Energy-saving Optimized Operation Method Of Low-dew-point Rotary Desiccant Cooling System Based On Machine Learning Algorithm

In industrial production,in order to improve process quality and promote product upgrading,enterprises are more and more stringent requirements for humidity in the production environment;the control of humility has become a non-negligible problem in part of the industrial upgrading process.Low-dew-point rotary desiccant cooling systems are widely used in industrial enterprises with ultra-low humidity production and storage requirements,at the same time,the dehumidification process is a high energy consumption process and excessive dehumidification often results in high production costs.Therefore,it is very important to realize the precise control of humidity in the control area and develop energy-saving dehumidification technology,both to ensure industrial production safety and to help enterprises reduce production costs.This paper proposes an energy-saving optimized operation method based on machine learning algorithm for the low-dew-point rotary desiccant cooling system.Taking the low-dewpoint rotary desiccant cooling system in a factory as a research object,analyze the operation mechanism and process requirements of the low-dew-point rotary desiccant cooling system,build the model of each part of the system in a data-driven way,and further optimize the system globally to achieve energy saving and optimization.In addition,the corresponding software platform is designed and developed for practical engineering applications.The main areas of research in this article include.(1)To address the problems of data loss and mutation caused by noise,network delay,equipment failure and other disturbances in data collection,transmission and storage processes,the use of physical and statistical discriminatory methods to process abnormal data,improve data quality,further data integration and normalization processing,and provide effective data for accurate modeling.(2)Establish models of low-dew-point rotary desiccant cooling system.According to the operating characteristics of the system,propose a desiccant wheel model based on RNN(Recurrent Neural Network).The EEP(Expected Error Percentage)and CV(Coefficient of Variation)of the first-stage desiccant wheel model are 3.77% and 5.46%,and the EEP and CV of the second-stage desiccant runner model are 2.60% and 4.14%;Propose an approximate characterization method for the wet load of the control workshop,and further establish a dew point temperature prediction model for the control workshop based on RNN,in which the models EEP and CV are 2.80% and 4.26%;Propose an electric power model and air volume model based on BP(Back Propagation)neural network.The EEP and CV of the electric power model are 4.44% and 5.49%,and the EEP and CV of the air volume model are 3.51% and 4.92%;Based on the air volume model,propose a cooling energy consumption model and a regeneration energy consumption model based on thermal balance equation.(3)Propose an energy-saving optimized operation method for a low-dew-point rotary desiccant cooling system based on an improved GA(Genetic Algorithm).Compared with the basic genetic algorithm,the system energy saving optimization based on the improved genetic algorithm improved the smoothness of the output parameters by 25.75%;Compared with the original control system,the dew point temperature stability of the control workshop was significantly improved as a result of the low-dew-point rotary desiccant cooling system parameter optimization simulation based on the improved GA,and the operating cost was saved by 28.77%.(4)Based on the actual conditions of the project,design the application process of the system energy-saving optimization method,and complete the design and development of the software platform for low-dew-point rotary desiccant cooling system energy-saving optimization operation which is convenient for operation and analysis.