At present, the problem of water pollution in China is becoming more and more serious. In view of the scarcity of water resources and deterioration of environmental quality, it is necessary to effectively control pollutants in water bodies. This paper mainly studies the causes of pollutants in small sample data through learning and experiment methods, and takes them as the precondition to connect with the actual environment. On this basis, the improvement measures based on indoor water pollution monitoring and prediction are constructed to analyze, sort out and model the above problems. The emission concentration distribution map obtained by MATLAB software combined with laboratory simulation is used to verify that the above theoretical model is feasible and effective to solve the harmful problems caused by water quality deterioration. The test results show that, based on small sample learning and data fusion technology, It has a certain effect on the water pollution prevention project and can monitor the water pollution.