Operational performance estimation of vehicle electric coolant pump based on the ISSA-BP neural network

Accurately estimating the operational performance of electric coolant pump (ECP) can support long-term sensorless operational monitoring and reduce the cost and energy consumption of a vehicle thermal management system. However, there are some problems such as low estimation precision of theoretical model and back propagation neural network (BPNN) models, and the input parameters of existing studies are difficult to obtain at the ECP. In this study, a novel ISSA-BPNN estimation model is proposed that combines a hybrid strategy improved sparrow search algorithm (SSA) with the BPNN after hyperparameter optimization, and for the first time analyzes and uses the total power easily obtained as the input data of the model. Multiple experimental results show that the estimation precision and reliability of the proposed ISSA-BPNN model are much higher than those of the present theoretical models and BPNN methods. The average training time of the proposed ISSA-BPNN model is 226.9 s, and the average real-time operation time is about 5 ms, which meets the real-time application requirements. The proposed model is also applicable to the operational state estimation of other types of integrated pumps.