The contribution of this research effort was to show that a reliable RPV could be built, tested, and successfully used for flight testing and parameter estimation purposes, in an academic setting. This was a fundamental step towards the creation of an automated Unmanned Aerial Vehicle (UAV). This research project was divided into four phases. Phase one involved the construction, development, and initial flight of a Remotely Piloted Vehicle (RPV), the West Virginia University (WVU) Boeing 777 (B777) aircraft. This phase included the creation of an onboard instrumentation system to provide aircraft flight data. The objective of the second phase was to estimate the longitudinal and lateral-directional stability and control derivatives from actual flight data for the B777 model. This involved performing and recording flight test maneuvers used for analysis of the longitudinal and lateral-directional estimates. Flight maneuvers included control surface doublets produced by the elevator, aileron, and rudder controls. A parameter estimation program known as pEst, developed at NASA Dryden Flight Research Center (DFRC), was used to compute the off-line estimates of parameters from collected flight data. This estimation software uses the Maximum Likelihood (ML) method with a Newton-Raphson (NR) minimization algorithm. The mathematical model used a traditional static and dynamic derivative buildup. Phase three focused on comparing a linear model obtained from the phase two ML estimates, with linear models obtained from a (i) Batch Least Squares Technique (BLS) and (ii) a technique from the Matlab system identification toolbox. Historically, aircraft parameter estimation has been performed off-line using recorded flight data from specifically designed maneuvers. In recent years, several on-line parameter identification techniques have been evaluated for real-time on-line applications. Along this research line, a novel contribution of this work was to compare the off-line estimation results with results obtained using a recently introduced frequency based on-line estimation method. Specifically, phase four focused on comparing the ML results with a frequency domain based on-line estimation technique. The RPV vehicle and payload was designed and constructed with the combined efforts of WVU researchers, graduate and undergraduate students of the Mechanical and Aerospace Engineering Department, and a private sub-contractor, Craig Aviation.
展开▼
机译：这项研究工作的贡献在于，可以在学术环境中构建，测试并成功地将可靠的RPV用于飞行测试和参数估计。这是迈向创建自动无人机（UAV）的基本步骤。该研究项目分为四个阶段。第一阶段涉及远程驾驶车辆（RPV），西弗吉尼亚大学（WVU）波音777（B777）飞机的建造，开发和初始飞行。此阶段包括创建机载仪表系统以提供飞机飞行数据。第二阶段的目标是从B777模型的实际飞行数据中估算纵向和横向稳定性并控制导数。这涉及执行和记录用于分析纵向和横向估计的飞行试验演习。飞行操纵包括由电梯，副翼和方向舵控制装置产生的操纵面双合翼。 NASA Dryden飞行研究中心（DFRC）开发了一个称为pEst的参数估算程序，用于根据收集的飞行数据计算参数的离线估算。该估计软件使用最大似然（ML）方法和牛顿-拉夫森（NR）最小化算法。数学模型使用传统的静态和动态导数累积。第三阶段着重于比较从第二阶段ML估计获得的线性模型与从（i）最小二乘最小二乘（BLS）和（ii）Matlab系统识别工具箱中的技术获得的线性模型。从历史上看，飞机参数估计是使用已记录的专门设计的飞行数据离线进行的。近年来，已经针对实时在线应用评估了几种在线参数识别技术。沿着这项研究路线，这项工作的新颖贡献在于将离线估计结果与使用最近引入的基于频率的在线估计方法获得的结果进行比较。具体而言，第四阶段着重于将ML结果与基于频域的在线估计技术进行比较。 RPV车辆和有效载荷是在WVU研究人员，机械和航空航天工程系的研究生和本科生以及私人分包商Craig Aviation的共同努力下设计和制造的。
展开▼
