Artificial neural networks simulate the human brain network by way of memory

processing information with high intelligence.In recent years

it is widely used in lead-acid battery remaining power prediction research

However

convergence is slow

sensitive to initial and easier to fall into local minimum is a single neural network algorithm shortcomings which difficult to resolve.In view of this situation

this project will improve the ant colony algorithm with BP neural network integration.First

training BP neural network weighting parameters globally with improved ant colony algorithm

and then

going to further partial learning with the BP neural network algorithm

so as to get the best BP neural network weights.Finally

has been certified that this method can promote the rate of convergence visibly and prove the accuracy of the prediction of the BP networks

and can predict the remaining capacity of the battery accurately.