Lithium battery core and surface temperature estimation method based on filtering

Abstract

The invention discloses a lithium battery core and surface temperature estimation method based on filtering, and belongs to the technical field of state estimation. According to the method, iteration of a system state is carried out by using a filtering method, prior distribution of system noise and disturbance does not need to be known in advance, and the applicability of a state variable method is improved; in a state variable iteration process, each particle in particle filtering is wrapped by using a multi-cell body, iteration of the particle filtering particles and iteration of a space body are synchronously carried out, and region limitation is performed on a diffusion range of the particles by using the multi-cell body, so that the distribution of the particles is closer to a real value and is denser; and the dense distribution of the particles means that the weights of the particles are closer, the number of the particles with small weights is smaller, the particles are prevented from being replaced in the resampling process, the diversity of the particles is reserved, and then the problem that the system robustness is poor due to particle shortage in the traditional particle filtering algorithm process is solved.

Description

technical field [0001] The invention relates to a method for estimating the core and surface temperature of a lithium battery based on filtering, and belongs to the technical field of state estimation. Background technique [0002] With its advantages in power density, energy density, cycle life, self-discharge rate and price, lithium battery energy storage system has become one of the main choices for clean electric energy buffer and electric vehicle power source. In order to make it work in a normal working state at all times, it is necessary to estimate its working temperature (core and surface temperature) in real time. [0003] As an intelligent algorithm for state estimation, particle filter algorithm is widely used in the field of state estimation because of its low dependence on system model and noise distribution. However, in the existing methods, due to the existence of noise, the diffusion of particles in the particle filter algorithm is uncontrollable, and small...

AI Technical Summary

Problems solved by technology

However, in the existing method, due to the existence of noise, the diffusion of particles in the particle filter algorithm is uncontrollable, and small weight particles far away from the true value will be generated, and the small weight particles will be replaced by large weight particles in the resampling process. Reduced the diversity of particle samples after the resampling process

In the existing methods, the Markov chain is often constructed, and the Markov Monte Carlo method is used to move the particles and then resample the particles, but the construction of the Markov chain is often quite difficult

[0004] Because the existing method reduces the diversity of particle samples after the resampling process, the particles with large weights are dominant in the iterative process, which is insensitive to system parameter changes and reduces the robustness of the system.

Images