Abstract: The quality of coke has a great impact on the production of blast furnace smelting.The establishment of coke quality prediction model with high accuracy and good adaptability is of great significance to enterprise production, in order to solve the problem that it is difficult to measure coke quality in real time, a coke quality prediction model based on chaotic sparrow search algorithm(TSSA) and optimized support vector regression(SVR) is proposed.Firstly, the improved Tent chaotic map is used to initialize the population and strengthen the global search ability of sparrow search algorithm(SSA), and then TSSA model is used to optimize the parameters of SVR model, which effectively overcomes the problem of parameter selection of traditional SVR model.Seven indexes such as moisture, ash and volatile in blended coal are selected as the input of the model, and the four indexes of coke quality, including crushing strength, wear resistance, reactivity index and strength after reaction, are selected as the output of the model, the TSSA-SVR model is verified according to the historical production data of coking plant, and compared with SSA-SVR and SVR models.The experimental results show that the proposed method has good accuracy and adaptability, and has certain practical value for coke production.