Abstract: The soil heavy metals-digital characteristics have a complex feature,and its pollution evaluation was a clustering problem of high dimensional nonlinear data,therefore,the classification of soil heavy metal pollution can be carried by using PPC model.In the application of PPC evaluation model for soil heavy metals,the best projection direction can directly affect the accuracy of the evaluation results.To obtain a better projection direction,this paper modular calculates the projection direction by bacterial foraging algorithm(BFA)、genetic algorithm(GA)、fish swarm algorithm(FSA)which were applied for the PPC model respectively,and the soil heavy metal pollution evaluation was constructed by using BEA-PPC、GAPPC、FSA-PPC model,respectively.The aforementioned 3models were then applied in a region,to evaluate the agricultural soil heavy metal,the projection direction and the corresponding evaluation results were compared,which show that,the 3evaluation models are applicable in the evaluation of the soil heavy metal pollution;and the projection direction of these three intelligent algorithm are reasonable,moreover the FSA accurately reflects the heavy metal in the evaluation process of the most important degree,followed by GA,BFA.