Simplified Soft Sensing Model Applied in the Centralized Regenerator of a Distributed Operating Liquid Desiccant Dehumidification System

Abstract

A distributed operating Liquid Desiccant Dehumidification System (LDDS) designed for commercial building applications allows one regenerator to handle multiple dehumidifier units, which provides the potential to apply low-grade energy in a centralized form for desiccant solution regeneration. An online soft sensing model with the Adaptive Network-based Fuzzy Inference Systems (ANFIS) structure is developed to satisfy the operating scheme. The number of input variables is reduced by the regeneration mass transfer model, and the fuzzy interference is optimized by Genetic Algorithm (GA) with a constrained objective function. The accuracy of the soft sensing models with different simplicity levels are validated and compared in case studies. Results show that the proposed soft sensing model is accurate, and the simplification approaches can reduce the size of the model significantly without affecting the predictive accuracy.