A method for liquid metal battery curve reconstruction and SOH estimation based on fragmented data

Abstract

The application discloses a kind of liquid metal battery curve reconstruction and SOH estimation method based on fragmentation data, first, the short discharge voltage-capacity fragmented data segment generated by power grid random scheduling is collected;Then, a curve reconstruction model composed of one-dimensional convolutional neural network layer, bidirectional long short-term memory network layer and weak physical information constraint layer is constructed, and the complete discharge curve is recovered from the sparse segment;Then, multi-dimensional health characteristics are extracted from the reconstructed curve, and after enhancement by canonical correlation analysis, input into the online sequential extreme learning machine model with self-feedback pseudo-label mechanism, the model generates self-feedback pseudo-label by exponential moving average adaptive bias correction, and the weight is updated in real time using recursive least squares method, so as to realize the adaptive tracking of capacity regeneration phenomenon in the whole life cycle of liquid metal battery and accurate estimation of SOH.The application effectively solves the estimation problem under extreme data loss, and significantly improves the accuracy and robustness of SOH estimation.

Images