Abstract: A method of lithium battery health assessment was proposed based on the framework of multi-scale data fusion with artificial neural network as the core. The internal constant resistance, the sample entropy of charging voltage and the isobaric discharge time were selected as typical characteristic parameters. The three-layer distributed artificial neural network was established for the multi-scale data fusion, and the calculated fitting output was used as reference value for the health assessment. The proposed method was verified by the national aeronautics and space administration（NASA） experimental datasheet. The results show that the method based on the lithium battery typical characteristic parameters and the multi-scale data fusion framework can rapidly iteratively converge to complete the evaluation and fitting of the health status of lithium battery. Comparing the calculation results of the proposed method with the test platform data, the average error is less than 3%, and the evaluation performance degradation trend is consistent with the actual deterioration trend.