The index system for security standard evaluation of water sources and the grading standards are established by selecting 9 indexes from 3 aspects of water security, water quality security and management security. Due to the shortage of the best projection direction PP model in practical application, a new bionic swarm intelligence algorithm, social spiders optimization (SSO) algorithm, is used to search the optimal projection direction PP model, and an evaluation model of SSO-PP is proposed. The SSO algorithm is verified by using 5 high-dimension complex functions and is compared with the particle swarm optimization (PSO) algorithm. The SSO-PP model is applied in the evaluation standards of an example. The results show that the SSO algorithm has better convergence precision and global optimization ability. The accuracy of PP evaluation model can be effectively improved by using the SSO algorithm to select the best projection direction of the PP model. The evaluation of Mudi River water source is “basically standard” and “standard” in 2010 and 2015 and “ideal” in 2020 and 2030.