Abstract:

The difficulties and methods of unmanned aerial vehicle radar detection technology are studied. It analyzes the model and algorithms of Quantum multi-Pattern Recognition Network（QPRN）. By Grover introducing algorithm optimization theory, Phase Rotation Quantum Multi-Pattern Recognition Algorithm（PRQMPRA） is proposed. The optimization algorithm avoids the defect that both phase rotations are π in the Redundancy Quantum Multi-Pattern Recognition Algorithm（RQMPRA）, which will lead to a decrease in the probability of successful search. Three types of data sets are used to analyze the pattern recognition ability of Error Back Propagation Algorithm（EBPA）, Cross-entropy function-Deep Autoencoder learning Algorithm（CDAA）, RQMPRA and PRQMPRA. In the case of determining the limit error, the results show that PRQMPRA has higher recognition rate and relatively faster operation speed. A multi-pattern recognition algorithm based radar target detection method is proposed to study the target detection problem by pattern classification. Using the above four algorithms for UAV target detection experiments, the results show that PRQMPRA has higher detection accuracy and can maintain a higher discovery probability in the case of low Signal to Noise Ratio（SNR）.

Key words: target detection, Unmanned Aerial Vehicle（UAV）, quantum computing, pattern recognition

摘要：

研究了现阶段无人机雷达探测技术的难点与方法，分析了量子多模式识别网络模型与算法，根据Grover算法优化理论，提出了基于相位旋转的量子多模式识别算法（PRQMPRA）。优化算法避免了在带冗余项的量子多模式识别算法（RQMPRA）中两个相位旋转均为[π]会导致搜索成功概率降低的缺陷。利用三种数据集对误差反向传播算法（EBPA）、基于交叉熵函数的深层自编码器学习算法（CDAA）以及RQMPRA与PRQMPRA进行模式识别能力分析，结果表明在确定限定误差的情况下PRQMPRA具有更高的识别率与相对较快的运算速度。提出了一种基于量子多模式识别算法的雷达目标检测方法，通过模式分类的方法研究目标检测问题。利用上述四种算法进行无人机目标检测实验，研究结果表明PRQMPRA具有更高的检测精度，在低信噪比的情况下可保持较高的发现概率。

关键词: 目标检测, 无人机, 量子计算, 模式识别