International Journal of Neural Networks and Advanced Applications
E-ISSN: 2313-0563
Volume 6, 2019
Notice: As of 2014 and for the forthcoming years, the publication frequency/periodicity of NAUN Journals is adapted to the 'continuously updated' model. What this means is that instead of being separated into issues, new papers will be added on a continuous basis, allowing a more regular flow and shorter publication times. The papers will appear in reverse order, therefore the most recent one will be on top.
Submit a paper | Submission terms | Paper format
Volume 6, 2019
Title of the Paper: Comparison of Artificial Neural Network Controller and PID Controller in on Line of Real Time Industrial Temperature Process Control System
Authors: Abdulgani Albagul, Hafed Efheij, Bileid Abdulsalam
Pages: 69-74
Abstract: The conventional PID (proportional-integral- derivative) controller is widely applied to industrial automation and process control field because its structure is sample and its robust is well, but it do not work well for nonlinear system, time-delayed linear system and time varying system. This paper provides a new style of PID controller that is based on artificial neural network and evolutionary algorithm according to the conventional one’s mathematical formula. Artificial Neural Network is an effective tool for highly nonlinear system. With the advent of high-speed computer system, there is more increased interest in the study of nonlinear system. Neuro control algorithm is mostly implemented for the application to robotic systems and also some development has occurred in process control systems. Process Control systems are often nonlinear and difficult to control accurately. Their dynamic models are more difficult to derive than those used in aerospace or robotic control, and they tend to change in an unpredictable way. This paper gives an example where a multilayered feed forward back propagation neural network is trained offline to perform as a controller for a temperature control system with no a priori knowledge regarding its dynamics. The inverse dynamics model is developed by applying a variety of input vectors to the neural network. The performance of neural network based on these input vectors is observed by configuring it directly to control the process. This paper compared the performance of PID controller with ANN based upon Set point change, Effect of load disturbances and Processes with variable dead time. The result shows that ANN outperforms the PID controller.
Title of the Paper: Three Link Rigid Manipulator Control using Improved Neural Network based PID Controller
Authors: Sherif G. Ahmad, Mohamed A. El-Gohary, Mohamed S. Elksas, Fayez G. Areed
Pages: 60-68
Abstract: This paper presents an artificial neural network based pid controller of a three link rigid manipulator. We develop neural network control algorithms to solve the nonlinear problems for compensating robot manipulator control with uncertainties so that accurate position could be achieved. The back propagation algorithm has been used for training a two layered feedforward artificial neural network. Our proposed controller is simply combining the ANN with other conventional control method and provide the network with more data about the structure and the behavior of the system, the neural network is trained with the data generated by pid controller. The simulation result shows that the controller works well and performs better than the conventional PID.
Title of the Paper: Artificial Neural Networks for Forecasting Passenger Flows on Metro Lines
Authors: Mariano Gallo, Giuseppina De Luca, Luca D’Acierno, Marilisa Botte
Pages: 53-59
Abstract: Forecasting users’ flows on transportation networks is a fundamental task of Intelligent Transport Systems (ITSs); indeed, most control and management strategies on transportation systems are based on the knowledge of user flows. For implementing ITS strategies, the forecast of user flows on some network links obtained as a function of user flows on other links (for instance, where the data are available in real-time) may be a significant contribution. In this paper, we propose the use of Artificial Neural Networks (ANNs) for forecasting metro onboard passenger flows as a function of passenger counts at stations’ turnstiles. We assume that metro station turnstiles allow collecting the number of entering passenger by means of an automatic counting system and these data are available every some minutes (temporal aggregation); the objective is to estimate on-board passengers on each track section of the line (i.e. between two successive stations) as a function of turnstiles’ data collected in the previous periods. The choice of the period length may depend on service schedules. Artificial Neural Networks are trained by using simulation data obtained with a dynamic loading procedure of the rail line; the proposed approach is tested on a real-scale case: the Line 1 of the Naples metro system (Italy). Numerical results show that the proposed approach is able to forecast the flows on metro sections with satisfactory precision.
Title of the Paper: Pedestrian Detection in Infrared Images using FastRCNN
Authors: Asad Ullah, Farooq Muhammad, Xie H. M., Sun Zhaoyun
Pages: 46-52
Abstract: Compared to visible spectrum image the infrared image is much clearer in poor lighting conditions. Infrared imaging devices are capable to operate even without the availability of visible light, acquires clear images of objects which are helpful in efficient classification and detection. For image object classification and detection, CNN which belongs to the class of feed-forward ANN, has been successfully used. Fast RCNN combines advantages of modern CNN detectors i.e. RCNN and SPPnet to classify object proposals more efficiently, resulting in better and faster detection. To further improve the detection rate and speed of Fast RCNN, two modifications are proposed in this paper. One for accuracy in which an extra convolutional layer is added to the network and named it as Fast RCNN type 2, other for speed in which the input channel is reduced from three channel input to one and named as Fast RCNN type 3. Fast RCNN type 1 ( original Fast RCNN ) has better detection rate than RCNN (60.3636 : 54.54318), with 3.5 x RCNN training and 2.5 x RCNN test speed. Compare to Fast RCNN, Fast RCNN type 2 has better detection rate (63.42% : 60.36%) while Fast RCNN type 3 is faster (having 1.018 x FastRCNN training and 1.04 x FastRCNN test speed).
Title of the Paper: Hyperspectral Image Content Identification using Kernel based Neural Network
Authors: Puttaswamy M. R., Balamurugan P.
Pages: 39-45
Abstract: Large dimension data of Hyperspectral Image (HSI) leads to high computation cost, more execution time and increase the memory demand; therefore, difficulty arises during the classification of HSI. Unsupervised-BSA (band selection algorithm) using linear projection (LP) dependent band metric similarity has considered for informatics band selection of HSI. However, space complexity and time constrain is very big challenge for many feature algorithm. In this paper, we will use monogenetic binary feature (MBF) to get the local and monogenic feature from each pixel. The extracted feature from the combined monogenetic parameter will be used for effective classification process. In post processing, the classification of extracted feature from the MBF is perform by the Neural Network (NN) and to provide an enhanced generalization ability we introduce a kernel based NN, and considering the unknown feature mapping. To validate the performance of proposed hyperspectral classification algorithm, we will compared with several state-of-the-art.
Title of the Paper: A Modular Intrusion Detection System based on Artificial Neural Networks
Authors: Antonios S. Andreatos, Vassilios C. Moussas
Pages: 32-38
Abstract: This paper proposes a novel Intrusion Detection System (IDS) based on Artificial Neural Networks (ANNs). The proposed multi-ANN system is modular, parallel and easily expandable in order to detect additional types of attacks. Three types of attacks have been tested so far: DDoS, PortScan and Web attacks. The experimental results obtained by analyzing and testing the proposed IDS using the CICIDS2017 dataset, show satisfactory performance and superiority in terms of accuracy, detection rate, false alarm rate and time overhead compared to existing single-ANN systems.
Title of the Paper: Optimum Process Parameters in Superfinishing Process using Artificial Neural Networks
Authors: Badea Lepadatescu
Pages: 25-31
Abstract: The work reported describes application of artificial neural networks (ANN) for the purpose of deriving a complex nonlinear relationship among several factors that influence the roughness of part surfaces obtained through superfinishing process according with different process parameters. The relationship is necessary to optimize the process parameters and predict the optimum values to obtain the roughness surfaces that are needed for the part manufactured. A feed forward two-layers ANN is designed and trained using experimental data. The model is tested for generalization and simulated in MATLABTM environment. The results are used to determine the best process parameters that must be used to have a high surface finish according with the technical requirements.
Title of the Paper: Multi-target Regression Approach for Predictive Maintenance in Oil Refineries using Deep Learning
Authors: Helmi Helmiriawan, Zaid Al-Ars
Pages: 18-24
Abstract: Modern oil refineries typically use a high number of sensors that generate a massive amount of data about various process variables in the infrastructure. This data can be used to perform predictive maintenance, an approach to predict impending equipment failures and mitigate downtime in refineries. This paper presents the use of multi-target regression approach for predictive maintenance. Multi-target regression is a modeling approach that aims to predict multiple targets simultaneously. The relationships between multiple process variables are modeled using deep learning methods, while the model error is evaluated using cumulative sum method to detect faults that might potentially become failures. Unlike many existing solutions, our approach does not rely on the availability of data that captures the presence of faults in the plant. The proposed approach is demonstrated using real industrial data from a crude distiller in Shell Pernis. The results show a speed-up in modeling time by 16x and an improved early fault detection time by 1.2x as compared with the single-regression approach. Furthermore, the proposed approach is also able to isolate the faults by producing higher errors in predicting faulty equipment compared with healthy equipment.
Title of the Paper: The Use of Computational Intelligence Paradigms in Smart Software Engineering: Techniques, Applications and Challenges
Authors: Kariman Ramzy ElHelow, Abdel-Badeeh M. Salem
Pages: 12-17
Abstract: Computational Intelligence (CI) is an efficient paradigm for development intelligent systems. This paradigm has resulted from a synergy between cognitive computing, fuzzy sets, Rough sets, bio-inspired computing, machine learning, computer science, engineering, statistics, mathematics, physics, psychology and social sciences. Recently, many researchers have attempted to develop CI methods and algorithms to support the decision-making in different tasks and domains. There has been a recent research in the application of CI paradigms, approaches and techniques to address software engineering(SE) problems .CI offers smart models and intelligent algorithms that can contribute greatly to design formalization and automation. In this paper we clarify many important SE issues, review some of CI techniques and their applications and also highlight challenges.
Title of the Paper: Use of Artificial Intelligence Techniques to Determine Dental Caries: A Systematic Review
Authors: Romany F. Mansour, Abdulsamad Al-Marghilnai, Zahar A. Hagas
Pages: 1-11
Abstract: Background: Dental caries is a chronic pathological condition affecting an estimated 36% of global population in their permanent teeth. It is characterized by demineralization of hydroxyapatite crystals and destruction of collagen matter in dental tissues. Various conventional methods for early detection of dental caries are used by dentists all over the world, such as Visible Light- Enhanced Techniques, Electronic Conductance Measurements, Electrical Impedance Spectroscopy, Digital radiography, Laser Fluorescence System and Ultrasound Caries Detector. However, the shortcomings of these techniques alarms the need to adopt a better method for early detection of caries. Objective: The present study provides a systematic review of accuracy to use Raman spectroscopy as a method for caries detection at an early stage. Absence of sample penetration makes the method simple and hence is widely used. Methodology: Significant information related to Raman spectroscopy has been extracted and utilized in the presentation of systematic review paper. Various parameters have been taken into account, such as type of Raman spectroscopy, central wavelength, optical power, description of the system, scan rate, description of Raman micro-spectroscopy, Raman imaging, Raman peak, Peak intensities of laser polarization direction, Depolarization ratio and polarization anisotropy. Results and Discussion: The different research studies use varied system and central wavelength range that affects the result of each one of them. However, studies by M.T. Kirchner, et.al.(1997), Alex C.-T. Ko, et.al. (2006) and Alex C.-T. Ko, et.al. (2008) indicated maximum number Raman peak intensities corresponding to different functional groups, providing more information than other studies.
