In underwater target search path planning,the accuracy of sonar models directly dictates the accurate assessment of search coverage.In contrast to physics-informed sonar models,traditional geometric sonar models fail ...In underwater target search path planning,the accuracy of sonar models directly dictates the accurate assessment of search coverage.In contrast to physics-informed sonar models,traditional geometric sonar models fail to accurately characterize the complex influence of marine environments.To overcome these challenges,we propose an acoustic physics-informed intelligent path planning framework for underwater target search,integrating three core modules:The acoustic-physical modeling module adopts 3D ray-tracing theory and the active sonar equation to construct a physics-driven sonar detection model,explicitly accounting for environmental factors that influence sonar performance across heterogeneous spaces.The hybrid parallel computing module adopts a message passing interface(MPI)/open multi-processing(Open MP)hybrid strategy for large-scale acoustic simulations,combining computational domain decomposition and physics-intensive task acceleration.The search path optimization module adopts the covariance matrix adaptation evolution algorithm to solve continuous optimization problems of heading angles,which ensures maximum search coverage for targets.Largescale experiments conducted in the Pacific and Atlantic Oceans demonstrate the framework's effectiveness:(1)Precise capture of sonar detection range variations from 5.45 km to 50 km in heterogeneous marine environments.(2)Significant speedup of 453.43×for acoustic physics modeling through hybrid parallelization.(3)Notable improvements of 7.23%in detection coverage and 15.86%reduction in optimization time compared to the optimal baseline method.The framework provides a robust solution for underwater search missions in complex marine environments.展开更多
针对水下目标方位(Direction of Arrival,DOA)估计准确性实时性的要求,理论分析了互质阵列模型、压缩感知DOA估计的原理,设计实现了基于FPGA的互质阵列压缩感知算法DOA估计系统。首先介绍了系统开发环境,包括平台选择、开发流程等;其次...针对水下目标方位(Direction of Arrival,DOA)估计准确性实时性的要求,理论分析了互质阵列模型、压缩感知DOA估计的原理,设计实现了基于FPGA的互质阵列压缩感知算法DOA估计系统。首先介绍了系统开发环境,包括平台选择、开发流程等;其次,介绍了硬件系统的整体框架,重点说明了PS与PL之间的数据传递流程和硬件各模块实现过程,并仿真验证了该系统的正确性。在Xilinx FPGA平台上进行了湖试数据的处理,完成了数据运算参数的统计收集,验证了DOA估计的有效性,并计算了运算耗时。结果表明,所设计的系统能够正确完成DOA估计并满足实时性要求。展开更多
基金supported by Natural Science Foundation of Hu'nan Province(2024JJ5409)。
文摘In underwater target search path planning,the accuracy of sonar models directly dictates the accurate assessment of search coverage.In contrast to physics-informed sonar models,traditional geometric sonar models fail to accurately characterize the complex influence of marine environments.To overcome these challenges,we propose an acoustic physics-informed intelligent path planning framework for underwater target search,integrating three core modules:The acoustic-physical modeling module adopts 3D ray-tracing theory and the active sonar equation to construct a physics-driven sonar detection model,explicitly accounting for environmental factors that influence sonar performance across heterogeneous spaces.The hybrid parallel computing module adopts a message passing interface(MPI)/open multi-processing(Open MP)hybrid strategy for large-scale acoustic simulations,combining computational domain decomposition and physics-intensive task acceleration.The search path optimization module adopts the covariance matrix adaptation evolution algorithm to solve continuous optimization problems of heading angles,which ensures maximum search coverage for targets.Largescale experiments conducted in the Pacific and Atlantic Oceans demonstrate the framework's effectiveness:(1)Precise capture of sonar detection range variations from 5.45 km to 50 km in heterogeneous marine environments.(2)Significant speedup of 453.43×for acoustic physics modeling through hybrid parallelization.(3)Notable improvements of 7.23%in detection coverage and 15.86%reduction in optimization time compared to the optimal baseline method.The framework provides a robust solution for underwater search missions in complex marine environments.
文摘针对水下目标方位(Direction of Arrival,DOA)估计准确性实时性的要求,理论分析了互质阵列模型、压缩感知DOA估计的原理,设计实现了基于FPGA的互质阵列压缩感知算法DOA估计系统。首先介绍了系统开发环境,包括平台选择、开发流程等;其次,介绍了硬件系统的整体框架,重点说明了PS与PL之间的数据传递流程和硬件各模块实现过程,并仿真验证了该系统的正确性。在Xilinx FPGA平台上进行了湖试数据的处理,完成了数据运算参数的统计收集,验证了DOA估计的有效性,并计算了运算耗时。结果表明,所设计的系统能够正确完成DOA估计并满足实时性要求。
