受环境影响的虚拟树木生长预测与仿真

Prediction and Simulation of Virtual Tree Growth Influenced by the Environment

虚拟树木建模并对其生长过程进行模拟是个难题。本文提出一种不需要设定特殊规则,而且不需要迭代运算的基于树型结构的建模方式,并通过循环神经网络RNN(Recurrent Neural Network)生成仿真序列。由于循环神经网络处理长时间依赖效果并不理想,本文采用GRU(Gated Recurrent Unit)单元模型来弥补循环神经网络的不足。实验表明,本文提出的方法可以模拟树木受到外界环境影响后的生长过程,并且对于复杂模型能通过并行处理进一步提高仿真速度,达到了预期效果。

Modeling and simulating the growth process of virtual trees is a difficult problem.This paper proposes a modeling method based on the tree structure with no requirements of special rules or iterative computation,which can generate the simulation sequence through RNN（Recurrent Neural Network）.As RNN is not efficient enough to handle the long-time dependence,this paper adopts GRU（Gated Recurrent Unit） unit model to make up for the deficiency of RNN.The experiment results show that the proposed method can simulate the growth process of trees under the influence of the external environment,and further improve the simulation speed of complicated models through parallel processing.