Abstract: In order to meet the demand of accurate perception of geological anomaly distribution in intelligent precision mining of working face，and further improve the inversion accuracy of electromagnetic wave CT of working face,an intelligent inversion model of electromagnetic wave CT under the constraint of multi-source prior information was proposed.Based on the functional optimization transformation of the sparse matrix equations of the electromagnetic wave CT inversion of the working face,the optimization and evolution mechanism of the single genetic algorithm (SGA) was analyzed,and the search optimization performance of the SGA intelligent algorithm by different genetic parameters was discussed.The objective function solution mechanism of electromagnetic wave CT inversion for geological abnormalities of working face based on Multi-population adaptive genetic algorithm (MAGA) was designed.In addition,three constraint models of range constraint,nearest constraint and average value constraint were proposed.The multi-source prior geological information revealed in the production process,such as roadway exploration,drilling and mining,was taken as constraint conditions,and the evolution direction of geological anomaly inversion objective function search can be continuously adjusted.Finally,the electromagnetic wave CT inversion model based on MAGA intelligent algorithm and multi-source prior information constraint was formed,and it was tested and verified by numerical simulation and engineering example.The test results show that the MAGA intelligent algorithm can fully combine the multi-group co-evolution and the adaptive adjustment strategy of genetic parameters,and has higher global and local search accuracy and stability than SGA; With the increase of external constraint data,the MAGA intelligent algorithm based on average value constraint can obtain a better convergence result of the objective function of electromagnetic wave CT inversion.Taking the electromagnetic wave CT geological anomaly inversion and interpretation within 8208 working face of Datong mining area in Shanxi as an example to test the built model,the predicted results were basically consistent with the actual revealed results,which further verifies that the intelligent inversion method under the constraints of multi-source information is reliable in engineering practice.The proposed model provides a new idea and method for the accurate inversion and prediction of geological anomalies within the working face.