Prediction of current required for Cathodic protection system by artificial neural network

العناوين الأخرى

التنبؤ بالتيار المطلوب لمنظومة الحماية الكاثودية باستخدام الشبكات العصبية الصناعية

مقدم أطروحة جامعية

Muhsin, Samar Muhammad

مشرف أطروحة جامعية

Rida, Mustafa M.
Hammadi, Nawal J.

أعضاء اللجنة

Muhammad, Ahmad A.
Atiq, Adnan A.
Slaymun, Abbas Hamid

الجامعة

جامعة البصرة

الكلية

كلية الهندسة

القسم الأكاديمي

قسم الهندسة الكيميائية

دولة الجامعة

العراق

الدرجة العلمية

ماجستير

تاريخ الدرجة العلمية

2012

الملخص الإنجليزي

The present study deals with the experimental and theoretical studies for a case study cathodic protection (CP) system.

The first part, deals with installation of a mini cathodic protection system for submerged steel pipe segment.

Several variables affecting submerged metallic structure in natural water were simulated by covering the range of conductivity values using sodium chloride solution from (0-3.5)% NaCl, pH from (5-9), under dynamic condition provided by stirring the solution starting from the static conditions up to 2000 rpm, and at a temperature of 30oC.

The values of current required for protection (Iprot) of the pipe segment at the above specified studied conditions were measured and recorded.

The result obtained in sodium chloride solution were compared with the measurements of the (Iprot) in natural sea water sample.

The second part includes the theoretical study using the experimental data obtained to build an Artificial Neural Network (ANN) model employing MATLAB R 2010 b programming, to predict the current required at any other specified set of conditions within the studied range.

The experimental results showed that the (Iprot) increased with increasing both of the solution concentration and stirring rate.

Also, (Iprot) increased slightly as the pH value decreased.

It was found that the turbulence factor showed a remarkable effect on (Iprot), and to a lesser extent the concentration and pH values.

However, the values of (Iprot) obtained for natural sea water were found less than that observed in case of using 3.5% NaCl sodium chloride solution.

Neural network results showed good convergence with the experimental data as 8E-5 mean square error was achieved with regression correlation of 0.9997 by two hidden layers network configuration of 7 nodes for each hidden layer.

التخصصات الرئيسية

الكيمياء

عدد الصفحات

110

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-746500