基于投影寻踪的污水处理厂运营绩效评价研究

Operation Performance Evaluation of Wastewater Treatment Plant Based on Projection Pursuit

由于国内污水处理厂运营效果的衡量指标涉及范围广、种类多,很难找出单一指标对不同污水厂的运营绩效进行对比分析,因而使得不同污水厂运营绩效的评价难度较大。在构建衡量污水厂运营绩效指标体系的基础上,应用加速遗传算法和投影寻踪模型,将衡量污水厂运营绩效的多维评价指标投影到一维数据空间,实现了复杂系统的横向对比。为了证明该评价结果的有效性,并将其与主成分分析法的结果进行了对比。结果表明:投影寻踪方法可以应用于污水厂运营绩效的评价;最佳投影方向客观地反映了各指标值所占权重,在对运营绩效较差的污水厂进行改进时,应首先考虑具有较大投影方向值即较大权重的指标;投影特征值的大小反映了污水厂的综合运营绩效,特征值越大其运营绩效越好,反之越差。研究结果有助于污水厂的自我诊断以及监管单位比较不同污水厂的运营绩效,对于排名在前的污水厂,可给予适当的激励,对于排名靠后的污水厂,应该查明原因以加强管理,促使其逐步改善运营绩效。

Evaluating the operation performance of wastewater treatment plant （WWTP） by single index is difficult, due to a wide range and diversity of the evaluation indexes. Based on the foundation of operation performance index system of WWTP, genetic algorithm and projection pursuit model were used to project muhi-dimensional evaluation indexes to one-dimensional data space, and horizontal comparison of complex systems was achieved. In order to prove the validity, the results of principal component analysis and projection pursuit were compared. The results showed that： ① Projection pursuit could be applied to the evaluation of WWTP operation performance; ② Optimum projection direction objectively reflected the weight of each index value. An index having a larger weight value should be considered when impro- ving the poor operation performance of a WWTP ; ③ The magnitude of the projection eigenvalue reflected the comprehensive operation performance, the larger the eigenvalue, the better the operation performance. The research results can contribute to the continuous self-improvement of WWTP and compare operation performance of different WWTPs. Top-ranked WWTPs could find appropriate encouragement using the indexes while lower-ranked WWTPs should identify the reasons to strengthen management and improve operation performance.