基于欠定盲源分离和Copula统计的船舶电力系统健康评估建模研究

Aiming to the key technologies of initial value generates and optimization mechanism for artificial bee colony optimization algorithm, chaotic maps was used to generate the initial value of optimization parameters, and the improved optimization mechanism was also proposed. On the above basis, a method of underdetermined blind source separation with the order based on artificial bee colony optimization mechanism was constructed; it will eliminate the order ambiguity of separated signals. The health assessment model of marine electric power system based on underdetermined blind source separation and Copula statistics was emphatically studied. Firstly, application underdetermined blind source separation in data acquisition of marine electric power system work monitoring, it can evaluate or predict the health situation of marine electric power system by combining with the organics of sate assessment and prediction based on Copula statistics theory, and then enable ship can be repaired in the best berthing condition and working environment. Research works of this project were validated by using theory analysis, software simulation, platform experiment and entity ship measured study. At present, both home and abroad are committed to intensify efforts to develop integrated power system for ship, the electric propulsion system substitute for diesel engine propulsion system and it become main shipboard propulsion system. However, the normal power system is the premise of that ship safely driving at sea. The contribution of our work will improve ship technical performance, and it also meets the need which China step into strong maritime national.

首先针对人工蜂群优化算法的初值产生和寻优机理关键技术，提出混沌映射技术产生初值法和相应改进的寻优机理；在此基础上，建立基于人工蜂群优化机理的欠定盲源有序分离技术，以消除源信号分离次序的模糊性；着重研究基于欠定盲源分离技术和Copula统计的船舶电力系统健康评估模型，它首先将欠定盲源分离技术应用于船舶电力系统工作监测数据采集,然后与基于Copula统计理论的状态评估和预测机理有机相结合来评估或预测船舶电力系统健康情况，使船舶能在良好停泊条件或工作环境下进行维护；该项目采用理论分析、软件仿真、平台实验和实体船舶实测进行验证研究。 目前，国内外都致力于加大"综合电力系统"型船舶的开发力度，电力推进系统挑战了柴油机动力的统治地位，成为船舶推进的主动力系统。正常工作的电力系统是船舶在海上安全行驶的前提，因此本项目的研究对提高船舶技术具有重大的理论意义和应用价值，也符合我国建设海洋强国国策的的需求

近年来，电力推进系统挑战了柴油机动力的统治地位，成为船舶推进的主动力系统，而正常工作的电力系统保证船舶在海上安全行驶的前提。然而，复杂的海洋和天气多变的恶劣环境使得船舶电力系统产生故障是不可避免的，因此及时给出电力系统的健康状态信息，可以有效地避免故障的发生，它也是保证船舶电力系统及推进系统长期不间断有效稳定运行的关键。本项目围绕船舶电力系统健康评估建模的研究目标，开展如下方面的研究：(1) 提出了一种改进的人蜂群优化算法，该算法利用Logistics混沌映射技术来产生待优化参数的初始值，还根据目前蜂群成员的寻优质量采用两种探索方案。(2) 为了消除源信号分离次序的模糊性,提出了一种基于人工蜂群优化机理的欠定盲源有序分离方法,该方法能按差分峰度的绝对值的降序分离源信号。(3) 研究一种基于欠定盲源离技术的船舶电力系统工作监测数据采集方法，它可以解决船舶机舱空间有限与采集电力系统工作数据需要装置过多传感器之间的矛盾。(4) 根据电力系统工作数据的统计特性，建构了一种基于Copula理论的船舶电力系统健康状态评估模型；根据电力系统健康演变的动力学行为特征，建构了基于动力学理论的船舶电力系统健康状态评估的初步构思。它们都可以用于评估船舶电力系统当前处于哪一种工作状态。