面向武器精度一致性评估的等效试验设计与建模方法

Equivalent tests are introduced as assisting techniques in high-performance complex equipment system due to the restriction of test requirements. However, the main shortage of equivalent tests is that their results are somehow precise for local performance of the system, so generally it is always difficult to convert those results to authentic system performance criterion especially under the environment of actual combat operation. Since most of the existing methods are unfit for consistency evaluation of various equivalent tests under the multidimensional constraint conditions, the project studies the optimal design methodology to optimize the allocation of test resources and achieve the goal of high precision consistency evaluation conclusion in the actual combat condition..The main contents include: (1) We first construct the complete space of the meta-model based on the physical background of the system, where the integrated response surface for different types of equivalent tests embedded; (2) Secondly, we propose the uncertainty metrics with respect to the requirements of consistency evaluation, then perform sensitivity analysis between model parameters and different test types and assess the model prediction ability; (3) Thirdly, we introduce the optimal design method for multi-dimensional constrains in equivalent tests to update and verify the model and parameters, then finally evaluate the model precision under the actual combat operations; (4) Finally, we aim to use our method in the application of consistency evaluation of the guided weapon. We are capable to utilize all kinds of the equivalent tests more efficiently, the conclusion of our project will provide scientific theories and methodologies for integration modelling and global optimal design, which will optimize the evaluation scheme and provide the conclusion of high precision consistency evaluation.

导弹试验受条件限制常引入等效试验，其试验结果难以直接转化为与实战状态一致的系统性能指标。针对多类型等效试验在多维度复杂约束下难以直接应用现有理论的问题，本项目研究等效试验约束条件下的最优设计模型，以期优化试验资源配置，并达成实战状态下的高精度一致性评估结论的目的。具体内容包括：（1）基于系统物理背景，构造半数学半物理的基元模型完备空间，针对异源等效试验建立响应曲面整体模型；（2）结合一致性评估需求构造不确定性量化指标，分析模型对异源等效试验的响应敏感程度及预测能力；（3）针对等效试验多类型约束构建约束最优设计方法，对模型进行迭代校验；（4）在复合制导武器精度的实战状态一致性评估中获得应用。本项目在元模型的扩展形式、不确定性量化模型辨识、约束最优试验设计上有创新构建，研究成果可为武器性能评定提供全局优化设计的试验方案，试验与应用的一致性评估结论，为我国导弹试验评估提供关键技术支撑。

开展导弹的试验评估本质上是利用各类试验数据，确定导弹性能影响因素的最优水平组合，建立指标与因素之间的高可信度响应模型。本项目基于“建模、设计、评估”的研究流程，以优化等效试验设计为指导思想，以制导武器精度在实战状态下的一致性评估为目的，开展导弹武器试验设计和试验评估方法。建立导弹试验的复杂响应曲面元模型：通过末制导物理机理，选取合适的影响变量，建立半物理半数学末制导误差基元模型；针对导弹编队模式，开展基于系统脆弱性的元胞自动机-图模型方法研究，提高弹群作战能力；开展区域代理模型技术研究，在减少样本需求量的同时提高模型构建的针对性。利用不确定性方法对模型进行进一步优化，开展了基于信息流、KL散度、组合高斯核函数等方法研究，在异源数据条件下提高模型的稳健性和预测能力。基于导弹试验的高低精度试验数量配比、样本量、试验范围等实际约束条件，开展嵌套空间填充设计、序贯贝叶斯填充设计、基于可卷偏差的均匀试验设计等方法研究，在约束条件下寻找最优试验设计方案，提高试验设计效率，最后在多源试验数据条件下，进行了等效试验的一致性评估方法研究。研究成果可为武器性能评定提供全局优化设计的试验方案，试验与应用的一致性评估结论，为我国导弹试验评估提供关键技术支撑。