基于惯性姿态匹配的自适应滤波器构建及船体变形测量方法研究

In modern large ships, some weapon systems & sensors or measuring equipment are equipped. Among combat units or measuring units, a unified attitude reference which can help improve working efficiency of whole ship must be constructed, and ship hull deformation measurement is key to keep attitude reference unified. Inertial attitude matching method based on laser gyroscope units is very important in military application in measuring ship hull deformation, with some advantages of high precision in real time dynamically and flexible installation without optical visibility. .Considering some problems such as dimension selection of ship hull deformation filtering model, adaptive identification of model parameters, regression and bias error suppression of filtering estimation, the project mainly systematically studies three key scientific problems, which are filtering model dimensionality reduction by using prediction divergence criterion estimation, adaptive identification of filtering model parameters based on generalized method of wavelet moments, inertial attitude matching adaptive-filtering structure based on residual resampling technique..The feature of the project is systematically studying ship hull deformation measurement based on inertial attitude matching adaptive filter, which is more perfect with enough theoretical principal and is probably implemented with high precision and autonomous measurement in engineering application.

现代大型舰船上各种武器装备和测量设备之间必须建立统一的姿态基准才能发挥整船最大的工作效能，船体变形测量是实现姿态基准统一的关键。基于激光陀螺组合体的惯性姿态匹配测量法可以对船体变形直接进行高精度的动态实时测量，具有不需要光学通视、安装灵活和环境适应性好的优点，对于提高整船的作战效能具有重要的军事意义。针对惯性姿态匹配测量中存在船体变形滤波模型的阶数及其参数的自适应辨识、滤波估计的退化及偏置误差的抑制等问题，本项目将针对：（1）基于预测偏差估计准则的船体变形滤波模型阶数选择；（2）基于广义小波矩估计的船体变形滤波模型参数自适应辨识；（3）基于残差补偿重采样的惯性姿态匹配自适应滤波器构建等关键科学问题进行研究。本项目的特色是完整地构建出基于惯性姿态匹配的自适应滤波器及其船体变形测量方法，为实现高精度的自主变形测量和惯性姿态匹配测量法的成果转化及工程应用提供充分的理论基础和技术支持。

现代大型舰船上各种武器装备和测量设备之间必须建立统一的姿态基准才能发挥整船最大的工作效能，船体变形测量是实现姿态基准统一的关键；基于激光陀螺组合体的惯性姿态匹配测量法可以对船体变形直接进行高精度的动态实时测量，具有不需要光学通视、安装灵活和环境适应性好的优点，对于提高整船的作战效能具有重要的军事意义。针对惯性姿态匹配测量中存在船体变形滤波模型的阶数及其参数的自适应辨识、滤波估计的退化及偏置误差的抑制等问题，本项目在①基于预测偏差估计准则的船体变形滤波模型阶数选择及检验、②基于广义小波矩估计的船体变形滤波模型参数的自适应辨识、③基于残差补偿重采样的惯性姿态匹配自适应滤波器的构建及试验验证等三个方面进行了深入研究；取得了惯性姿态匹配卡尔曼滤波器的框架设计、卡尔曼滤波中状态方程参数的辨识算法、卡尔曼滤波中观测方程的残差补偿重采样设计、惯性测量匹配法的船体变形测量精度水平的粗略估计方法等四个方面的重要结果；积累了“远望三号”测量船某航次的搭载试验实测数据集、根据光学变形测量数据生成的各类含误差陀螺数据、广义小波矩估计源代码包的测试数据等关键数据。本项目的科学意义是完善了惯性量匹配测量的船体变形滤波器构建理论，表现为：通过引入鲁棒广义小波矩估计方法于惯性姿态匹配的卡尔曼滤波器中，使得双高斯-马尔可夫模型的参数辨识统一在小波方差框架中，具备鲁棒性和自适应调整尺度权重的优点；通过引入残差补偿重采样技术于惯性姿态匹配的卡尔曼滤波器中，可以避免双高斯-马尔可夫模型的参数辨识时出现退化。本项目的应用前景是通过采用广义小波矩估计+残差补偿重采样结合的方法，使得双高斯-马尔可夫模型形式的惯性姿态匹配卡尔曼滤波器具备自主、实时测量船体变形的应用价值，预计为大型舰船的传递对准技术提供方案参考和技术支持。