摘要
提出了一种基于多模型组合的冷凝器污垢预测新方法。该方法采用经验模型、自适应指数平滑模型、灰色模型、T%DS模糊模型等多种模型预测污垢的增长,并通过遗传算法对模型参数、各模型输出之间的组合系数进行自适应滚动优化调整,以适应冷凝器水质及工况参数的动态变化,从而取得比单个预测模型更好的预测精度。试验结果表明:该方法短期污垢预测效果好,中长期污垢预测精度较高,是实现冷凝器污垢预测的有力工具。
A novel prediction approach based on multi-model combination for condenser fouling is proposed . In the approach, experiment model, adaptive index smoothing model, grey model and T-S model are adopted to predict the fouling trend. Adaptive rolling optimization technique based on genetic algorithm is employed to adjust model parameters and combination coefficients between the model output in accordance with the variation of cooling water quality and condition parameters in condenser. Based on it, a test on an actual condenser is conducted. The results show the approach achieves fouling prediction precisely under the conditions of short-term, medium-term and long-term fouling prediction and can be considered as a powerful tool for fouling prediction.
出处
《传感技术学报》
CAS
CSCD
北大核心
2005年第2期225-229,共5页
Chinese Journal of Sensors and Actuators
基金
国家自然科学基金项目60075008
湖南省科技攻关项目04GK3049资助
关键词
冷凝器污垢预测
多模型组合
遗传算法
自适应滚动优化
condenser fouling prediction
multi-model combination
genetic algorithm
adaptive rolling optimization
