Abstract: The trajectory distribution patterns of pilot operation gesture are learned based on an improved self-organizing neural network model. Wilcoxon rank sum test and the edit distance technology are introduced to determine the matches in the internal net neighborhood, and the cross-validation technique is used to get the threshold for abnormal detection. Using the learned patterns, this paper considers both local and global anomaly detection as well as object behavior prediction. Experimental results demonstrate the effectiveness of this approach.

Key words: pilot gesture, hierarchical self-organizing neural network, anomaly detection, behavior prediction, Wilcoxon rank, cross-validation

摘要： 为准确学习飞行员操作手势的轨迹分布模式，提出一种改进的层次自组织映射方法。引入 秩和检验技术，结合编辑距离判断内部网的匹配程度，通过交叉验证使验证集获得误差最小，从而自适应取得判断异常的阈值。根据训练得到的轨迹分布模式检测操作过程中的局部异常，判断运动轨迹所表示的事件是否为异常事件，并预测手势将来行为轨迹。实验结果验证了改进方法的有效性。

关键词: 飞行员手势, 层次自组织神经网络, 异常检测, 行为预测, 秩, 交叉验证

CLC Number: 

• TP18