In order to expand the application of intelligent algorithm and projection pursuit (PP) fusion model in comprehensive evaluation of water quality, a new group intelligence algorithm-electrostatic discharge algorithm (ESDA) was introduced. Six typical test functions were selected to simulate and verify the ESDA. The simulation results of the current optimization teaching and learning optimization (TLBO) algorithm, the gray wolf optimization (GWO) algorithm and the whale optimization algorithm (WOA) were compared. The ESDA-PP evaluation model was constructed. The water quality evaluation of the water function area of the four sections of the Panlong River in Wenshan Prefecture from 2015 to 2017 was carried out as an example. The results showed that the accuracy of ESDA optimization is much better than that of TLBO, GWO and WOA, and it has excellent optimization precision and global search ability. The ESDA-PP model has a comprehensive evaluation rate of 93.3%, 83.3%, 83.3% and 100% for the water quality of the Tianshengqiao, Lishizhai, Huaqiao and Tianbao Bridges, respectively, which can meet or basically meet the water quality protection target requirements of the water function area. The water quality of Tianshengqiao, Lituzhai and Huaqiao in the dry season is generally better than that in the flood season, and the non-point source pollution is the dominant factor. The water quality of Tianbao Bridge during the flood season is generally better than the dry season, and point source pollution is the dominant factor.