[matlab project practice] prediction of remaining service life of lithium ion battery based on convolutional neural network and bidirectional long short time (cnn-lstm) fusion

Abstract
For the remaining service life of lithium ion battery (remaining useful life,RUL) The accuracy and stability of traditional prediction methods are low , Fused convolutional neural network ( convolutionalneural network,CNN) And two-way long-term and short-term memory ( bidirectional long short-termmemory,BiLSTM) Features of neural networks , A method for predicting the remaining service life of lithium ion batteries is designed . In order to make full use of the time series characteristics of battery data , Using convolutional neural networks (convolutional neural network,CNN) Extract the deep features of lithium-ion battery capacity data , utilize BiLSTM The memory function of neural network preserves important information in data , Predictive battery RUL Trends . By adopting NASA ( National Aero-nautics and Space Administration) Lithium ion battery data . The experimental results show that ,CNN-BiLSTM It has higher prediction stability and accuracy .
Problem description
Lithium ion batteries are mainly composed of positive electrodes 、 Insulating sheet 、 The negative electrode and diaphragm are composed of , Pictured 1 Shown . Lithium ion battery cycles 2 It refers to when the lithium-ion battery completes a complete charging and discharging cycle , The number of cycles is increased by 1.
A healthy state (state of health,SOH) Through the internal resistance of lithium ion battery 、 Lithium ion battery capacity 、 Characterization of physical quantities such as lithium ion battery power and lithium ion battery peak power , Used to judge the degradation degree of lithium ion battery performance . Lithium ion battery capacity has a good trend of continuous degradation, and capacity is an important parameter that can directly characterize the current power storage capacity of lithium ion batteries , It is widely used as lithium ion battery RUL Predicted health indicators . In this paper, the ratio of the current lithium-ion battery capacity to the initial capacity is taken as SOH Define the standard , Can be defined as (1).

among ,capacity, by t Always fully charged , The capacity of lithium-ion batteries ;capacity. Is the rated capacity of the lithium ion battery at the initial time .
Lithium ion battery RUL It refers to the lithium battery after a certain charge and discharge process , The number of charge and discharge cycles required to decay from the current capacity to the failure threshold , The definition is as follows :

among ,N It refers to the current moment , Number of cycles of lithium ion battery . End of life of lithium ion battery (end of life,EOL) It is defined as lithium ion battery SOH Number of cycles when the value first drops to the specified failure threshold . When SOH Decay to 70% when , Then the lithium ion battery reaches the end of its service life ( The failure threshold ).

CNN-BiLSTM Fusion neural network

Training results ：



RMSE =0.7911
** Download link ：** Code link