In order to solve the problem that samples of minority class of critical risk sources can't be effectively identified due to the deliberate attack samples finite unbalance, a multi-classifier ensemble weighted balanced distribution adaptive method for critical risk sources identification is proposed. Firstly, ensuring that the minority samples are fully selected, the source domain multi sample training set is obtained by random sampling, and different initial weights are given to the samples to fully train the minority samples. Then, the classifier of the sample set in the source domain is trained, and the pseudo label of the target domain is optimized and the weight matrix is updated after many iterations. Finally, the selected base classifiers are integrated into strong classifiers through the strategy of multi classifier integration, and the recognition performance of classifiers is evaluated by macro average and micro average evaluation indexes. The global terrorism database(GTD) data is used to verify the proposed method, which can effectively identify a small number of samples while ensuring the overall accuracy.