A new approach for Lead-Acid batteries modeling by local cosine

In this paper a new approach based on the Local Cosine Bases is proposed in order to obtain an easy and improved Lead-Acid battery modeling so avoiding the training process of RNN and the need of big amount of relative data training sets. The wavelet packet analysis give us a tools to achieve major improvements on data discrimination and analysis. In particular the Local Cosine Bases transform allows us to sensitively reduce the number of significant coefficients, it is useful to synthesize a complex signal with an high degree of approximation of the original signal.