Projection pursuit model for evaluating drought based on improved artificial fish swarm algorithm of Sanjiang Plain

Projection pursuit model for comprehensive evaluation of drought was established by using artificial fish swarm. It was superior to the existing methods which might cause many problems such as discrete evaluation grade and barely recognizable results. Artificial fish swarm algorithm (AF) was introduced to optimize the function and seek the optimum projection vector, and it was improved by using the adaptive artificial fish step and crowded degree factor. Compared to the previous methods, it overcame the shortcomings which were easily to fall into local extreme optimum and improved the global search ability and convergence speed. Taking Hongxinglong Branch Bureau of Sanjiang Plain as the research objects, the departure ratio of rain-fall, Z index and homogenization of rain-temperature were selected to establish drought assessment model based on above methods. The results showed that the model could avoid the incompatibility of single indexes. The improved projection pursuit method is feasible to evaluate the drought.