The practicability and operability of online calculating model of boiler efficiency is highly dependent on the composition and type of fuel. In order to reduce the complexity and diversity of coal quality, the analysis error of on-site coal quality and the influence of the inaccuracy of artificial offline input parameters on the boiler thermal efficiency online calculating, a virtual coal quality database which can be used for online calculation of thermal efficiency of coal-fired boiler was proposed. Through the linear regression of the compositions, sources, and types of coal commonly used in industrial boilers and the use of mathematical algorithms such as statistical analysis and cluster analysis, a virtual coal quality database was constructed. In order to verify the applicability of the coal quality database, the calculated calorific value of the coal was compared and analyzed with the help of the neural network algorithm. The results show that the error is within the measurement error range of the industrial coal. The built coal-fired database can effectively realize the online calculation of the thermal efficiency of the boiler when is operating under variable conditions.