The invention relates to a power system load classification field, and provides an adjustable high-energy load identification method based on combination of a fish swarm optimization algorithm and a likelihood c-means clustering algorithm. Since a high-energy load has certain adjustable characteristics, the adjustable capacity is large and the speed of responding to a dispatching command is fast,reasonably using the high-energy load can effectively solve the problem of wind curtailment due to low peak shaving capacity of a conventional unit and insufficient power transmission capacity of an outer conveying channel. The method comprises the steps of firstly, carrying out standardized processing on wind power and transformer substation comprehensive load sample data, classifying the comprehensive loads of a transformer substation connected with high-energy load feeder lines into adjustable high-energy loads and uncontrollable loads through a clustering method based on combination of thefish swarm optimization algorithm and the likelihood c-means clustering algorithm based on continuous adjustable characteristic parameters of the high-energy loads, and finally identifying the proportion of the adjustable high-energy loads in the comprehensive loads.