The lower heating value (LHV, also called net calorific value, NCV) of the coal entering the furnace is one of the most important variables for process optimization. However, in most cases, it is not online measurable in coal-fired power plants. In this paper, an online identification approach of the LHV is proposed based on the boiler-side whole process models together with the real-time data acquired by the distributed control system (DCS). The boiler-side whole process models cover typical facilities of a power plant, and are established based on dynamic mass and energy balances. The accurate estimation of the mass flow rate of the coal entering the furnace and the total output energy of the boiler is the highlight of the approach. Rolling identification of the LHV of the coal is presented to take the intrinsic time variant characteristics of the coal combustion and heat transfer processes into account. To validate the proposed approach, pseudo-online simulations are carried out with the history data of a real 300 MW coal-fired power plant. The mean absolute relative error between the calculated LHVs and their original assays is found less than 5.0%, and the mean absolute relative error becomes less than 1.1%, if the rectified assays are used. (C) 2015 Elsevier Ltd. All rights reserved.