PSO-无偏灰色马尔科夫模型在船舶交通流量预测中的应用

为提高船舶交通流量的预测精度,利用具有全局搜索能力的粒子群算法(Particle Swarm Optimization, PSO)对无偏灰色马尔科夫模型进行优化,构建船舶交通流量预测的PSO-无偏灰色马尔科夫模型。该模型可综合考虑预测中的残差序列、状态区间、状态转移概率,自适应选取最优白化系数,用以准确表征船舶交通流量的发展趋势。以东营港2012—2017年船舶交通流量季度统计数据为例,选取前21个季度数据对模型进行训练,后2个季度数据对预测结果进行分析,与传统的无偏灰色模型和无偏灰色马尔科夫模型相比,该模型能显著地提高船舶交通流量的预测精度,其拟合精度和预测精度分别为91.439%和95.959%,验证后该模型具有科学性与有效性。

Application of PSO-Unbiased Grey Markov Model in Ship Traffic Flow Prediction

Particle Swarm Optimization (PSO) with global search capability is used to optimize the unbiased gray Markov model. The improved model can synthetically consider the residual sequence, the state interval and the state transition probabilities in the prediction, and select the optimal whitening coefficient adaptively to accurately characterize the development trend of the ship traffic flow. The quarterly statistical data of ship traffic flow from 2012 to 2017 in Dongying Port are taken to demonstrate the process. The data of the first 21 quarters are selected to train the model and the data of the latter two quarters are used for checking the forecast results. In comparison with the traditional unbiased gray model and the unbiased gray Markov model, the proposed model shows significant improvement of the prediction accuracy. The fitting accuracy and the prediction accuracy reach 91.439% and 95.959%, respectively, proving the validity of the model.