Bioaerosol particles are widely spread in the air， and high concentrations of bioaerosols are extremely hazardous to human health. In order to realize the early warning prediction of atmospheric bioaerosol concentration， this paper takes fluorescence lidar as the detection tool， and on the basis of obtaining the bioaerosol concentration contour， combines the relevant parameters of the atmospheric environment， optimizes the support vector machine （SVM） by using the particle swarm （PSO） and the genetic algorithm （GA） and establishes a prediction model for the bioaerosol concentration contour， by adopting the temperature， humidity， PM2.5， PM10， CO₂， SO₂， NO₂， O₃， wind speed and other related parameters as inputs and bioaerosol concentration contour data as outputs for training， determining the parameter configuration of the prediction model， and reintroducing new atmospheric parameters， predicting the bioaerosol concentration contours by using the trained model and comparing it with that detected by fluorescence lidar， and analyzing the results of the bioaerosol concentration contours predicted by the optimized model with different algorithms. The bioaerosol concentrations predicted by the optimized model and their relative errors are also analyzed.