Optimal Capacity Matching And Control Evaluation Of Centralized Heating System With Air Source Heat Pumps

With the increasing emphasis on environmental protection and the growing demand for energy,air source heat pumps are receiving more attention as a new low-carbon and environmentally friendly heating system.The application of air source heat pumps is also expanding,especially in large buildings and small communities or parks,where the use of multiple parallel air source heat pump water units is becoming more common.However,in practical applications,there are issues of unreasonable unit capacity matching and inefficient operation control when multiple parallel heat pumps are used,resulting in a loss of unit performance.This article conducts in-depth research on the capacity matching and operational strategies of multiple parallel air source heat pump water units based on the characteristics of air source heat pumps.First,based on typical building heat load curves and unit performance curves,a method is proposed to calculate the average performance of different selection schemes.By analyzing the impact of unit load ratio and unit capacity on different selection schemes,the unit performance under different selection schemes is calculated.The results show that when the total number of units is 5 and the unit capacity ratio is 2∶5∶6∶7∶8,the average coefficient of performance(COP)for the entire heating season can be increased by 6% compared to the traditional scheme of selecting multiple units of the same capacity.Next,a thermal capacitance-resistance(RC)model is established for the building and end radiator.The RC model is constructed for the enclosure structure and end structure of a single heat pump household,and genetic algorithms are used to reverse identify the model parameters,obtaining the most realistic parameters for the building RC model and radiator RC model.The identification results of the model indicate that different RC models have a certain influence on the accuracy of simulation results,among which the 4R3 C model is the closest to the actual situation and can effectively reflect the actual heat transfer process.This model can serve as the fundamental physical model for the operation control of the units.Finally,an optimized control strategy for the units is obtained based on reinforcement learning.Building on the RC model of the building and end radiator,a reinforcement learning control strategy is proposed with the objective of improving COP as the reward function while maintaining indoor temperature stability.The control scheme for optimizing unit performance is trained.The training results are compared with reward functions that aim to reduce energy consumption and decrease heat generation to validate the rationality of the reward function.The test results show that the reinforcement learning optimization control scheme improves performance by2.9% compared to traditional feedback control methods,while achieving an overall energy consumption reduction of 8.4%.