基于耦合的新船价格波动影响因素分析

针对新船价格波动机理,通过分析船舶市场运行过程中影响新船价格的各种因素所占权重,以及这些因素作用下最终形成市场价格的过程,研究新船价格波动背后的形成过程和作用体系.首先,为解决多元数据求极值问题,提出动态粒子平衡的均匀采样方法,并在高维空间的单位超球面上生成权重方向向量,得到映射到不同方向向量上的投影序列.其次,为消除不平稳、不规则数据的影响,运用经验模态分解方法对投影序列进行处理,生成不同度量尺度下的平稳序列.再次,考虑到随机抽样的离散特性,以本征模函数与目标序列的判定系数为优化目标,加入了遗传算法得到在连续超球面上的优化结果.最后,通过对新船价格及其各相关因素的历史数据进行分析,得出拟合度较高的曲线及其所对应的各因素所占权重.分析表明该方法能够有效解决多元序列求极值和多元非平稳序列的相关性分析问题,能够综合考虑更多因素描述与新船价格的波动机理,为以后船舶价格预测提供一种新的方法.

Analyzing the influencing factors of new ship price fluctuation based on coupling

Aiming at the new ship price fluctuating mechanism, by analyzing the weight and way of factors that influence the volatility of ship price, this article researched the formation process of new ship price volatility and its interaction system. Firstly, in order to solve the extreme value problem of multivariate data, a uniform sampling method was put forward based on Newton's law of motion and corresponding weight vectors were generated on a unit sphere in the high-dimensional space, then time series were translated into projection sequences through mapping on different weight vectors. Secondly, this paper applied empirical mode decomposition (EMD) to analyze projection sequences and generated stationary series in different measurement scales, which reduced the effects of non-stationary and irregular data. Thirdly, considering the discrete features of random sampling, R-square of intrinsic mode functions (IMFs) was chosen as optimization goal, which was solved on continuous unit sphere by applying genetic algorithm. Finally, through analysis on the multivariate correlation between new ship price and its related factors, a good fitting curve was drawn while the weights of factors were got. The result showed that this method can effectively solve the extreme value problem of multivariate sequence and correlation analysis problem of multiple nonstationary sequences. It considered more associated factors to describe the curve of new shipbuilding price, and provided a new method for the price prediction of ships.