نوع مقاله : مقاله پژوهشی
عنوان مقاله English
نویسندگان English
The purpose of this research is to monitor the condition and troubleshoot the gas turbine of marine propulsion system by developing artificial intelligence models. Therefore, artificial intelligence algorithms are divided into two categories: machine learning and deep learning. The use of machine learning methods usually requires the extraction of signal features by the user, while in deep learning methods, the raw signal can be provided to the neural network to perform feature extraction and then regression or classification (Orhan & Celik, 2023). In this research, after the general description of the marine propulsion system, we use a dataset related to the turbine and compressor of the ship's propulsion system. Each data contains 16 operational parameters of the system and two decay coefficients related to the turbine and compressor of the system. The 16 parameters will be the features and the two decay coefficients will be the label or output of the model (Coraddu et al., 2014). Regression models should be developed due to the continuous value of wear coefficients. In order to develop the model using Python programming language, classical machine learning methods such as decision tree regression, k-nearest neighbor, polynomial regression, Gaussian and Lasso, and in the next step, multi-layer neural regression model, are used to develop the model .Finally, the performance of these methods was compared and the polynomial regression method and the decision tree had the best performance among the rest of the models.
کلیدواژهها English
References
