摘要
针对信号采集卫星系统中,采集目标的优先级和携带的信息量在星地资源调度过程中不断随时间变化的特点,提出了一种基于二次映射编码的动态调度差分演化算法(QMDSDE)。算法将采集目标的优先级和携带的信息量进行加权,作为采集的收益。在迭代求解时,根据卫星可见时间窗口的周期性特点设置固定的检测频率,以当前时刻所有目标收益的平均值和敏感系数的乘积作为检测环境变化的阈值。当检测算子的值大于阈值时,算法通过前向预测模型动态调整搜索方向,寻找问题的实时Pareto最优解。算法采用二次映射编码消除星地资源组合中的无效解,降低了算法的搜索空间,提高了算法的求解速度。经仿真校验,该方法在信号采集过程中,可以得到收益较好的星地资源调度方案,且具备较快的收敛速度。
A dynamic schedule differential evolutionary algorithm based on quadratic-mapping coding( QMDSDE) is proposed in this paper for the schedule of onboard and ground resources in signal collection satellite system with changing collecting targets in priority and amount of information. It collects and weights the targets ' priority and amount of information synthetically as their income. In the iteration process,it sets fixed detection frequency based on the periodicity of satellites' time-windows. And it regards the product of the average income of all targets and sensitivity coefficient as threshold to detect whether the environment has changed. When the value of detection-operator is larger than the threshold,algorithm dynamically adjusts the searching direction using forward-prediction model to seek real-time Pareto optimal solutions. Quadratic-mapping code eliminates useless solutions in combination of onboard and ground resources,which will greatly reduce the searching space and improve the solving speed of the algorithm. Simulations have demonstrated that the method proposed can receive onboard and ground resources schedule plans with cracking income and quick converge.
出处
《宇航学报》
EI
CAS
CSCD
北大核心
2016年第3期348-356,共9页
Journal of Astronautics
基金
国家自然科学基金(41571403
61472375)
关键词
信号采集卫星系统
检测频率
前向预测模型
实时Pareto最优解
二次映射编码
Signal collection satellite system
Detection frequency
Forward-prediction model
Real-time Pareto optimal solutions
Quadratic-mapping code