A novel optimized control strategy based on BP neural network for dynamic heating of sorption energy storage system

The discharging process of the sorption energy storage system (SES) is complex and affected by multiple parameters, making it difficult to match the building load with fluctuation. To achieve the optimized control of the SES operation and the effective energy releasing, a control strategy based on bp neural network is proposed. By the heating load parameter as the input values and the reactor inlet operation parameter as the output values, the bp neural network model is established. Energy analysis shows that the optimized case with bp neural network can reduce the excessive reactor heat release in the early heating stage and reduce 53.31 kWh energy loss. The heating time is extended by 3.20 h, and the total energy supplied is increased by 67.52 kWh. The reactor energy utilization efficiency and system energy efficiency can reach 75.12 % and 71.03 % of optimized case, which are 23.01 and 20.94 percentage points higher than the reference case. The SES is a potential alternative to electric heating. The optimized case can save 5836.24 $ per year, which is 2.31 times that of the reference case. In terms of CO₂ emission reduction, the optimized case is 36.69 % higher than the reference case.