Abstract: Against the backdrop of the vigorous development of the new energy industry, lithium-ion batteries, as the core power units of electric vehicles and energy storage devices, pose a key technical challenge in ensuring the safe operation of energy systems, which lies in accurately monitoring the degree of performance degradation of lithium-ion batteries. Building a high-precision battery State of Health（SOH） evaluation system can not only optimize energy management efficiency and slow down the battery aging process but also significantly reduce operation and maintenance costs, thereby curbing resource consumption and ecological pollution. Especially during the advancement of the dual-carbon strategy, this technology holds vital engineering significance for realizing the full-life-cycle value mining of power batteries. Traditional methods for constructing lithium-ion battery SOH estimation models have several issues, such as complicated partial differential equations in chemical models and high computational requirements, as well as strict measurement environmental requirements for parameter identification in equivalent circuit models, which limit their application scope. This study employs Long Short-Term Memory（LSTM） networks to build a high-precision SOH estimation model to accurately assess the health state of lithium-ion batteries. By learning from the battery’s historical usage data through algorithms, this approach can estimate the full-life-cycle SOH without requiring an understanding of the specific meanings of the internal chemical reaction aging mechanisms of the battery and complex mathematical model expressions. Simulation tests have been carried out to verify the performance of the designed algorithm.