Due to the complexity of coal type and the large fluctuations of coal calorific value in power plant， the calculation of positive balance coal consumption has difficulty in accuracy. Therefore， by using the corresponding relation between load and the amount of design coal in design data， as well as the actual load， the actual amount of coal， the design lower calorific value of coal， the prediction model of coal’s calorific value was built based on the technical indicators of coal-fired boiler. The corrective loop for the real-time calorific value of burning coal was increased into the coordinated control system （coordinated control system， CCS） of units， and the real-time calorific value of burning coal was used for the regulation of steam temperature and pressure， which could effectively improve the performance of automatic generation control （AGC）. To the problem that the target coal amount of run back （RB） could not completely correspond with load when the calorific value of burning coal changed greatly， the real-time calorific value of burning coal was used before RB action to correct the target coal amount of RB， which ensureed that the target coal amount of RB in all conditions could meet the changes of coal quality， and could correspond with the target load of RB and improve the operation accuracy of RB. During the real-time computing process of positive balance coal consumption in energy loss differential analysis system， the test value was replaced by coal’s real-time calorific value measured in soft-sensing， which could greatly improve the accuracy of coal consumption calculation. At present， the project has been widely promoted in all thermal power enterprise of DatangInternational system.