The invention discloses a wind power generation short-term load forecast method of a least squares support vector machine. The method comprises the following steps of 1, preprocessing original data; 2, carrying out principal component analysis on an original data sequence which is input to the least squares support vector machine by a principal component analysis method, and analyzing and extracting a key impacting indicator of wind power loads; 3, building a mathematical model of the least squares support vector machine; 4, inputting the analyzed and extracted key impacting indicator to the mathematical model of the least squares support vector machine to be used as a training sample and a testing sample; 5, carrying out forecast on testing sample data by the mathematical model of the least squares support vector machine to obtain a forecast result. According to the wind power generation short-term load forecast method of the least squares support vector machine, the principal component analysis method and the mathematical model of the least squares support vector machine are combined, the calculated amount is reduced, the operability is increased, and the whole forecast performance and the whole forecast accuracy are improved.