摘要
污水排放会对环境和生态系统产生负面影响,准确的水质预测可及时预警污染风险,为水环境保护提供决策支持。基于此,提出一种基于适应度平衡改进算术优化算法(FAOA)和鲁棒集成深度随机向量函数链接网络(ORedRVFL)的污水水质预测模型(FAOA-ORedRVFL)。首先,针对算术优化算法(AOA)引入适应度平衡策略,增强其寻优能力;其次,利用FAOA优化ORedRVFL模型的超参数,增强模型的预测精度和鲁棒性。同时,采用互信息(MI)选择与待预测变量相关程度较高的变量,提高数据建模效率。结果表明,FAOA-ORedRVFL模型具有较好的水质预测性能,与ORedRVFL模型相比,出水氨氮和总氮的均方根误差(RMSE)分别提升了17.88%和22.19%;通过设计噪声干扰实验,进一步验证了FAOA-ORedRVFL模型的抗干扰性能。FAOA-ORedRVFL模型为复杂污水环境下的水质预测提供了高精度、强鲁棒性的解决方案。
Wastewater discharge has negative impacts on the environment and ecosystem.Accurate water quality predictions can provide timely warnings of pollution risks and offer decision support for water environmental protection.A wastewater water quality prediction model(FAOA-ORedRVFL)was proposed based on the fitness-distance balance arithmetic optimization algorithm(FAOA)and the outlier-robust ensemble deep random vector functional link network(ORedRVFL).Firstly,an adaptability balance strategy was introduced into the arithmetic optimization algorithm(AOA)to enhance its optimization capabilities.Secondly,the FAOA was used to optimize the hyperparameters of the ORedRVFL model,thereby enhancing the prediction accuracy and robustness of the model.Additionally,mutual information(MI)was employed to select variables with higher relevance to the target variable,thereby improving data modeling efficiency.The results indicated that the FAOA-ORedRVFL model demonstrated excellent water quality prediction performance.Compared with the ORedRVFL model,the root mean square error(RMSE)for ammonia nitrogen and total nitrogen in the effluent improved by 17.88%and 22.19%,respectively.Furthermore,through the design of noise interference experiments,the anti-interference performance of the proposed model was further validated.The FAOA-ORedRVFL model provided a high-precision,robust solution for water quality prediction in complex wastewater environment.
作者
张楚
葛宜达
李燕妮
李正波
彭甜
ZHANG Chu;GE Yida;LI Yanni;LI Zhengbo;PENG Tian(Faculty of Automation,Huaiyin Institute of Technology,Huai’an 223003,China;Jiangsu Permanent Magnet Motor Engineering Research Center,Huai’an 223003,China)
出处
《工业水处理》
北大核心
2025年第9期71-77,共7页
Industrial Water Treatment
基金
国家自然科学基金项目(62303191,62306123)
江苏省高校自然科学基金面上项目(23KJD480001)
江苏高校“青蓝工程”资助项目(苏教师函〔2024〕14号)
江苏省研究生科研与实践创新计划项目(SJCX24_2140,SJCX24_2141)
淮阴工学院研究生科研与实践创新计划项目(HGYK202412,HGYK202413)。
关键词
水资源污染
水质预测
互信息
鲁棒
算术优化算法
water resource pollution
water quality prediction
mutual information
robustness
arithmetic optimization algorithm
