The Research On Multi-Objective Optimization Of Forest Spatial Structure Based On GAPSO

The health and stability of forest is closely related to the forest space structure, there is important theoretical and practical significance on the study of forest structural optimization which is based on intelligent information processing technology space for our country’s sustainable management of forests and "two types" of social development. The forest spatial structure involves many aspects, the multiple objectives of forest spatial structure optimization conflict with each other and have complicated constraint conditions, it is actually a nonlinear multi-objective programming problem, so conventional mathematical programming methods have greater limitations and is difficult to achieve good effect. Particle swarm optimization (PSO) and Genetic Algorithm (GA) are global optimization algorithms which have been developed in recent years. Particle swarm algorithm uses the past experiences of individual particles in the groups and experiences of other particles to get valid information, for most optimization problems, PSO has a faster convergence speed and needs fewer parameters to set, while it is not so good in distribution and convergence of the solution set, so less involved in the practical application.GA is widely used in all kinds of complex optimization problem because of it’s implicit parallelism and high robustness.This article puts on an improved genetic, hybrid multi-objective particle swarm optimization algorithm (MO-GAPSO) solving multi-objective optimization problem on the basis of a comprehensive analysis of the spatial structure of the forest. Testing the performance of the proposed algorithm and simulating actual stand spatial structure optimization through classical optimal performance function, the paper verifies the feasibility and effectiveness of the improved hybrid algorithm. The main research work and research conclusion includes:(1) The paper introduces7expression factors of the forest spatial structure about Forest stand scale based on the research:mixed degree, competition index, relative Angle scale, forest layer, the density of space index, open ratio and the size ratio of the number, setting up the objective function and constraint conditions of forest spatial structure optimization, establishing the basic mathematical model of multi-objective forest spatial structure optimization; (2)This paper treats each tree’s location, mixed degree, competition index, relative Angle scale, forest layer, the density of space index, open ratio and the size ratio of the number in the stand space as the genes mapped to chromosome. As a result, each tree can be seen as an independent chromosome, the paper sets the spatial structure optimization of multi-objective function to the fitness function of genetic algorithm and builds a forest ideal spatial structure based on MO-GA optimization model, then makes the multi-objective optimization problem of forest spatial structure into the population evolutionary genetic optimization problem. And verified by the examples:the model based on MO-GA can effectively find out the ideal space structure of the forest;(3)The paper introduces genetic algorithm mechanism in PSO algorithm, a mathematical model is proposed to improve the basic algorithm MO-GAPSO forest spatial structure of multi-objective optimization. Every tree in the forest stand can be seen as a solution of PSO solution space, the entire stand space is mapped to the target particle solution space, the trees’s space coordinates in the stand mapped to the positions of ’particles’ in the solution space, the paper sets fitness function about particle populations through multiple objective functions of stand space, establishes an optimization model for forest spatial structure based on the MO-GAPSO, makes Multi-objective problem about positioning forest spatial structure of the weak links into iterative optimization problems on particle swarm. The results show that, the improved algorithm can find out the global optimal solution rapidly and the model can accurately locate the target tree which affects the overall spatial structure;(4) Using Rosenbrock function and Rastrigin function which are used to test algorithms’ performance on optimizing, to test the performance of the improved MO-GAPSO algorithm, and make comparative analysis with the standard particle swarm algorithm. Verified by the experimental result:The improved PSOGA algorithm’s searching accuracy is improved by more than40%, convergence rate increased by more than20%, fitness function curve is smoother and the minimum is smaller, The algorithm’s ability to get rid of local optimal is improved, its performance is superior to the standard particle swarm optimization (pso) algorithm.