güllü, mevlüt afyon kocatepe üniversitesi - mühendislik fakültesi - harita mühendisliği bölümü, Turkey

Soft Computing Model in Geodetic Coordinate Tranformation

27814

In Turkey, the geodetic and values are produced in ED50 until 2001. With the increase in the use of satellite techniques since 1990, the coordinate values began to be obtained in WGS84 system. Naturally, it has become necessary to make the transformation between these datums. The objective of this study is using Artificial Neural Networks (ANN) in the coordinate transformation. Over the test network that was constituted from the selected points in Afyonkarahisar and the surrounding provinces, the similarity transformation and the Back Propagation Artificial Neural Networks (BPANN) are compared. The differences between the known coordinates and the coordinates calculated by the similarity transformation and BPANN are evaluated, in terms of root mean square error and better results were achieved by BPANN.

655

Coordinate transformation , Artificial neural networks , Back propagation artificial neural Networks , Similarity.

Afyon Kocatepe University Journal Of Science an‎d Engineering

659