30113

پيش بيني عملكرد غشاهاي نانوفيلتراسيون و اسمز معكوس با استفاده از مدلهاي يادگيري ماشين جهت حذف ميكرو آلاينده ها

كارشناسي ارشد

مهندسي شيمي- پديده‌هاي انتقال و فرايند جداسازي

1400

1402/7/15

عليرضا همتي

احد قائمي

مهندسي شيمي، نفت و گاز

در اين پژوهش با توجه به اهميت تصفيه آب، به بهينهسازي فرآيندهاي غشايي براي حذف ميكروآلايندهها پرداختيم. امروزه هوش مصنوعي يك فناوري پيشرو با قابليت تركيب رفتار و هوش انساني در ماشينها يا سيستمها است. بنابراين، مدلسازي مبتني بر هوش مصنوعي، كليد ساخت سيستمهاي خودكار و هوشمند با توجه به نيازهاي امروزي است. براي حل مسائل دنياي واقعي، انواع مختلفي از هوش مصنوعي مانند هوش مصنوعي تحليلي، عملكردي، تعاملي، متني و بصري را ميتوان براي افزايش هوش و قابليتهاي يك برنامه كاربردي به كار برد به همين منظور از سه الگوريتم شبكه عصبي مصنوعي، جنگل تصادفي و XGB براي پيش بيني عملكرد غشاهاي نانوفيلتراسيون و اسمز معكوس پرداختيم و پس از يادگيري و تست مدل شبكه عصبي مصنوعي با ميانگين خطاي مطلق 139/0 بهترين عملكرد را براي ما داشت.

1402/08/30

Predicting the performance of nanofiltration and reverse osmosis membranes using machine learning models to remove micropollutant

1/1/1900 12:00:00 AM

Water and wastewater treatment is a process that improves the quality to make it more suitable for a specific purpose. Today, one of the main applications of membranes is water and wastewater treatment. Micropollutants are pollutants found in small concentrations in water that are persistent and bioactive, meaning they are not completely dissociable and cannot be removed by conventional water treatment methods. For this reason, their detection and elimination has challenged the scientific community. These micropollutants have caused great concern because their presence in water supply systems endangers the health of humans and animals. To develop efficient techniques for their removal, it is necessary to understand their physical and chemical properties and to know all the processes capable of removing them. Membrane separation processes are used as a suitable method to remove micropollutants. Today, artificial intelligence is an advanced technology with the ability to combine human behavior and intelligence in machines or systems. Therefore, modeling based on artificial intelligence is the key to building automatic and intelligent systems according to today's needs. To solve real-world problems, different types of AI such as analytical, functional, interactive, textual, and visual AI can be used to enhance the intelligence and capabilities of an application. The removal of pollutants by nanofiltration and reverse osmosis membranes is a multidimensional process that includes the selection of membrane materials and the optimization of experimental conditions. It is difficult to discover the contribution of factors affecting the removal rate by trial and error experiments. However, the advanced machine learning method is a powerful tool to simulate this complex process, which includes 4 traditional learning algorithms (regression, support vector machines, artificial neural network, K nearest neighbor) and 4 group learning algorithms (random forest, decision tree, and gradient amplification) is used to predict the pollutant removal efficiency. The results have shown that group learning models make significantly better predictions than traditional models.

هوش مصنوعي , غشا اسمز معكوس

artificial intelligence , reverse osmosis membrane

Parnian Masoodi

Dr. Alireza Hemmati