The invention discloses a medium-and-long-term power market risk assessment method based on machine learning. The method comprises the steps of collecting the power market risk related data, the system operation constraint condition data and the quotation data; establishing a medium-and-long-term power market risk assessment index system according to the acquired data, performing quantitative analysis on the hazard degree of the typical risk events occurring in the medium-and-long-term power market, and performing division according to the total hazard amount; carrying out data evaluation on the obtained data by using a Delphi method to obtain the medium and long term power market risk evaluation data; evaluating the medium and long term power market risks based on machine learning, establishing a general comprehensive evaluation model for the medium and long term power market risks, and learning according to existing market risk evaluation data to establish a learning machine; and feeding back a calculation result of the established medium-and-long-term market risk assessment learning machine, and assessing the medium-and-long-term market risk. According to the invention, an effective risk early warning effect can be provided for the market operators.