The rise of intelligent network-connected new energy vehicles signifies that the automobile industry is developing in the direction of intelligence, network connectivity and electrification. Based on the digital twin technology to build a multilayer system architecture, the virtual network mapping problem belongs to the combinatorial optimization problem of the optimization problem, the improved chaotic system is applied to the F function in the Feistel encryption framework in order to achieve good obfuscation and diffusion effects, and the cyclic shift control bits are added to the seed key, and a new dynamic key generation method is proposed. For new energy vehicle power battery fault prediction, a fault prediction model is established based on the LS-SVM algorithm, and the correlation calculation of power battery faults is carried out to compare the performance of SVM and LS-SVM methods. The encryption algorithm proposed in this paper has good security, plaintext sensitivity and fast encryption time; the SOC jump alarm, high-voltage interlock alarm, under-voltage alarm of on-board energy storage device, over-voltage alarm of on-board energy storage device have strong positive correlation with other faults, and the results of the Least-Squares Support Vector Machine regression prediction show that there will be a good tracking effect at the four points of time after consecutive prediction, and after the four points of time there will be a good tracking effect with real value offset, realizing new energy vehicle power battery fault prediction.
