Under the current power deviation assessment mechanism of China & aposs southern power market, electricity sales companies with more accurate power load forecasting technology will have less deviation power in the current medium and long-term market, to gain a competitive advantage at a lower cost. In order to explore what factors have influence on the power load forecasting and how to use the forecasting model to achieve better power load forecasting precision in the mature power spot market, four kinds of factors affecting electric load are summarized: historical load data, meteorological factors, time factors and economic factors. The load forecasting model for long-term and short-term Memory Network ( LSTM) which takes consideration of multi-variable factors is constructed. Taking Queensland, Australia, as an example for prediction analysis, the results show that the error of power load forecasting method based on LSTM is lower than that of the power load forecasting method based on ARIMA model, and the prediction effect is better. Among the influencing factors, the comparison of the results of the calculation examples shows that the date type has the strongest influence on the forecasting results of power load forecast, followed by the electricity price factor, and finally the minimum and maximum temperatures. This prediction method can also be used in the PJM power market similar to the United State.