Elektromanyetik açıcının bazı fiziksel özelliklerinin ölçümü, sonlu elemanlar yöntemi ile analizi ve karakteristiğinin makine öğrenmesi ile kestirimi = Measurement of some physical properties of an electromagnetic plunger, its analysis with finite element method, and estimating the characteristic using machine learning

Abstract

Elektromanyetik açıcılar, sanayide, günlük hayatta karşılaşılan ve sık kullanılan cihazlardan biridir. Bu cihazlar sıvı, gaz ve yağ gibi akışkanları kontrol etmek için kullanılmaktadır. Sıvı geçişlerini kontrol etmek için açıcıların basıncı, debisi, çekme süresi, elektromanyetik açıcının boyutu, bobin, ve kullanılan manyetik malzemeler önem teşkil etmektedir. Bu parametreler tasarımın fiziksel büyüklüklerine, kullanılan malzemenin tipine, çalışma akımına, kullanılan bobine göre farklı değerlere sahip olmaktadır. Bu çalışmada tasarım özellikleri bilinen bir elektromanyetik açıcı kullanılarak mekanik ve elektriksel ölçümler gerçekleştirilmiştir. Bu kapsamda cihazın yay sabiti ve çekme süresi ölçülmüştür. Yay sabitini ölçmek için dinamometre kullanılmıştır. Daha sonra ise elektriksel ölçüm için osiloskop kullanılmış ve açıcının çekme süresi belirlenmiştir. Elektromanyetik açıcının mekanik özellikleri (boyutu, yay sabiti) ve elektriksel akımın zamana göre değişimi açıcının çekme süresine etki etmektedir. Elektromanyetik açıcı farklı gerilimlerde ve pistonun farklı çekme mesafelerinde çalıştırılarak çekme süresi ölçülmüştür. Daha sonra Sonlu Elemanlar Yöntemi (SEY) temelli çözümleyicilere sahip COMSOL Multiphysics programı ile açıcının benzetimi gerçekleştirilerek çalışma karakteristiği verileri alınmıştır. Yapılan benzetim ile açıcının çalışma karakteristiğinin elde edilmesi amaçlanmıştır. Benzetimlerle pistonun yer değişmesi, hızı, endüktans ve manyetik akı değişkenleri zamana bağlı olarak hesaplanmıştır. Açıcıda ölçüm ile elde edilen gerilim ve pistonun farklı hareket etme mesafeleri kullanılarak Yapay Sinir Ağı (YSA) eğitimi gerçekleştirilmiş ve akım karakteristiğin tahmini yapılmıştır. Hataların Karesinin Ortalamasının Karekökü yöntemi kullanılarak sonuçlar değerlendirilmiştir. Yapılan eğitimler sonucunda %2 – %5 aralığında hata oranı bulunmuştur. Ölçüm sırasında pistonun hızı ve yer değişmesi doğrudan belirlenemediğinden COMSOL Multiphysics programındaki benzetimler ile bu karakteristikler belirlenmiştir. Model sonuçları temel alınarak yapılan YSA eğitimleri ile söz konusu büyüklükler %3 – %6 arasında hata oranı ile tahmin edilmiştir. İki farklı manyetik karakteristikli nüveye sahip elektromanyetik acıcının çekme süresi ölçümleri yapılmış ve bu nüvelerin karakteristiğinin açıcının çekme süresi üzerindeki etkileri araştırılmıştır. Ölçülen çekme süresi ve uygulanan gerilim kullanarak kullanılan piston malzemesinin tipinin tahmin edilebileceği görülmüştür. Piston malzemesinin tahmin edilmesi ve sınıflandırılması için destek vektör makineleri ve yapay sinir ağları eğitim yöntemleri kullanılmıştır. Yapılan eğitimler sonucunda performans değeri %91 olarak elde edilmiştir.

Electromagnetic plungers (EMP) are electromechanical structures that convert electrical energy into mechanical movement. They work by applying electrical energy to control the motion of a plunger through the use of solenoid coils. Because of this, EMPs are also known as solenoid valves or solenoids. The term "solenoid" comes from the Greek words "solen" meaning pipe, and "eidos" meaning shape. An EMP consists of a coil wound around a ferromagnetic core, a plunger, and a spring. When an electrical current is applied to the coil, it generates a magnetic field that pulls the plunger, creating a linear displacement movement. If an electromechanical system rotates as a result of energy conversion, it is referred to as an electric motor, whereas if it creates a linear displacement movement, it is called an electromagnet. When the electrical current is removed, the spring returns the plunger to its original position. Valves are devices that offer several advantages in the automation applications, such as fast and safe switching, a long lifespan, and compatibility with the materials used in the environment. Valves are primarily used to control the flow of fluid, including liquids, air, and gas. They are frequently used in irrigation systems in gardens, appliances like washing machines or dishwashers, and in the pneumatic or hydraulic industrial fields, as well as in the automotive and space industries. Important parameters for controlling fluid flow include pressure, flow rate, and retraction time, which vary depending on factors such as the design's physical dimensions, the type of material used, the working flow, and the coil used. In electromagnetic actuators, the moment when the plunger opens and allows the fluid to flow through the pipe is referred to as the open the orifice,. This opening coincides with the moment of retraction, hence it is also known as the retraction time. The pull-in time of a solenoid valve has been influenced by several factors such as the valve design, its size, the strength of the solenoid, and the amount of electrical current applied. Solenoid valves can be categorized into two types based on their mode of operation: pilot-controlled and direct-acting solenoid valves. In direct-acting solenoid valves, a plunger is directly pulled by the solenoid. When electric current is applied, the plunger is pulled by the solenoid and the valve opens. The plunger has been used to allow fluid to pass through pipes. Pilot-controlled solenoid valves are controlled by a pilot valve. The pilot is the control element inside the valve. This control element has been used to control the function of the valve. Pilot-controlled solenoid valves are generally used to control parameters such as ambient temperature, pressure, or flow rate. In this study, two electromagnetic plungers with defined design properties have been used to determine their mechanical and electrical properties. The actuator plungers have been fabricated from various materials with varying magnetic permeability. The spring constant was measured with a dynamometer and the pull-in time was determined with an oscilloscope. The pull time was affected by changes in the mechanical properties and electrical current over time, and was measured at different voltages and plunger pull distances. Measurements have been repeated at 6 different voltages and 5 different displacement distances. Finite element method (FEM) - based programs have been widely used to avoid costly and unnecessary options when creating electromagnetic plunger models, to shorten the design process and conduct accurate analyses. Analyses performed using these programs have been known as finite element analysis (FEA) techniques. Simulations have been performed using the COMSOL Multiphysics program and a FEM-based solver to preserve actuator motion properties. The simulation was performed according to real measurement conditions in this program. The aim has been to obtain the working characteristic of the actuator through the simulation. In these simulations, the displacement, velocity, inductance, and magnetic flux density variables of the plunger were calculated. The Artificial Neural Network (ANN) has been trained using the voltage obtained during the measurement and various movement distances of the piston, and the estimated current characteristic has been calculated. All measurements and simulations have been performed out for 2 EMPs. The results were evaluated using the Root Mean Square Error method. In the estimations made for the measurement results, the error rate was found in the range of 2% – 5%.Since the velocity and displacement of the plunger could not be directly determined during the measurement, these variables were determined by simulation. The determined variables have been trained using an ANN and the displacement of the plunger was estimated. The error rate of 3% - 6% was calculated for the simulation results. The pull-in time of two electromagnetic actuators with different magnetic characteristics have been measured, and the effects of these characteristics on the pull-in time of the actuator have been investigated. It has been shown that the type of core material used can be estimated based on the measured pull-in time and applied voltage. Data for plunger material estimation has been trained on both support vector machines and artificial neural networks to classify material types. The input data for classification, including time, voltage, current, the spring used, and solenoid coil power used, has been entered into the system. Confusion matrices have been obtained to assess the accuracy of the classification results, and the accuracy, sensitivity, and specificity indices have been measured. As a result of the classification study, 91% success have been achieved. It has been determined that EMP is more efficient with shorter pull-in-time. Research in this area has also been conducted in literature. In this study, electrical and mechanical measurements have been performed using two different magnetic-characteristic plungers, which is different from the literature. The simulation of the actuator has been carried out in the COMSOL Multiphysics program, and the training has been performed using machine learning algorithms, based on the reviewed literature. In the second section of the thesis, general information about electromagnetic actuators has been provided, along with the electromagnetic model and state space equations of the plunger. In the third section, the structure of the electromagnetic actuator used in the study is analyzed, and the experimentally obtained inductance value of the actuator and its current characteristics are presented. Measurements have been performed using two different magnetic plunger samples. In the fourth section of the thesis, modeling and simulation of the electromagnetic actuators used in the study have been carried out using the COMSOL Multiphysics program. In the fifth section, the results of the measurement and simulation have been trained using machine learning algorithms, and various estimations have been made. The measurement and simulation results have been trained with ANN, and the current characteristic of the opener and the distance traveled by the piston have been estimated. The mean square error method has been used to carry out the error analysis. Subsequently, the classification of two different magnetic piston samples was performed using support vector machine and artificial neural network methods. The thesis has been concluded with the results section.