This paper designs thirty five indicators from such three dimensions as product output, industry investment and boom index closely related to manufacturing industry’s electricity consumption through its each micro industry, selects key influencing indicators in accordance with relevance principle and establishes forecast model of manufacturing industry’s electricity consumption by adopting QGA LSSVM algorithm.A forecast instance of Anhui’s quarterly cumulative manufacturing industry’s electricity consumption demonstrates that the results of this method are accurate and reliable, which is significantly super to methods dependent on manufacturing industry’s output or non key influencing indicators, which can provide an effective tool for electricity market forecasters and economic operation analysts.