基于改进人工蜂群算法的大青杨晚材率预测

为了更加准确地预测人工林大青杨(Populus ussuriensis)晚材率,通过对标准人工蜂群算法(artificial bee colony, ABC)的蜜源更新公式进行改进,提出了分段式蜜源搜索半径公式,并用改进的人工蜂群算法(AABC)对径向基(radial basis function, RBF)神经网络的初始参数进行优化,提出一种基于改进的人工蜂群算法和径向基神经网络算法结合的预测模型,并与粒子群(partical swarm optimization, PSO)优化的RBF神经网络预测结果进行对比。结果表明:传统的RBF预测模型不仅收敛速度慢,而且预测精度不高。基于改进的ABC算法优化RBF神经网络预测模型整体比PSO优化的效果相对较好,收敛速度从42步提升至28步,预测的平均相对误差从2.54%降低到0.95%。可见对ABC算法的改进是可行的,而且提高了晚材率预测的精度。

Prediction of Latewood Rate of Populus ussuriensis Based on Improved Artificial Bee Colony Algorithm

Abstract: In order to predict the latewood rate of Populus ussuriensis in plantations more accurately, the nectar source update formula of the standard artificial bee colony algorithm (ABC) was improved. Additionally, the segmented nectar source search radius formula was proposed, and the improved artificial bee Swarm algorithm (AABC) optimized the initial parameters of the radial basis function (RBF) neural network and proposed a prediction model based on the combination of improved artificial bee colony algorithm and radial basis function neural network algorithm, and optimizes it with particle swarm (PSO) The prediction results of the RBF neural network were compared. The results show that the traditional RBF prediction model not only converges slowly, but also has low prediction accuracy. Based on the improved ABC algorithm to optimize the RBF neural network prediction model, the overall effect is relatively better than that of the PSO optimization. The convergence speed is increased from 42 steps to 28 steps, and the average relative error of prediction is reduced from 2.54% to 0.95%. It can be seen that the improvement of the ABC algorithm is feasible, and the accuracy of latewood rate prediction is improved.