In this paper, aiming at the inconsistency of decline law and short echelon service life of power batteries, an optimization method of residual value of retired batteries based on recession rate prediction is proposed. Firstly, based on the combination of grey prediction and least squares support vector machine, the historical usage data of batteries are mined to predict the decline rule of retired batteries, secondly, taking the highest utilization benefit of retired batteries as the objective function and considering the loss cost of retired batteries, the dynamic operation scheme of retired batteries in the whole life cycle is proposed. Finally, the rolling prediction is realized by using dynamic data. Taking a bus demonstration station in Yangtze River Delta as an example, the results show that the proposed method can effectively predict the decline law of batteries, and the revenue of retired batteries can be increased by 10%.