Improve Apso-of Grnn-based Regional Logistics Demand Prediction

With the development of economic integration, logistics is increasingly important in optimized the allocation of resources,.Regional logistics are the main components of regional economic factors.From rationalization of a single enterprise logistics to rationalization of regional logistics is an inevitable trend of socio-economic development. Therefore, it's important to plan and set up a comprehensive network of regional logistics of, and forecasting research of regional logistics demand is one of an important aspect.Because traditional logistics demand forecasting methods unrecognized logistics capacity and factors of forecasting's highly non-linear, some vague uncertainties can not be processed, resulting in serious distortion of the results of forecasting, This paper is based on generalized regression neural network(Generalized Regression Neural Network,GRNN) to build a forecasting model, generalized regression neural network have good generalization, it has strong advantages in the approximation ability, classification ability and study speed, and with strong ability of non-linear fitting etc., it's suitable for the forecasting analysis. but the only adjustment parameters - smoothing factor is difficult to determine, smoothing factor's values have a great impact on pridiction of the network.Therefore, this paper presents an improved adaptive particle swarm optimization (Adaptive Particle Swarm Optimization,APSO) to determine the smoothing factor, smoothing factor is mapping for the particles,the algorithm based on the characteristics that the fitness value of particles is equivalent to location of particles, it's adaptive to adjust the inertia weight by comparing the fitness value of particle with the current value of the global optimum, and according to the average distance of current population,take part of the population to be mutation for overcoming the premature convergence, and find the smoothing factor of global optimum,so that determine the general regression neural network model.Through the empirical study proved that the APSO-GRNN prediction model which this article constructs is suitable to forecast regional logistics demand.