飞控系统多元数据融合虚拟试验仿真模型验证

Multivariate data fusion virtual test simulation model validation in flight control systems

为提高飞控系统虚拟试验仿真模型的有效性与可信性,解决虚拟试验中多元冗余与强耦合数据的问题,提出了一种基于多元数据融合的仿真模型验证方法。首先,通过多元仿真数据融合,减少噪声和误差;其次,运用皮尔逊相关系数分析多元输出数据的相关性,并结合时域卷积神经网络-长短期记忆(TCN-LSTM)方法融合数据特征,深入挖掘数据的时空关联性;最后,通过概率分布分析评估仿真数据与参考数据的差异,转化为可信度等级,实现仿真模型的有效验证。实验结果表明,80%的输出数据达到仿真模型可信标准,其中有50%的输出数据达到完全可信标准,该方法显著提升了仿真模型的验证精度与可靠性。

This study seeks to enhance the effectiveness and reliability of virtual test simulation models for flight control systems.To address the challenges posed by multivariate redundancy and strong data coupling in virtual testing,a simulation model validation method based on multivariate data fusion is proposed.Initially,multivariate simulation data is integrated to minimize noise and reduce errors.Subsequently,the Pearson correlation coefficient is applied to assess the correlations among the multivariate output data,while the TCN-LSTM approach is utilized to fuse data features,thereby uncovering the spatial-temporal relationships within the data.Finally,probability distribution analysis is conducted to quantify the discrepancies between simulation and reference data,which are then converted into credibility scores to facilitate the effective validation of the simulation model.Experimental results indicate that 80%of the output data satisfies the simulation model credibility criteria,with 50%of the output data classified as fully credible.These findings underscore the method's ability to significantly enhance the accuracy and reliability of simulation model validation,while effectively integrating the quantitative outcomes as required.