Showing papers presented at "International Symposium on Industrial Electronics in 2017"

Deep neural networks for energy load forecasting

Abstract: Smartgrids of the future promise unprecedented flexibility in energy management. Therefore, accurate predictions/forecasts of energy demands (loads) at individual site and aggregate level of the grid is crucial. Despite extensive research, load forecasting remains to be a difficult problem. This paper presents a load forecasting methodology based on deep learning. Specifically, the work presented in this paper investigates the effectiveness of using Convolutional Neural Networks (CNN) for performing energy load forecasting at individual building level. The presented methodology uses convolutions on historical loads. The output from the convolutional operation is fed to fully connected layers together with other pertinent information. The presented methodology was implemented on a benchmark data set of electricity consumption for a single residential customer. Results obtained from the CNN were compared against results obtained by Long Short Term Memories LSTM sequence-to-sequence (LSTM S2S), Factored Restricted Boltzmann Machines (FCRBM), “shallow” Artificial Neural Networks (ANN) and Support Vector Machines (SVM) for the same dataset. Experimental results showed that the CNN outperformed SVR while producing comparable results to the ANN and deep learning methodologies. Further testing is required to compare the performances of different deep learning architectures in load forecasting.

Feature selection in UNSW-NB15 and KDDCUP'99 datasets

Abstract: Machine learning and data mining techniques have been widely used in order to improve network intrusion detection in recent years. These techniques make it possible to automate anomaly detection in network traffics. One of the major problems that researchers are facing is the lack of published data available for research purposes. The KDD'99 dataset was used by researchers for over a decade even though this dataset was suffering from some reported shortcomings and it was criticized by few researchers. In 2009, Tavallaee M. et al. proposed a new dataset (NSL-KDD) extracted from the KDD'99 dataset in order to improve the dataset where it can be used for carrying out research in anomaly detection. The UNSW-NB15 dataset is the latest published dataset which was created in 2015 for research purposes in intrusion detection. This research is analysing the features included in the UNSW-NB15 dataset by employing machine learning techniques and exploring significant features (curse of high dimensionality) by which intrusion detection can be improved in network systems. Therefore, the existing irrelevant and redundant features are omitted from the dataset resulting not only faster training and testing process but also less resource consumption while maintaining high detection rates. A subset of features is proposed in this study and the findings are compared with the previous work in relation to features selection in the KDD'99 dataset.

Understanding and personalising smart city services using machine learning, The Internet-of-Things and Big Data

Abstract: This paper explores the potential of Machine Learning (ML) and Artificial Intelligence (AI) to lever Internet of Things (IoT) and Big Data in the development of personalised services in Smart Cities. We do this by studying the performance of four well-known ML classification algorithms (Bayes Network (BN), Naive Bayesian (NB), J48, and Nearest Neighbour (NN)) in correlating the effects of weather data (especially rainfall and temperature) on short journeys made by cyclists in London. The performance of the algorithms was assessed in terms of accuracy, trustworthy and speed. The data sets were provided by Transport for London (TfL) and the UK MetOffice. We employed a random sample of some 1,800,000 instances, comprising six individual datasets, which we analysed on the WEKA platform. The results revealed that there were a high degree of correlations between weather-based attributes and the Big Data being analysed. Notable observations were that, on average, the decision tree J48 algorithm performed best in terms of accuracy while the kNN IBK algorithm was the fastest to build models. Finally we suggest IoT Smart City applications that may benefit from our work.

A big data based deep learning approach for vehicle speed prediction

Abstract: Vehicle speed prediction plays an important role in Data-Driven Intelligent Transportation System (D2ITS) and electric vehicle energy management. Accurately predicting vehicle speed for an individual trip is a challenging topic because vehicle speed is subjected to various factors such as route types, route curvature, driver behavior, weather and traffic condition. A big data based deep learning vehicle speed prediction algorithm featuring big data analytics and Adaptive Neuro-Fuzzy Inference System (ANFIS) is presented in this paper. Big data analytics examines copious amounts of speed related data to identify the pattern and correlation between input factors and vehicle speed. ANFIS model is constructed and configured, based on the analytics. The proposed speed prediction algorithm is trained and evaluated using the actual driving data collected by one test driver. Experiment results indicate that the proposed algorithm is capable of accurately predicting vehicle speed for both freeway and urban traffic networks.

Cyber-physical production systems' design challenges

Abstract: Recent advances in information technologies and artificial intelligence are enabling the creation of intelligent and highly reconfigurable factories which will lead to unprecedented production development. The notion of Cyber-Physical Production System (CPPS), denoting a system where mechatronic components are coupled to a smart logical entity that enables these factory units to interact in an adaptive way, has been presented as one of the cornerstones of what is perceived to be the 4th Industrial Revolution. However, more than two decades of research have shown that such a vision is not as new as recent programmes suggest and that certain developments are only now reaching a maturity stage that renders them usable with limitations. These are due to a set of persistent design challenges that undermine their acceptance along with the fact that the preconditions for operating such systems globally are far from being satisfied. This paper provides an integrated, clear and critical view on pending design issues that still need to be satisfied. In this context, the paper concentrates on the conceptual and technical design barriers that relate to the development of cyber-physical production components and their operation in the context of a CPPS.

Using smart meters data for energy management operations and power quality monitoring in a microgrid

Abstract: Smart metering devices have become an essential part in the development of the current electrical network toward the paradigm of Smart Grid. These meters present in most of the cases, functionalities whose analysis capabilities go further beyond the basic automated meter readings for billing purposes, integrating home or building area networks (HAN/BAN), alarms and power quality indicators in some cases. All those characteristics make this widely spread equipment a free, accurate and flexible source of information that can replace expensive and dedicated devices. Therefore, this paper presents the integration of a commercial advanced metering infrastructure (AMI) in the context of a smart building with an energy management system (EMS). Furthermore, power quality monitoring based on this AMI is explained. All the details regarding the implementation in a laboratory scale application, as well as the obtained results, are provided.

Model and energy management system for a parallel hybrid electric unmanned aerial vehicle

Abstract: In this paper is presented a complete Simulink model and a control strategy for the energy management of a parallel hybrid electric UAV (Unmanned Air Vehicle) powertrain. The model consists of an internal combustion engine, a gearbox (which includes a planetary gear and a continuously variable transmission), an electric motor, which can work also as a generator, an electric drive (Inveter) and a Li-Po battery pack. The control strategy propose a near real-time iterative algorithm based on Dynamic Programming to solve an optimization problem for the optimal power management and torque-split of the powertrain with final state constraints on state variable. Aim of the study is to investigate new flight capabilities that derives from the use of a hybrid architecture, like silent mode using only electric motor, useful for military and civil purpose, and the capacity of reducing fuel consumption. Simulation studies are based on data of an existing UAV and a real flight mission.

Machine vision system errors for unmanned aerial vehicle navigation

Abstract: The use of laser scanners as machine vision systems in Unmanned Aerial Vehicle (UAV) navigation offers a wide range of advantages, when compared with camera-based systems. As one advantage, the measurement of real physical distances can be mentioned, which results in a reduction of measurement times and thereby fast image processing of the UAV surrounding medium. In previous work, a novel laser scanner namely Technical Vision System (TVS) was presented, which implements a continuous laser scan for determination of 3D coordinates of any object under observation. Also previous work has shown the advantage, when using high-quality instead of low-quality DC motors as actuators for positioning the laser ray in the TVS field-of-view. However, the static friction in the ball bearings of the motor shaft leads to a residual error not zero for the angular error in steady state. Present paper introduces a new approach of estimating this residual error by use of a friction model. Thereby, the friction model is described and used for simulation of the residual angular error when controlling the DC motor in open-loop configuration and validated by experimentation results.

Analysis of the requirements for offering industrie 4.0 applications as a cloud service

Abstract: Industrie 4.0 introduces a concept of digitalized production by allowing agile and flexible integration of new business models while maintaining manufacturing costs and efficiency at the reasonable level. In addition, cloud computing is one of the IT trends that is used nowadays to offer services on demand from a virtual environment in enterprise and office areas. The use of cloud computing in an industrial automation domain in order to offer on-demand services, such as alarm flood management or control as a service, is a promising solution. This study examines how the cloud-based applications can meet the Industrie 4.0 requirements concerning security, communication, self-configuration, reliability, and asset administration shell. Moreover, research challenges and existing gaps that need further investigation are identified and discussed.

A battery energy management strategy for UK enhanced frequency response

Abstract: Balancing the grid at 50 Hz requires managing many distributed generation sources against a varying load, which is becoming an increasingly challenging task due to the increased penetration of renewable energy sources such as wind and solar and loss of traditional generation which provide inertia to the system. In the UK, various frequency support services are available, which are developed to provide a real-time response to changes in the grid frequency. The National Grid (NG) — the main distribution network operator in the UK — have introduced a new and fast service called the Enhanced Frequency Response (EFR), which requires a response time of under one second. A battery energy storage system (BESS) is a suitable candidate for delivering such service. Therefore, in this paper a control algorithm is developed to provide a charge/discharge power output with respect to deviations in the grid frequency and the ramp-rate limits imposed by the NG, whilst managing the state-of-charge (SOC) of the BESS for an optimised utilisation of the available stored energy. Simulation results on a 2 MW/1 MWh lithium-titanate BESS are provided to verify the proposed algorithm based on the control of an experimentally validated battery model.

Architecture for security monitoring in IoT environments

Abstract: The focus of this paper is to propose an integration between Internet of Things (IoT) and Video Surveillance, with the aim to satisfy the requirements of the future needs of Video Surveillance, and to accomplish a better use. IoT is a new technology in the sector of telecommunications. It is a network that contains physical objects, items, and devices, which are embedded with sensors and software, thus enabling the objects, and allowing for their data exchange. Video Surveillance systems collect and exchange the data which has been recorded by sensors and cameras and send it through the network. This paper proposes an innovative topology paradigm which could offer a better use of IoT technology in Video Surveillance systems. Furthermore, the contribution of these technologies provided by Internet of Things features in dealing with the basic types of Video Surveillance technology with the aim to improve their use and to have a better transmission of video data through the network. Additionally, there is a comparison between our proposed topology and relevant proposed topologies focusing on the security issue.

DC link capacitor optimization for integrated modular motor drives

Abstract: In this paper, selection of optimum DC link capacitor for Integrated Modular Motor Drives (IMMD) is presented. First, a review of IMMD technologies is given and current research and future prospects are studied. Inverter topologies and gate drive techniques are evaluated in terms of DC link performance. The urge for volume reduction in IMMD poses a challenge for the selection of optimum DC link capacitor. DC Link capacitor types are discussed and critical aspects in selecting the DC links capacitor are listed. Analytical modeling of DC link capacitor parameters is performed and it is verified by simulations conducted using MATLAB/Simulink. Optimum selection of DC link capacitor is achieved based on the electrical, thermal and economical model.

SmartFingerBraille: A tactile sensing and actuation based communication glove for deafblind people

Abstract: This paper presents a smart tactile sensing and actuation-based finger-glove to be used by visual and hearing impaired people for communication and learning. It is based on the concept of finger Braille and supports both face-to-face and distant communication. The device comprises of a smart finger-glove worn on both hands in the ring, middle and index finger and communicates with mobile devices using Bluetooth technology. Six tactile sensors and actuators each were used and a pair embedded in the glove in the ring, middle, and index fingers of both hands to represent the six dots of the Braille code. The user wears the smart finger-glove and taps the correct finger combination corresponding to the Braille code on any surface to compose and send messages to mobile devices or other gloves. Messages can equally be received with the glove from a mobile device in the form of vibrations on the fingers corresponding to the Braille codes. The facile and low-cost nature of the device will make it attractive to Braille users and makes learning Braille an entertainment activity and could spur the interest of sighted people also to learn Braille.

Tensor product-based model transformation for position control of magnetic levitation systems

Abstract: This paper presents an application of the Tensor Product-based model transformation to the real-time position control of magnetic levitation systems. Three cases that depend on the number of singular values are presented. All case studies are validated by experiments conducted related to the sphere position control of a laboratory magnetic levitation system.

Innovative learning in engineering education: Experimenting with short-term project-oriented research and project-based learning

Abstract: This paper presents a reflection on the results of an experimental curricula design which combined short-term project-oriented research and project-based learning in the field of electrical engineering and renewable energy. The experiment was based on the application of the Project-Based Learning methodology. The problem to be solved arose from a local company and was carried out at a master level in the Polytechnic Institute of Braganca in Portugal. This experiment aimed to contribute to the innovation and modernization of educational projects of professionally oriented programmes in higher education institutions. It also demonstrated some of the advantages and challenges of learning innovation. Feedback from the students was encouraging but the results of this experiment mainly pointed to the need that everyone must be prepared for change. The study contributes to understanding both why and how we need to foster educational shift in electrical engineering to educate students and enhance their engineering skills for the twenty-first century.

Li-ion battery parameter identification with low pass filter for measurement noise rejection

Abstract: The advent of Energy Management (EM) and Electric Vehicles (EV) have completely changed the use of batteries. Accurately estimating the remaining power in batteries has become increasingly important. In order to estimate precise battery state of charge (SOC)/state of health (SOH) value, accurate parameter identification is essential when constructing an accurate battery model. Even though we are able to exactly identify battery parameters offline, the precision of online parameter identification usually suffers from measurement noise, which is an unavoidable phenomenon. In this paper we investigate how battery parameter identification is influenced by measurement noise. The selection of a low pass filter is also discussed and a fourth order Butterworth filter is adopted to effectively reject high frequency measurement noise. This algorithm can help with the identification of battery parameter that rejects measurement noise and maintains the accuracy of online battery parameter identification for future online model-based battery SOC/SOH estimation.

Peer-to-peer decentralized control structure for real time monitoring and control of microgrids

Abstract: In order to integrate a large number of distributed energy resources in distribution grids a robust decentralized information and communication control structure is required. This paper proposes an overlay peer-to-peer (P2P) architecture for controlling and monitoring microgrids in real time, which has a great capacity of adaptation to the demanding network requirements of these environments. The proposed concept has been implemented and experimentally tested on a microgrid. Real data analysis demonstrates that the system meets with the network performance parameters proposed for microgrids, such as latency and bandwidth, showing that peer-to-peer overlay networks are useful for energy grids in practice.

Seamless handover in industrial WLAN using IEEE 802.11k

Abstract: In the era of increasing connectivity driven by IoT and CPS concepts, wireless networks are gaining importance in industrial applications. One of their obvious benefits is the support of mobile clients. In complex network infrastructures consisting of many access points, roaming becomes an important aspect because many applications rely on realtime communication and therefore need seamless handover between access points. In IEEE 802.11 WLANs, however, the roaming today is mostly reactive, i.e., a connection is kept as long as the quality of the link provides the possibility to transmit data. If the quality decreases too much the connection is lost, and the client must scan the wireless channels and look for a new access point in order to establish a new connection. The scanning procedure typically takes a long time and thus prohibits seamless roaming. This paper investigates a simple, proactive roaming method using the IEEE 802.11k standard to collect information about the wireless environment prior to roaming. Actual handover is then based on an assessment of the signal strength (RSSI). Experimental results show that the proposed method can reduce the handover time by a factor 30.

Wind speed and direction measurement based on time of flight ultrasonic anemometer

Abstract: In this paper, an ultrasonic anemometer based in the time of flight is presented. The wind speed and direction is obtained using four ultrasonic transducers in an orthogonal configuration. Two transducers are used in each orthogonal direction as transmitter and receiver of a short sequence of pulses and the time of flight is recorded in both ways. The system is based on a low-cost Arduino-Nano board and the overall system was tested in a wind tunnel, where accuracy and linearity were analyzed. The developed anemometer is intended to be used in a scale model sailboat to support its autonomous navigation.

Behavior of the flying capacitor converter under critical operating conditions

Abstract: The Flying Capacitor Converter (FCC) offers an attractive alternative to conventional 2-IeveI converter topologies due to the easily acquired high number of voltage levels and the increased effective switching frequency. However, balancing of the flying capacitor (FC) voltages is crucial in practice since a deviation from the nominal voltage levels increases harmonics in the output voltage and, more importantly, jeopardizes the integrity of the converter due to overvoltages across the power transistors. Modulation inherent FC balancing techniques (termed natural/passive balancing) have been thoroughly analyzed in literature, however only for stationary operating conditions. In this paper, the behavior of the FCC and the effectiveness of passive balancing will be analyzed in detail regarding specific operating conditions present in typical industry applications such as converter start-up, shut-down, standby and operation under fault conditions. The basis for the analysis is a 5-level, 2 kW FCC embedded in two typical industry applications: single-phase PV inverter and single-phase PFC rectifier.

Case-based reasoning approach applied to surveillance system using an autonomous unmanned aerial vehicle

Abstract: The advanced use and the evolution of technologies regarding autonomous Unmanned Aerial Vehicle (UAV) have increased the availability of information and resources to perceive the environment, allowing its application in various activities, such as inspection and military. However, the intelligence level of these kind of systems needs to be improved in order to fit them in modern tasks. In this sense, this work proposes a high-level processing approach to be embedded in UAV system for understanding human activities in real environment. Additionally, the Case-Based Reasoning (CBR) methodology is also applied to allow the adaptation of the flight plan and the fully autonomous surveillance in limited areas. In order to enhance the solution, the proposed architecture is inspired in the biologic model of the human cognitive system and comprises low, middle and high levels to enable the perception of the environment as well as comprehension of the scene. The experiments have shown technical feasibility and effectiveness of the architecture. Moreover, the use of UAV has reduced the number of cameras and operators, being also capable to reach difficult areas.

A multi-output capacitive charger for electric vehicles

Abstract: Multi-output converters are widely used in various electronic devices and Electric Vehicles (EVs) due to their benefit in terms of cost, efficiency and space for installation. On the other hand, Wireless Power Transfer (WPT) methods including Inductive Power Transfer (IPT) and Capacitive Power Transfer (CPT) have been applied increasingly in EVs since they are more convenient and safe as compared to conventional conductive chargers. Due to replacing copper wires and permeable materials of inductive coupling method with cheap metal plates, implementing a CPT system is more cost effective and structurally simple system as compared to IPT. However, researches on multi-output chargers relating to CPT are rarely presented. This paper proposes a new concept for a multi-output charger system by adopting CPT for EVs application. The proposed system can charge independently several batteries of EVs at one time by adopting only one full bridge inverter at the primary side. The mathematical analysis supported by simulation results are presented.

A novel single-phase five-level active rectifier for on-board EV battery chargers

Abstract: This paper presents a novel single-phase active rectifier for applications of on-board EV battery chargers. The proposed active rectifier, with reduced number of semiconductors, is constituted by four MOSFETs and four diodes, and can produce five distinct voltage levels, allowing to reduce the passive filters used to interface with the electrical power grid. An almost sinusoidal grid current with unitary power factor is achieved in the grid side for all the operating power range, contributing to preserve the power quality. The principle of operation, the current control strategy and the modulation technique are presented in detail. Simulation results in different conditions of operation are presented to highlight the feasibility and advantages of the proposed active rectifier.

A methodological use of inertial navigation systems for strapdown navigation task

Abstract: This paper presents a study of mathematical description for inertial navigation systems and integration of virtual sensors implementation. Virtual sensors allow to estimate quantities detecting events or changes in its environment, to calculate variables such velocity, position and attitude on rigid or mobile bodies of navigation systems. Recently, the creation of smaller devices and lightweight micro-machined electromechanical systems (MEMS) has opened the possibility to create artifacts that solve the positioning and orientation problems of navigation systems. This has increased the interest on topics of inertial navigation systems, typically in areas as artificial vision and robotics, especially on mobile robotics. The basic idea of mobile robotics is defined by transportation to reach an object or a specific position. Therefore it is necessary to provide environment value in real time for interpretation. This paper presents an implementation of navigation systems and a design of a Strapdown Inertial Navigation System (INS). Consequently, a simulation is conducted to verify and compare the performance between the implemented INS equations and a MATLAB SIMULINK 6DoF Euler Block. Furthermore an analysis of the simulation results is performed.

A vision system for traffic sign detection and recognition

Abstract: The paper presents an automatic traffic sign recognition system using the videos recorded from an on-board dashcam. It is based on image processing, bilateral Chinese transform, and vertex and bisector transform techniques. The images captured from the dashcam are processed with the histogram of oriented gradients to form feature vectors, followed by support vector machines to detect the traffic signs. The bilateral Chinese transform and vertex and bisector transform are used to extract the area of traffic sign from images. Finally, a neural network is adopted to identify the traffic sign information. In this work, we test the algorithms using the images captured from the camera mounted behind the front windshield. The experiments are evaluated with real traffic scenes and the results have demonstrated the effectiveness of the proposed system.

Mobile robot navigation in natural environments using robust object tracking

Abstract: In this paper the authors introduce a method focusing on the robustness improvement of the landmark tracking system for mobile robot operation in natural environments. We extract feature points from the data obtained by a stereo vision system with CenSurE (Center Surround Extremas for Realtime Feature Detection and Matching) used as a detector, and FREAK (Fast Retina Keypoint) as a descriptor. RANSAC (RANdom SAmple Consensus) is used to remove outlier data from the feature points in order to increase precision. For self-localization, landmarks are selected from the surroundings. These landmarks are tracked by a template matching method using ZNCC (Zero-Mean Normalized Cross-Correlation) complemented with visual odometry based motion estimation. For performance purposes, this is combined with UKF (Unscented Kalman Filter) for narrowing the landmark search areas. A template update strategy suitable for long range tracking is also introduced. Finally, for increasing robustness in long range operation, we solve the issue of obscured/temporarily out of frame landmark tracking by estimating their position based on nearby visible landmarks.

Identifying vulnerabilities in smart gric communication networks of electrical substations using GEESE 2.0

Abstract: The use of Intelligent Electronic Devices (IEDs) in electrical substations has brought a great improvement in terms of speed, costs, maintenance, and reliability. In this context, the use of IEC 61850 standard appears as one of the main recommendations for substation automation in smart grids that digitally perform the electrical network control and protection. Nevertheless, this improvement may be prone to a number of security issues. Therefore, security threats in substation networks must be detected in order to improve cyber security solutions. GEESE is a command-line packet generator software following the IEC 61850 standard. In this work, we propose GEESE 2.0, an extension to GEESE that provides a graphical user interface that easily allows substation cyber security threat evaluation. Moreover, GEESE 2.0 can sniff, capture, modify and send GOOSE packets on the network, so that it can create different attacks in order to evaluate the substation security. We have locally deployed a substation network emulation and used GEESE 2.0 to perform network attacks in order to analyze the impact of an attack in a substation protection scheme.

SRM converter topologies for EV application: State of the technology

Abstract: The popularity of electric vehicles is still growing. The investment in R&D has been focused on the development of new and cheaper motors like Switched Reluctance Machines (SRM). This article presents an overview of SRM converter topologies critically compared for EV application in terms of controllability, costs, fault tolerance and complexity. The studied topologies are classified into two groups: hard-switching and soft-switching. But also, other features like the amount of components (switches, diodes, etc.) and modes of operation are taken into account.

Towards predictable execution model on ARM-based heterogeneous platforms

Abstract: Many today's real-time applications, such as Advanced Driver Assistant Systems (ADAS), demand both high computing power and safety guarantees. High computing power can be easily delivered by, now ubiquitous, multi-core CPUs or by a heterogeneous system with a multi-core CPU and a parallel accelerator such as a GPU. Reaching the required safety level in such a system is by far more difficult because the commercial-of-the-shelf (COTS) high-performance platforms contain many shared resources (e.g. main memory) with arbiters not designed to provide real-time guarantees. A promising approach to address this problem, known as PRedictable Execution Model (PREM), was introduced by Pellizzoni et al. [1]. We are interested in applying PREM to ARM-based heterogeneous platforms, but so far, all PREM-related work has been done on x86 or PowerPC. In this paper, we introduce several building blocks that are needed for implementing PREM on NVIDIA Tegra X1 platform. We propose a modification of the MemGuard tool to be practically usable on ARM platforms. We also analyse a throttling mechanism of Tegra X1 memory controller, that allows controlling memory bandwidth of non-CPU clients such as the GPU. We show that this mechanism can be used to make the execution time of CPU tasks more predictable.

Real-time continuous action detection and recognition using depth images and inertial signals

Abstract: This paper presents an approach to detect and recognize actions of interest in real-time from a continuous stream of data that are captured simultaneously from a Kinect depth camera and a wearable inertial sensor. Actions of interest are considered to appear continuously and in a random order among actions of non-interest. Skeleton depth images are first used to separate actions of interest from actions of non-interest based on pause and motion segments. Inertial signals from a wearable inertial sensor are then used to improve the recognition outcome. A dataset consisting of simultaneous depth and inertial data for the smart TV actions of interest occurring continuously and in a random order among actions of non-interest is studied and made publicly available. The results obtained indicate the effectiveness of the developed approach in coping with actions that are performed realistically in a continuous manner.