Abstract: The timely and effective detection of f-CaO content in cement clinker plays an important role in the control and optimization of cement clinker performance.Presently, the traditional method of manual sampling and testing in the laboratory is mostly used in cement plants to detect the clinker f-CaO content.However, the measurement results have a large lag.To address the problem of lagging results of traditional f-CaO content detection, this paper developed a model for f-CaO content prediction using the cascade forest algorithm based on the site data of cement production.First, the fourteen model parameters, for example, raw material feeding rate, decomposing furnace outlet temperature, rotary kiln current, etc.were selected according to the production process of cement clinker and the formation causes of clinker f-CaO production, and then the time series were constructed as model inputs using sliding time windows to include more improved time-series information.Finally, the proposed prediction method was compared with three classical machine learning models, namely support vector machine regression(SVR), k-nearest neighbor (KNN),random forest(RF).Results show that the built method can provide better prediction accuracy and generalization capability in predicting the f-CaO content of cement clinker.