Abstract: Aiming at the problem that a single set of battery external characteristic data or specific signals cannot evaluate the battery health state (SOH) nor characterize the essential changes of battery internal characteristics, this paper proposes a monitoring method and system that comprehensively considers both internal and external characteristic parameters of batteries, so as to realize multi-parameter inspection of batteries in energy storage power stations. Specifically, the external characteristic detection mainly relies on an intelligent inspection robot to conduct fixed-point detection and periodic inspection, and acquires battery temperature and image information through non-contact infrared temperature measurement, with the battery temperature monitored based on the information entropy method. For internal characteristics, a collaborative CNN-LSTM-Attention framework is adopted to predict and fit the attenuation curve of battery internal SOH with high precision. This method can comprehensively capture the changes of battery states, improve the accuracy and response speed of early warnings, and thus better guarantee the operational safety of energy storage.