Experimental and numerical studies on lithium-ion battery heat generation behaviors

Current predictions of battery HGR (heat generation rate) mainly rely on Bernardi’s empirical equations, which suffer from limitations of adaptability for thermal use. A novel scheme based on experiments and BPNN (BP (back propagation) neural network is a multilayer feedforward neural network trained according to the error back propagation algorithm) is proposed in this paper. In the experiments, the thermal generation behavior of 18650 batteries under 108 different operating conditions was investigated, and the Ta (ambient temperature) as well as ID (discharge current) were extracted as inputs to the BPNN model. The experimental results show that the heat generation performance of the battery is strongly influenced by Ta and ID. The higher Ta and ID, the more obvious this effect is. The maximum error of the battery HGR obtained by using Bernardi’s empirical equations can reach 45.2 %. The predicted battery HGR using BPNN has good accuracy with error control within 5%. Accordingly, this paper provides the technical basis for proposing a promising thermal management solution. •A novel method for predicting the heat generation rate of batteries is proposed.•Discover defects in the empirical equations for battery heat generation.•Explore the factors affecting battery heat generation.•Match battery simulated heat generation rate and actual heat generation rate.