The invention relates to a thermal power plant intelligent combustion control method based on cloud data and cloud computing, and the method comprises the steps of: firstly, constructing a cloud database, conducting abnormal value processing on input data and output data in the constructed cloud database, and then performing principal component analysis, training a heat exchange surface wall temperature prediction model, a denitration reactor inlet NOx concentration distribution prediction model and a boiler efficiency prediction model by adopting a principal component analysis result, and comparing a training result with data of a model calibration system, and if the root-mean-square error is within 15%, according to the heat exchange surface wall temperature prediction value, the NOx concentration prediction value and the boiler efficiency prediction value provided by an integrated prediction model, conducting calculation through an intelligent operation control module to obtain optimal operation parameters, thus realizing dynamic optimization control over boiler combustion and pollutant generation. The boiler efficiency and the adaptability to coal type changes are improved, the average emission load of NOx is reduced by 10% or above, and power generation benefits are maximized.