برآورد بهینه دقت مشاهدات در شبکه‌های کلاسیک جابه‌جاسنجی

روش برآورد مولفه های واریانس کمترین‌مربعات زمانی که تنوع مشاهداتی در شبکه وجود داشته باشد کارایی خوبی از خود نشان می‌دهد. با استفاده از این روش برای هر دسته از مشاهدات مختلف یک ضریب مقیاس محاسبه می‌شود. در این تحقیق از روش وزن‌دهی برآورد مولفه های واریانس کمترین‌مربعات استفاده شده است. این بهبود دقت برای مختصات نقاط شبکه به‌نحوی است که مقدار نیم قطر بزرگ بیضی خطای مطلق نقاط در حالت استفاده از برآورد مولفه‌های واریانس کمترین ‌مربعات برابر 29 میلی‌متر، در حالی‌که با استفاده از روش فاکتور وریانس ثانویه این مقدار به دو برابر افزایش می یابد. علاوه بر این در هنگام استفاده از روش برآورد مولفه های واریانس کمترین‌مربعات اثر ماتریس کوواریانس مجهولات برابر 0.8 میلی‌متر می باشد که نسبت به روش فاکتور وریانس ثانویه مقدار آن به اندازه دو برابر کاهش می یابد. در واقع مزیت روش برآورد مولفه های واریانس کمترین‌مربعات برآورد واقع بینانه ای از دقت پارامترهای مدل و ابعاد بیضی خطای مطلق می باشد.

Optimized Estimation of Observation Precisions In Classical Displacement Network

Any infrastructure such as dams need constant monitoring for the detection of risks of failure and/or to plan civil engineering maintaining work. A recent approach considers precise geodetic instruments and satellitebased geodetic monitoring as a method to estimate potential deformation of such structures. A growing need for a fully automated and continuous monitoring of structural and ground deformations has created new challenges for design and analysis of the monitoring schemes, where multisensor geodetic systems can provide essential aid. Combination of different geodetic data helps determining displacements with high precision, hence, the risk of damages is reduced. Corresponding authorities of large manmade structures are faced with the safety problem, as all have aim to reduce risk and cost. Designers try to design large structures to tolerate against different forces like wind, traffic load, temperature, flood, earthquake, land uplift etc. Using geodetic instruments and techniques, we are able to monitor the deformation behavior or deflection in the mentioned structures and eventually provide a structural failure alarm capability (Andersson 2008). It is important to select appropriate sensor and methods to detect the deformation. Slow deforming dams require submillimeter to millimeter level accuracy to monitor the displacement and deformation (Lindenbergh et al. 2005). Reaching this level of accuracy is not costly, if geodetic sensors are integrated with other sensors (e.g. geotechnical sensors, and precise total stations, see Hwang et al. 2012). It might be to implement other sensors (e.g. laser scanner and Total Station). Using point clouds data for deformation monitoring is almost new. Gonzalez et al. (2012) studied on point clouds accuracy for applications in civil engineering e.g. deformation monitoring. They showed that the results appear suitable for deformation monitoring, with accuracies less than 1 mm. Bagherbandi et al. (2009) studied on various techniques to find the optimal design of a deformation network using various criteria such as precision, cost and reliability. Better results can be achieved using the control network, provided that an optimal network design is performed for detecting deformations (Kuang 1996). In addition, the methods of geodetic network process can affect the results (Bagherbandi 2016). The aim of this study is primarily to evaluate different deformation monitoring methods and possibilities to physically interpret the deformation and evaluate the risk of failures. In this research, the idea of assigning weights for the observations by least square variance components estimation (LSVCE) is used (AmiriSimkooei 2007; Teunissen and AmiriSimkooei 2008) in order to improve accuracy of adjustment results, which differs from the applied method in Bagherbandi (2016) to determine the variance components. Some issues and parameters should be investigated in LSVCE such as the effect of variance components estimation on the observations final accuracy, the absolute error ellipsoid estimation, the study of the necessary conditions in a network to achieve higher accuracy and its effect on obtaining real results from the reliability matrix. All results obtained from adjustment by element, LSVCE, and Tikhonov regularization are compared using a simulated geodetic network and real data. Results from this study provide important information in studying deformation that can be used to interpret the deformation mechanism, which may reduce the risk of potential disasters in large structures. We will evaluate the abovementioned methods in Jamishan dam in Iran and utilize the geodetic techniques and observations to monitor the deformation of the dam.