动态误差谱理论及其在高超声速飞行器高准确度性能评估中的应用研究

Hypersonic vehicle high accuracy performance evaluation is a strategic issue of the hypersonic vehicle design-typification and hypersonic vehicle system test-analysis. However, the existing hypersonic vehicle performance evaluation methods have two major disadvantages (i.e., evaluation results are inaccurate and resources are wasted) in real-time evaluation during a data within the outlier; and thus, the new theory, new model and new method are desired to be presented in solving the above problems. This project involves the dynamic error spectrum theory and its application on the hypersonic vehicle high accuracy performance evaluation. First, we propose several dynamic error spectrum measures to the hypersonic vehicle high accuracy performance evaluation. Second, a new data preprocessing method is proposed to solve the “diverse population” problem of the hypersonic vehicle data based on cloud model. Furthermore, two algorithms, using the above preprocessing data, are presented to overcome the dynamic error spectrum computation problem under low-and high-dimensional data conditions. The first one is the finite mixture model algorithm, which uses the finite mixture model to estimate the error distribution of low-dimensional data for the hypersonic vehicle. The second one is the Dirichlet process Gaussian mixture mode algorithm, which is similar to applying the Dirichlet process Gaussian mixture mode to obtain the hypersonic vehicle data error distribution in the high-dimensional condition. Finally, a high accuracy performance evaluation method is designed to the hypersonic vehicle high accuracy performance evaluation based on the dynamic error spectrum, which utilizes to solve the real-time evaluation problem and the accurate assessment problem of the hypersonic vehicle performance evaluation. Expected results will provide new theory and methods for the hypersonic vehicle high accuracy performance evaluation.

高超声速飞行器高准确度性能评估是其设计定型与试验分析的核心问题。现有的高超声速飞行器性能评估方法在实时地评估含异常值的数据时存在评估结果不准确和资源浪费的问题，亟需研究新的理论、模型和方法解决这些问题。因此，本项目开展动态误差谱理论及其在高超声速飞行器高准确度性能评估中的应用研究。首先研究动态误差谱度量方法。然后针对“异总体”的高超声速飞行器数据，提出基于云模型的高超声速飞行器数据预处理方法。再利用得到的预处理数据，提出基于有限混合模型的动态误差谱算法和基于Dirichlet过程高斯混合模型的动态误差谱算法，解决在低维和高维数据条件下动态误差谱无法计算的难题。最后将动态误差谱用于高超声速飞行器的高准确度性能评估，提出基于动态误差谱的高超声速飞行器高准确度性能评估方法，解决数据中有异常值时高超声速飞行器难以实时准确评估的难题。预期成果将为高超声速飞行器高准确度的性能评估提供新理论及方法。

高超声速飞行器高准确度性能评估是其设计定型与试验分析的核心问题。现有的高超声速飞行器性能评估方法在实时地评估含异常值的数据时存在评估结果不准确和资源浪费的问题，亟需研究新的理论、模型和方法解决这些问题。因此，通过研究动态误差谱理论及其在高超声速飞行器高准确度性能评估中的应用，取得的创新性成果有：首先构建了动态误差谱度量数学模型、体积、区间、面积和指数形式的动态误差谱模型。然后提出了基于排列组合和相关系数的动态误差谱算法，提出了基于t分布混合模型的动态误差谱算法和提出了基于共享子空间和EM算法的动态误差谱算法，解决了低维和高维数据条件下动态误差谱无法计算的难题。进一步，针对“异总体”的高超声速飞行器数据，提出了基于组合贴近度的云模型数据预处理方法。最后将动态误差谱用于高超声速飞行器的高准确度性能评估，提出了基于误差谱理论的高超声速飞行器高准确度性能评估，基于动态误差谱的高超声速飞行器导航精度评估和基于动态误差谱和改进TOPSIS的高超声速飞行器高准确度性能评估方法，解决了数据中有异常值时高超声速飞行器难以实时准确评估的难题。研究成果将为高超声速飞行器高准确度的性能评估提供新理论及方法。