International Journal of Recent Technology and Engineering (IJRTE) ISSN: 2277-3878, Volume-8 Issue-5, January 2020 881 Published By: Blue Eyes Intelligence Engineering & Sciences Publication Retrieval Number:E6041018520 /2020©BEIESP DOI:10.35940/ijrte.E6041.018520 Abstract: The application of hall sensors in Magnetitic Flux Leakage (MFL) has played an important role in Above Storage Tank (AST) on detection of defect caused by corrosion to improve productivity and to avoid catastrophe. The MFL sensor measured magnetic flux distribution in three axes B x , B y and B z. Currently, there are several signal monitoring methods constructed by analysing MFL signal distribution upon defect detection. This paper presents the methodology of optimized Integrated Kurtosis-based Algorithm for Z-filter (I-Kaz TM ) Coefficient using multilevel signal decomposition technique to analyse the MFL signal distribution on the defect in the correlation of MFL scanning device speed and position. The MFL scanning device comprises 11 hall effect sensors position in array coupled with a linear guide to ensuring a constant velocity of scanning. In order to obtain an optimum signal distribution, I-Kaz TM 3D is proposed as one of the derivatives of I-Kaz TM to analyse multiple velocities of scanning. The characterization of the defect can be estimated by analysing the deflection of magnetic flux leakage in the y-axis, B y as the scanner approach the defect region before being analysed by I-Kaz TM from the beginning until the end of the workpiece. Keywords : Magnetic Flux Leakage, Kurtosis-based alghorith, I-KAZ TM , Hall effect sensors I. INTRODUCTION The Magnetic Flux leakage (MFL) is one of the reliable Non-Destructive (NDT) testing applied widely in flaw detection of plate and pipelines [1] resulted from long-term wear out and surrounding conditions. The deformed pipe and plate bring difficulty in safe operation due to corrode, cracked and deformed and it will bring a severe situation if it caused an oil-gas leakage [2]. MFL technique detects the defect volume of metal loss is frequently used in above storage tank (AST) by analyzing the shape of detected flaws before it can be classified to identify the severity of the damage [3]. The MFL detection system comprises two basic elements which are the method of detecting a leakage field and a method for magnetization. The presence of anomaly captures by the Revised Manuscript Received on January 15, 2020 * Correspondence Author Nor Afandi Sharif *, Industrial Automation and Control, Department of Electrical Engineering, German Malaysian Institute, Bangi, Malaysia. Rizauddin Ramli, Centre for Materials Engineering and Smart Manufacturing (MERCU), Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia. Mohd Zaki Nuawi, Centre for Integrated Design for Advanced Mechanical System (PRISMA), Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, Khairulbadri Ahmad, Industrial Automation and Control, Department of Electrical Engineering, German Malaysian Institute, Bangi, Malaysia. sensor in the magnetic property of the material if any defect presence. A statistical model often has been used by the researcher to evaluate and translate the signal into approximate data of depth, with and length due to indirect measurement of MFL type of evaluation[4]. In 2008, Sophian et al. [5] used Principle Component Analysis (PCA) as one of a statistical technique for extraction and classification of Pulsed Eddy Current (PEC) data. Within this technique, frequency analysis and peak point can be examined to identify defect characterization. More recent, Ding et al. (2016) [6] applied a Response Surface Methodology (RSM) to optimized input parameter values and validate the relationship between variable and response by using statistical fitting method before using Finite Element Analysis (FEA) to validate the optimized parameter. Furthermore, approximate linearity between the relation of defect length and distribution of the leakage field has also been studied by other researchers. The length and depth of the defect is interrelated because both of the components contributes to the magnitude of flux leakage [7]. In this paper, we propose a methodology for characterizing the defect of length and depth in the metal plate with various thick and designated defects using a novel statistical approach by implementing optimized Integrated Kurtosis-based Algorithm for Z-filter (I-Kaz TM ) [8]. The acquired signal in the review criteria experiment will be decomposed into three frequency ranges comprises low-frequency (LF) range of 0 – 0.25fmax, a high-frequency (HF) range of 0.25fmax – 0.5fmax and a very high- frequency (VF) range of 0.5fmax. On the other hand, Neural network (NN) signal analysis technique as one of artificial intelligent (AI) tools also been used widely in the MFL inspection due to its ability to provide solution in complex data analysis [9]. The defect characterization has been improved by implementing NN type analysis in a previous study [10][11]. Joshi (2008) [12] and Ramuhalli (2002) [13] studied the estimation of a 3D geometry defect in predicting maximum allowable operating pressure (MAOP) in a pipe due to inaccurate signal interpolation by applying wavelet neural network to identify the correlation between defect MFL signal and geometric parameters [14]. It shows that NN has faster convergence rate and fit better for the approximation of the multivariable function as long as the defect area is properly targeted [15]. The application of Finite Element Method (FEM) based simulation is another technique for investigate the behavior of MFL signal on the defect region in Characterization of Defect for Magnetic Flux Leakage in Non-Destructive Test using I-Kaz tm Nor Afandi Sharif, Rizauddin Ramli, Mohd Zaki Nuawi, Khairulbadri Ahmad