关键词: 软测量, 化学需氧量（COD）, 污水处理过程

Abstract: Wastewater treatment plant is a complex process affected by many uncertain factors, which leads to frequent fluctuations of operating conditions. An online soft sensor of effluent COD（Chemical Oxygen Demand） is presented to deal with the problem that single model cannot guarantee the accuracy of soft sensor. The soft sensor is composed of simplified mechanistic model and error compensation model, where simplified model based on ASM1（Activated Sludge Model No.1） is used to describe the basic characteristics of dynamic mechanism in wastewater treatment process; error compensation model using integrated linear model is adopted to compensate model errors under different operating conditions. The number of the clusters can be achieved by synchronous clustering algorithm that considers the time interval between input and output data and the relevance of adjacent data. Linear model in error compensation model reduces the computational cost and improves the real-time performance of soft sensor. Simulations show that online hybrid soft sensor of COD has better prediction accuracy under multiple operating conditions.

Key words: soft sensor, Chemical Oxygen Demand（COD）, wastewater treatment process