Cooperative Estimation Method of Remaining Power of Li-ion Battery and Sensor Bias Based on Neural Network and Unscented Kalman Filter

Abstract

The invention discloses a method for cooperatively estimating the residual power of a lithium electronic battery and sensor deviation based on a neural network and an unscented Kalman filter, comprising the following steps: S1: conducting a lithium electronic battery charging and discharging experiment and collecting sample data, including training Data and test data; S2: determine the input and output variables of the neural network, and establish the RBFNN model of the SOC; S3: learn the parameters of the established RBFNN based on the training data set, and obtain an accurate RBFNN model; S4: use the test data to establish the RBFNN model RBFNN conducts an independent accuracy test; S5: Set the SOC as an internal state, and design RBFNN-UKF to realize real-time estimation of SOC under the condition of uncertain initial SOC; S6: Set the sensor bias as an expansion state, in the original RBFNN-UKF Based on the design and expansion of RBFNN‑UKF, the collaborative estimation of SOC and unknown sensor bias is realized. The invention can realize the cooperative estimation of SOC and sensor deviation, and has the advantages of fast convergence speed, high precision and small error.

Description

technical field [0001] The invention relates to the field of electric vehicle batteries, in particular to a method for cooperatively estimating the remaining power of lithium electronic batteries and sensor deviations based on a neural network and an unscented Kalman filter. Background technique [0002] The rapid development of the automobile industry has indirectly led to a large amount of energy consumption and the deterioration of the environment. Therefore, battery-powered electric vehicles have gained widespread attention in the new energy vehicle industry due to their advantages of low energy consumption, zero emissions, and high cost performance. Due to the advantages of high energy density, low self-discharge, fast charging, and long life, lithium-ion batteries are widely used in portable appliances such as laptops, cameras, and mobile communications, and thus become the first choice for new energy electric vehicles. [0003] The battery management system is one of...

AI Technical Summary

Problems solved by technology

The shortcomings of the existing methods are as follows: 1. The ampere-hour integration method needs to know the initial value of the SOC, and at the same time as an open-loop algorithm, due to uncertain interference such as temperature and current measurement deviation, there will be an error accumulation effect in the SOC calculation process; 2. The open-loop voltage method can only estimate the SOC when the battery is in an open circuit and under long-term static conditions, and cannot be estimated online, and the static process generally takes several to ten hours; Training; 4. The traditional Kalman filter algorithm is only suitable for linear systems, so it is not suitable for lithium-ion batteries with highly nonlinear systems

Images