Abstract: In order to overcome the difficulty of determining the optimal projection direction in practical application of projection pursuit (PP) model, we propose a method of searching the optimal projection direction by using Social Spiders Optimization (SSO) algorithm, and hence building a SSO-PP model of water quality assessment. Four monitoring sections of Dianchi Lake watershed in Yunnan province from 2003 to 2013 were taken as case study. Fiver factors which have big influences on water quality were selected as assessment indicators: NH₃-N, TN, COD_Mn, BOD₅, and TP. The rating criteria of water quality were obtained according to index standard thresholds. Results showed that the assessment results of SSO-PP model were consistent with those of single-factor analysis, some superior for section 1^#, 2^#, and 4^#. Kendall statistical test showed that the water quality of section 3^# and section 4^# improved apparently. The results of SSO-PP model are objective and reasonable, indicating that the model can be applied to assessing water quality effectively.

Key words: water quality assessment, projection pursuit, social spider optimization algorithm, swarm intelligence, Dianchi watershed