Application of neural network and electrodynamic sensor as flow pattern identifier

Abstract

PurposeSolid particles flowing in a pipeline is a common mode of transport in industries. This is because pipeline transportation can avoid waste through spillage and minimizes the risk of handling of hazardous materials. Pharmaceutical industries, food stuff manufacturing industries, cement, and chemical industries are a few industries to exploit this transportation technique. For such industries, monitoring and controlling material flow through the pipe is an essential element to ensure efficiency and safety of the system. The purpose of this paper is to present electrical charge tomography, which is one of the most efficient, robust, cost‐effective, and non‐invasive tomographic methods of monitoring solid particles flow in a pipeline.Design/methodology/approachProcess flow data are captured by fitting an array of 16 discrete electrodynamic sensors about the circumference of the flow pipe. The captured data are processed using two tomographic algorithms to obtain tomographic images of the flow. Then a neural network tool is used to improve image resolution and accuracy of measurements.FindingsThe results from the above technique show significant improvements in the pipe flow image resolution and measurements.Originality/valueThe paper presents electrical charge tomography, which is one of the most efficient, robust, cost‐effective, and non‐invasive tomographic methods of monitoring solid particles flow in a pipeline.

Image Reconstruction Algorithm for Electrical Charge Tomography System

“…FBP method is accurate to detect the high concentrations area but has the problem to detect more charges in sensing area. As a result the image provides by FBP is looking like a huge point with uniform pattern and the value of concentration is reduced away from the sensor location to the center of sensing area (Rahmat et al, 2010). The problem with LS method is no unique solution (Isa and Rahmat, 2009).…”

“…In order to solve the inverse problem better, a systematic approach of image reconstruction algorithm is very important. Even though the image reconstruction had been discussed by many researchers (Yan and Byrne, 1997;Machida and Scarlett, 2005;Rahmat et al, 2010;Yan et al, 1995), but in EChT system, this aspect is still much to be explored.…”

“…Three methods have been introduced for image reconstruction in EChT before, namely Linear Back Projection (LBP) (Yang and Peng, 2003), Filtered Back Projection (FBP) (Yan and Byrne, 1997) and least square methods (Rahmat et al, 2010). However, LBP method is no-linearity with the sensing mechanism used in the electrical charge tomography, thus image reconstruction is focusing only on the sensor location (Machida and Scarlett, 2005;Rahim et al, 2008).…”

“…These particles will then be screened and monitored by a suitable transducer together with the charge detection circuit. This process is known as electrodynamics sensor (Rahmat et al, 2010). This sensor will detect the condition of the electrostatic charge on the moving particles.…”

“…Process tomography system by electrical charge is given one measurement from each of the sensors; the amount of information available is equal to the number of the sensors (Rahmat et al, 2010). Therefore, forward modeling of the pipe and the sensing area are mapped equally to the number of the sensor i.e., sixteen by sixteen-rectangular array consisting of 256 pixels or elements.…”