摘要
针对海洋动态目标的搜索问题,引入具有复杂测量特性的侧扫声呐探测概率模型,利用更加灵活的长时域模型预测控制策略规划自主式水下航行器的搜索航迹。首先,构建了真实的侧扫声呐探测概率模型,引入Sigmoid函数并进一步定义二次探测概率,另外考虑了海底起伏地形对探测事件造成的影响。其次,基于构建出的侧扫声呐探测概率模型,利用Chapman-Kolmogorov方程和贝叶斯公式实时预测更新探测事件发生后的动态目标概率图。最后,将大范围预期收益引入传统的模型预测控制方法(MPC),提出长时域模型预测控制方法(FMPC),有效提高搜索效率。仿真实验表明,所提的FMPC方法相比于传统MPC方法在搜索收益方面提高50%,具有更高的搜索效率。
In this paper, a detection probability model of side-scan sonar with complex measurement characteristics is introduced to search for dynamic target in the ocean, and a more flexible future-dependent model predictive control strategy is used to plan the search trajectory of autonomous underwater vehicle(AUV). Firstly, the real detection probability model of side-scan sonar is constructed. The Sigmoid function is introduced, and the second detection probability is further defined. In addition, the influence of seafloor topography on detection event is considered. Secondly, based on the detection probability model of side-scan sonar, the Chapman-Kolmogorov equation and Bayes formula are used to predict and update the dynamic target probability map after the detection event. Finally, a future-dependent model predictive control method(FMPC) with a wide range of expected reward is proposed to solve the local optimal problem in MPC. Simulation results show that the proposed FMPC method has a higher search efficiency than the traditional MPC method, and the search reward is improved by 50%.
作者
姚鹏
邱立艳
YAO Peng;QIU Liyan(College of Engineering,Ocean University of China,Qingdao 266100,China)
出处
《无人系统技术》
2022年第6期48-56,共9页
Unmanned Systems Technology
基金
国家自然科学基金(51909252)。
关键词
自主式水下航行器
动态目标搜索
侧扫声呐
探测概率模型
长时域模型预测控制
Autonomous Underwater Vehicle(AUV)
Dynamic Target Search
Side-scan Sonar
Detection Probability Model
Future-dependent Model Predictive Control(FMPC)
