A Pattern-Recognizer Artificial Neural Network for the Prediction of New Crescent Visibility in Iraq

Various theories have been proposed since in last century to predict the first sighting of a new crescent moon. None them uses concept machine and deep learning process, interpret simulate patterns hidden databases. Many these use interpolation extrapolation techniques identify regions through such data. In this study, pattern recognizer artificial neural network was trained distinguish between visibility regions. Essential parameters moon were collected from sight datasets used build an intelligent system recognition conditions. The ANN learned with accuracy more than 72% comparison actual observational results. simulation gives clear insight into three regions: invisible (I), probably visible (P), certainly (V). is suitable for building lunar calendars, so it four-year calendar on horizon Baghdad. built compared official Hijri Iraq.