Sensorless temperature estimation of lithium-ion batteries by integrating physics with the state-dependent model

State of temperature (SOT) is a key state quantity that affects the performance of batteries and the prevention of thermal runaway. The large-scale battery system is still limited by insufficient temperature sensors to obtain SOT for each individual battery. In this article, a new method for sensorless SOT estimation of batteries through the feedback of terminal voltage is proposed. Firstly, a state-dependent autoregressive model with exogenous inputs (SD-ARX) that combines a linear autoregressive structure with function-type coefficients is constructed to capture the dynamic nonlinear relationship between the terminal voltage and the state quantities (SOT, state of charge, and current) under different temperature ranges and working conditions. On this basis, this model is coupled with the thermal model to realize the SD-ARX-thermal coupling model. Finally, the SOT is estimated via the unscented Kalman filter, and the state of charge is estimated simultaneously. Meanwhile, the observability matrix of the system is derived, which further lays the theoretical foundation of the proposed sensorless SOT estimation method. Experimental validation has been carried out in different temperature ranges and working conditions, and the proposed method has satisfactory estimation accuracy even with different initial values and measurement noises.