Affiliations: School of Environment, Beijing Normal University,
Beijing, 100875, China | The State Key Laboratory of Water Environment
Simulation, Beijing Normal University, Beijing 100875, China | School of Hydropower Engineering, Sichuan University,
Chengdu 610065, China
Abstract: One of the difficulties frequently encountered in water quality
assessment is that there are many factors and they cannot be assessed according
to one factor, all the effect factors associated with water quality must be
used. In order to overcome this issues the projection pursuit principle is
introduced into water quality assessment, and projection pursuit cluster (PPC)
model is developed in this study. The PPC model makes the transition from high
dimension to one-dimension. In other words, based on the PPC model, multifactor
problem can be converted to one factor problem. The application of PPC model
can be divided into four parts: (1) to estimate projection index function
Q(a); (2) to find the right projection direction a; (3) to calculate
projection characteristic value of the i th sample z_i, and
(4) to draw comprehensive analysis on the basis of z_i. On
the other hand, the empirical formula of cutoff radius R is developed, which is
benefit for the model to be used in practice. Finally, a case study of water
quality assessment is proposed in this paper. The results showed that the PPC
model is reasonable, and it is more objective and less subjective in water
quality assessment. It is a new method for multivariate problem comprehensive
analysis.
Keywords: projection pursuit, cluster, model, water quality assessment
