,an unsupervised mining research process of "pretreatment-association rule mining-energy-saving strategy analysis" was established for the factors affecting the drying performance of dishwasher heat pump drying system and energy-saving operation optimization analysis,and the overall drying performance and various types of tableware were screened out The correlation rules between the performance of tableware and comparative analysis from the perspective of numerical data verify the interpretability of the correlation rules between the drying indexes;the state parameters and drying performance of the heat pump drying system are analyzed between the drying index and various influencing factors The correlation between the three types of external influence factors,such as charge volume,ambient temperature,and air supply method,and the drying index correlation rules were extracted,and the influence of the three external factors on the drying performance was analyzed.Analyze the correlation rules between the heating performance and drying performance of the heat pump system,combine various influencing factors with the drying performance,and summarize the energy-saving operation strategy of the dishwasher heat pump drying system Analysis of the relationship between drying performance and heat pump energy saving by two major influencing factors of heat,and a complete energy-saving operation strategy is obtained.Finally,for the diagnosis and identification of refrigerant leakage and other faults of the heat pump system,a "feature extraction-fault diagnosis and identification" based on supervised data mining process was established,and the actual refrigerant leakage and other faults of the simulated heat pump system were collected through experiments Data,the original data is preprocessed to obtain a data set containing 41 original data parameter features.Using the Relief F feature selection method and PCA method to extract the features of leakage and other faults,the 10-dimensional new parameter features with the highest correlation weight of leakage faults are screened out.The PCA feature extraction method uses the spatial linear transformation of the original 41-dimensional parameter features to obtain Characterization of leakage faults or other faults.Finally,the PCA-SVM model for leakage and other fault diagnosis and identification was established,and the recognition accuracy of the PCA-SVM model was verified in the two-class and multi-class identification modes,respectively.The accuracy rate of the leak identification represented by the Model-pca5 model was as high as 100% Model.The effects of different faults and leakage rates on the diagnosis and recognition performance of the model are studied,and the PCA-SVM model optimized by the Reflef F feature selection algorithm is verified and compared,and the optimized PCA-SVM model is obtained.