With the background of the rapid development of distributed generators, realizing fast and accurate fault location of Active Distribution Networks (ADNs) faces challenges of universality and fault-tolerance. First, the applicability of the existing fault location methods is analyzed according to the relationship between the fault locations of ADNs and over-current alarm information. Based on improved switch function and fitness function, a mathematical model of fault location is constructed. Then taking the Customer Electric Information Acquisition System (CEIAS) as the second information source for fault location, a fault auxiliary partition method using multi-source information is proposed, and the discrete intelligent optimization algorithm is used to solve the problem. Finally, several kinds of fault with different kinds of information distortion scenarios are used for simulation. The test results of different fault location methods such as whether multi-source information is used and whether discrete optimization algorithm is used are compared and analyzed, so as to verify the applicability, accuracy and efficiency of the proposed method for the fault location problem of ADNs. This work is supported by Science and Technology Project of State Grid Shandong Electric Power Company (No. 52061718004F) and National Natural Science Foundation of China (No. U1866206).