Comparison of Selected Machine Learning Algorithms for Industrial Electrical Tomography

Abstract

The main goal of this work was to compare the selected machine learning methods with the classic deterministic method in the industrial field of electrical impedance tomography. The research focused on the development and comparison of algorithms and models for the analysis and reconstruction of data using electrical tomography. The novelty was the use of original machine learning algorithms. Their characteristic feature is the use of many separately trained subsystems, each of which generates a single pixel of the output image. Artificial Neural Network (ANN), LARS and Elastic net methods were used to solve the inverse problem. These algorithms have been modified by a corresponding increase in equations (multiply) for electrical impedance tomography using the finite element method grid. The Gauss-Newton method was used as a reference to machine learning methods. The algorithms were trained using learning data obtained through computer simulation based on real models. The results of the experiments showed that in the considered cases the best quality of reconstructions was achieved by ANN. At the same time, ANN was the slowest in terms of both the training process and the speed of image generation. Other machine learning methods were comparable with the deterministic Gauss-Newton method and with each other.

On the Design of a New Simulated Inductor Using a Contactless Electrical Tomography System as an Example

“…In order to compare the image reproduction effect, three image quality indexes were used [21]: mean squared error (MSE), relative image error (RIE) and image correlation coefficient (ICC). MSE is evaluated according to Equation (56).…”

“…Currently, contactless electrical tomography (CET) has received increasing attention in the field of process tomography [16,17]. The latest research progresses have displayed the potential and broad perspective of the contactless electrical tomography (CET) techniques [18,19,20,21,22,23,24,25,26,27]. However, the current CET systems still exist a drawback, i.e., the existence of the coupling capacitances (formed by the two electrodes, insulating pipe, and the measured fluid) are adverse to the detection of the useful measurement signal (the details will be discussed in Section 5).…”

Logistic Regression for Machine Learning in Process Tomography

“…For this reason, the voltages take positive and negative values. A more precise method of generating simulation data in the form of pseudocode is presented in the article [47].…”

“…Other than this approach, currently the most commonly used methods employ machine learning algorithms [27,44]. The popular methods of machine learning used in tomography include neural networks [31], deep learning [27] and statistical methods [47]. Logistic regression is counted among statistical methods of machine learning.…”

On the Use of a Rotatable ECT Sensor to Investigate Dense Phase Flow: A Feasibility Study

“…The only capacitance residual error was observed to select the desired typical number of iterations for all cases according to the formula given by Yang and Peng [ 6 ]. It is also worth mentioning that the IBP algorithm is time consuming compared to LBP (~120 times slower), which makes it so far only applicable off-line [ 30 ].…”