A novel BA-ABC-ELM model for estimating state of health of lithium-ion batteries

Accurate state of health ( SOH ) estimation for lithium-ion batteries (LIBs) is a primary concern while estimating state of charge ( SOC ) of the battery and the cruising range in electric vehicles. Our research group has previously studied the construction mechanism of the Bat Algorithm-Extreme Learning Machine (BA-ELM) model and applied the model to estimate the SOH of batteries for the first time. To improve the global exploration and local exploitation abilities of the model, three methods are proposed to construct a BA-Artificial Bee Colony (ABC)-ELM model. Firstly, the excellent exploration and search mechanism of ABC is investigated to increase the diversity of global solutions. Thus, a method to give the bat a trial before they fly randomly is proposed. Secondly, considering the limitation that the search frequency of BA is generated randomly without considering various positions of individuals, a method based on the differences in Euclidean distance between individuals and the optimal position is proposed to adaptively change the search frequency. Thirdly, to avoid premature convergence or convergence difficulty of the model, an adaptive inertia weight coefficient is introduced to balance the global exploration and local exploitation abilities. Finally, the fitness function, the average velocity modulus length of bats, the LIB’s capacity estimation error and the model training time are used to systematically evaluate the performance of SOH estimation models. The results show that the root mean square error and mean absolute error during SOH estimation for BA-ABC-ELM model are only 62 % and 57 %, respectively, of those for BA-ELM model. The average training time of the two models differed by 1.57 %. This indicates that the computational efficiency of the two models is comparable, and the BA-ABC-ELM model has higher SOH estimation accuracy and better generalization performance.