Lithium Ion Battery Health Prediction via Variable Mode Decomposition and Deep Learning Network With Self-Attention Mechanism

Abstract

Battery health prediction is very important for the safety of lithium batteries. Due to the factors such as capacity regeneration and random fluctuation in the use of lithium ion battery, the accuracy and generalization ability are poor when using a single scale feature to predict the health state of lithium ion battery. To solve these problems, we propose a comprehensive prediction method based on variational mode decomposition, integrated particle filter, and long short-term memory network with self-attention mechanism. Firstly, the capacity data of lithium ion battery is decomposed by variational mode decomposition to obtain the residual component which can reflect the global degradation trend of lithium ion battery and intrinsic mode functions component that can reflect the local random fluctuation. Then, the particle filter algorithm is employed to predict the residual component, and the long short-term memory network with self-attention mechanism is proposed to predict the intrinsic mode functions component. Finally, the prediction results of each subcomponent are reconstructed to obtain the final prediction value of lithium ion battery health state. The experimental results show that the prediction method proposed in this article has good prediction accuracy and stability.

Remaining useful life prognostics of lithium-ion batteries based on a coordinate reconfiguration of degradation trajectory and multiple linear regression

“…Yu et al (2022) presented an improved Euclidean distance method and a cosine similarity method from the perspective of similarity for online diagnosis of multi-fault inseries connected battery packs. Ge et al (2022) established a comprehensive lithium-ion battery SOH estimation method based on variable mode decomposition, particle filter, and long short-term memory with a self-attention mechanism. Hu and Zhao (2022) presented a RUL prognostics method for lithium-ion batteries based on the wavelet threshold denoising and transformer neural network.…”

“…Lithium-ion battery manufacturers must conduct extensive life testing to obtain the design formulations, structural parameters, and operating environments that maximize battery life. Predicting the life of newly developed batteries based on the full-life degradation data of similar batteries (reference batteries) can significantly reduce product development time and cost (Finegan and Cooper, 2019;Fan et al, 2021b;Ge et al, 2022). Therefore, the remaining useful life (RUL) prognostics of lithium-ion batteries are beneficial for properly using and maintaining batteries and assisting in designing new products during the R&D phase.…”

Research on ERS point registration of large-size measurement field based on neural network non-uniform temperature field

“…Research on artificial neural networks continues to advance and has significantly progressed in various areas (Chen et al , 2020; Wang et al , 2021; Ge et al , 2022a; He et al , 2022; Zhou et al , 2022). In response to the problem of environmental pollution caused by aviation, scholars have developed a neural network prediction model for aviation emissions to assess the impact of aviation emissions on the environmental climate (Ge et al , 2022b). For the problem of thermal deformation caused by the temperature of machine tool processing, scholars use neural networks to build prediction models to predict accurately and compensate for the thermal deformation error of machine tools (Yan and Yang, 2009; Li et al , 2019; Zhang et al , 2019).…”

A similarity-based remaining useful life prediction method using multimodal degradation features and adjusted cosine similarity

“…On the other hand, the real health degradation of the system is often subjected to phenomena such as random disturbances and degradation regeneration [17,18]. Since the original sensor signals are usually non-stationary and non-linear, the health degradation curves obtained through sensor fusion have degradation regeneration and instantaneous mutations issues which may result in large errors in the similarity-matching process.…”

“…The technique has received increasing attention because it solves the problems of endpoint effects in EMD. Ge et al [17] proposed an integrated prediction method based on VMD and long short-term memory networks with selfattention mechanisms to mine the degradation information for battery. However, VMD requires a human-defined modal number k, which has a strong effect on the accuracy of the results.…”