Identification of Mine Water Sources using a Multi-Dimensional Ion-Causative Nonlinear Algorithmic Model

Abstract

Abstract This paper investigates the water quality system using the Piper diagram method to qualitatively analyze the distribution, characteristics, and laws of typical ion content. It also compares the differences between ion components of each aquifer with the closeness of the water body to be discriminated by combining statistical characteristic values. The R factor is utilized to simplify the index attributes, reduce input information dimension, and construct the coupled R-SVM discriminative model of Zhaogezhuang Mine. The input index is established as typical ion content, and the output is water source type. The discriminative model automatically establishes the mapping relationship between water quality indexes and discriminative criteria by learning the inherent property law between water quality samples. The independence components after dimensionality reduction are used as new discriminative indexes. The accuracy of the coupled model classification was 90.90% in the verification of the water source discrimination example of Zhaogezhuang mine. The coupled quantitative discriminant model based on R-factor and support vector machine provides an auxiliary verification and scientific decision for qualitative water chemistry analysis and provides a new idea for water source identification. Compared with the traditional qualitative way of water chemistry characteristics, Fisher function discrimination method, and single support vector machine model, this method improved the accuracy by more fully exploiting the internal laws of the data.