Scholarlink

<p>Non-normal residuals in rule-based wastewater controllers undermine reliability and hinder statistical monitoring. This study resolves the issue by fusing a zero-order Sugeno fuzzy-inference system with a real-coded genetic algorithm that jointly tunes rule weights and membership functions while steering errors toward Gaussian form. Fuzzy-cognitive mapping reduces the candidate rule set to five dominant rules, which are then optimized on a training–testing split from a full-scale plant. The resulting controller lifts the treated-water-quality index from 51.08 to 69.28, lowers mean-squared error and attains a test RMSE of 0.03; the residual standard deviation is virtually identical, confirming a near-normal error distribution.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

From chaotic errors to natural curves: real-coded genetic calibration of wastewater treatment systems

  • 期刊论文

  • 发表日期：

    2025年8月15日

查看全文
  • H. Esmaeili,
  • M. A. Afshar Kazemi,
  • R. Radfar,
  • N. Pilevari

摘要

Non-normal residuals in rule-based wastewater controllers undermine reliability and hinder statistical monitoring. This study resolves the issue by fusing a zero-order Sugeno fuzzy-inference system with a real-coded genetic algorithm that jointly tunes rule weights and membership functions while steering errors toward Gaussian form. Fuzzy-cognitive mapping reduces the candidate rule set to five dominant rules, which are then optimized on a training–testing split from a full-scale plant. The resulting controller lifts the treated-water-quality index from 51.08 to 69.28, lowers mean-squared error and attains a test RMSE of 0.03; the residual standard deviation is virtually identical, confirming a near-normal error distribution.