29th Annual Technical Conference Research Papers

Abstract – Thermal power plants; coal-fired steam, combined cycle, gas turbines, and reciprocating engines serve a large portion of the electricity demand in Sri Lanka, while large and small hydropower plants, and converter-and-inverter-based generation such as wind and solar, serve the balance. The power system of Sri Lanka is islanded and smaller in size, with a lesser amount of connected synchronous machines, compared with larger power systems. Accordingly, the power system of Sri Lanka can be classified as a low-inertia power system. Even before Non-Conventional Renewable Energy (NCRE) additions, the power system was largely dependent on underfrequency load shedding (UFLS) to recover after a disconnection of a large generator. Introducing more NCRE to the power system has worsened the situation further. On the path to achieving 70% of the renewable energy target by 2030, higher penetration of solar and wind generation can be expected soon. Apart from the advantages of renewable dominance, Sri Lanka would face additional challenges such as maintaining frequency stability due to reducing power system inertia. This study is on possible methodsto improve frequency stability. The possibility of using grid-scale BESS to improve frequency stability was studied through dynamic modelling. Costs were analyzed and comparative costs of different BESS utilization scenarios were evaluated using forecast BESS costs. Finally, the technical and economic viability of BESS for improving the frequency stability is presented and discussed.

Keywords—Battery Energy Storage Systems, Frequency Response, Non-conventional Renewable Energy, PSS/E

Authors —Nilan Hemachandra, Asanka S. Rodrigo,Tilak Siyambalapitiy

Abstract – Integration of distributed generators into power systems is enhanced immensely due to the high demand for energy consumption which causes huge problems in power system operations. One of the adverse effects is reactive power variation which leads to insecure and unbalanced operation of the power system. Shunt capacitors and Static VAR compensators are some existing reactive power compensation equipment to regulate reactive power in power systems. Instead of using traditional approaches, a novel reactive power optimizing method will be introduced to ensure the safe and reliable operation of the system. The key objectives are to reduce the active power loss and enhance the power quality of the system while maintaining the voltage profile within the standard limit. To achieve the goals, a Genetic Algorithm based solution will be proposed for the IEEE-39 bus test system. In this paper, a novel reactive power optimization model is simulated using a standard genetic algorithm-based test system. The final simulated outcomes will illustrate that the proposed algorithm is attainable and effective by enhancing the quality of power systems.

Keywords—Reactive Power, Genetic Algorithm, Power Systems, IEEE 39 Bus System

Authors —J.A.D.C.A.Jayakody,Edirisinghe G.A.D.D.Ganepola

Abstract – The paper presents a method to segment a time series by developing an estimate for the trend between two points in time and an estimate of the variation in the difference between the two points as a variation around this trend. The trend is estimated between two points in the time series and relates the amount of linear change between the points to the number of observations between them. In this paper, a time series of stock prices is used to estimate and detect changes in the trend. The variation in the prices provide information about the uncertainty in the decisions taken by the investors and the risks they faced at the time. The model provides the means to estimate future values in the time series and control the uncertainty or risk in that estimate measured as the variation around the trend by choosing the interval or time period over which to make the estimate. This model of a time series and trend is used to estimate the amount of information conveyed by the prices measured in an interval as the variation in the prices in that segment. The period over which a future value in the time series is estimated determines the accuracy in its estimate and the variation or uncertainty in the estimate leading to the observation that future values are best estimated over periods where past trends can be extended or over periods where future behavior is expected to be similar to the past.

Keywords—Information; risk; random walk; insert (key words)

Authors —Asoka Korale,

Abstract – Extensive adaptation of intelligent automation systems for industrial applications can be observed in the modern era. Nevertheless, best results can be obtained only when all these contemporary automation technologies are incorporatedtogether. Hence, it is essential to have a broad understanding about the integration of all these smart automation systems for future developments and expansions in industrial applications. Thus, the authors have developed a novel laboratory learning facility for teaching students about Integrated Intelligent Automation Systems as a prototype of a chocolate manufacturing factory. In this setup, several modern day technologies such as controllers, Android applications, Internet of Things(IoT), Cloud Services and Artificial Neural Networks (ANN) are utilized together for handling the customer orders for chocolates. In this learning facility, several tasks based on these automation technologies are assigned for students to be completed with in an allocated time . The effectiveness of this laboratory Learning Facility is tested through a survey of several questions presented as a questionnaire at the end of practical sessions. The analyzed responses from students justify the effectiveness of this learning platform in terms of achieving the learning objectives.

Keywords—Laboratory learning facility, automation, PLC controllers, android application, IoT & cloud service, ANN

Authors —M. M. S. M. Gunasena, N. H. R. H. Perera , K. B. J. Anuradha,P. M. S. Indrajith,R. M. R. Priyadarshani, A. G. B. P. Jayasekara

Abstract – In this paper, we study designing and implementing a high-performance IoT sensor node using a popular Raspberry Pi Single Board Computer (SBC). Modern IoT applications use multiple sensors to sense different physical parameters, including camera sensors. Connecting multiple sensors in a single node drastically reduces the computing power and uptime of the node. Also, it may not get expected performances when it works in real-time. During the design, multiple Raspberry Pi boards are interconnected as a cluster to improve the system’s computing power. And a parallel processing algorithm called Massage Transfer Interface (MPI) to share the load within the node. The performances of the cluster have been tested, and results are presented.

Keywords—Raspberry Pi Clustering, Internet of Things, Sensor node design, Massage Transfer Interface (MPI)

Authors—YR Samarawickrama, NT Jayatilake, MWP Maduranga,Ashen Wanniarachchi,WMSRB Wijayarathne

Abstract – Transient instability is one of the most severe types of power system instability, which have serious socioeconomic consequences if avoided. Fast on-line transient stability assessment in modern power systems is limited by traditional methods such as time-domain simulations and direct methods. The invention of phasor measurement units has created the path for artificial intelligence-based pattern detection and categorization for transient stability assessment. There are several categorization techniques for measuring transient stability have been documented in the literature. This research seeks to provide information on which algorithm is best for determining power system stability for a specific dataset. In a comparative examination of datasets, neural networks, support vector machines, and deep learning are evaluated for their capacity to address the binary stability classification problem. To simulate an IEEE-12 bus test system, the above datasets were constructed using MATLAB and Simulink.

Keywords—Transient stability assessment, neural network, support vector machine, deep learning

Authors —J.Birannaa, K.V.D.M.Dineatth, A.Nithurshan ,W.D.Prasad

Abstract -— This research project is designing and fabricating a Wireless Beach Cleaning Robot. The coastlines of Sri Lanka are one of the country’s most popular tourist destinations. That is polluted due to irresponsible human behaviors. Although the Sri Lankan local government institutions have cleaned up the coast, some areas are neglected for various reasons. Cleaning is complex and requires a considerable amount of time and resources. Sewage sedimentation caused by coastal breezes has become a significant issue. That makes identifying the contamination challenging. Workers find itdifficult to clean up as they dig the beach to collect that waste.Some organizations and government agencies take steps to remove the debris accumulated along the coast effectively. Many local and foreign tourists are often attracted to the beach for relaxation. People usually throw plastic on the beach without knowing the consequences. It dramatically affects the marine environment. The design of the robot uses wireless technology, such as Radiofrequency applications and the Internet of Things (IoT). The source of power for the robot is the 12VDC lithium-ion polymer battery. This system uses the real-time dashboard to monitor the level of the garbage bin. An IoT system is an interconnected network of intelligent devices that can sense and communicate with other systems. The Arduino IDE gathers the sensor data from the Ultrasonic sensors. The robot also indicates the level of the garbage bin through the blynk mobile application. It helps an operator to monitor the garbage bin level. This research workaims to design and fabricate a wireless beach cleaning robot with less human effort. The machine is constructed with a simple, economical design for easy maintenance and use. The machine is environmentally friendly and can operate in any beach condition.

Keywords—Beach pollution, Internet of Things, Beach Clean, Cleaning robot

Authors —H.M.R.G. Herath, G.C.B. Weerasinghe, M.W.S.B. Jayasooriya, K.K.N. Dharmarathna

Abstract -— This paper introduces an IOT based hydroponic planting box solution which is called as Grow N Glow. This can be controlled wirelessly from anywhere in the world with the concept of IOT. This is complete IOT solution with sensors and actuators that can be controlled by any smart device, desktop, or laptop. This developed solution includes light control, exhaust controls, water pump controls, oxygen pump controls nutrition pump controls and so on. As the result complete hydroponic system was developed. Using this solution lot of advantages can be achieved. It will maximize the space, saves water, maintain a microclimate for the requirements of the plants. This solution does not use soil which is a major advantage because there will be no harm for the plants or crops. The crops will grow fast, and the harvested crops will be fresher and healthier comparing to normal crops. There are some vegetables that require specific conditions or else they will not grow. With this solution common plants can be grown regardless of the weather conditions.

Keywords—IOT (Internet of things), Wireless, Hydroponics

Authors —Thiwanka Cholitha Hettiarachchi, Shashika Lokuliyana, Anuradha Jayakody

Abstract -— In Sri Lanka, traditional streetlamps currently consume 150 Gigawatt-hours of electricity every year. This is a significant amount of electric energy that has been wasted. To overcome this problem Multifunction Smart IoT Streetlamp system introduces as the viable solution. The system is designed to save electric energy. This system can be monitored and controlled remotely. For that, the traditional streetlamp system is replaced by using modern smart streetlamps. In this streetlamp system, lights can be turned on when needed off when not needed. In addition, the brightness of the lights can be controlled according to the requirements which increase the lifespan of the light, and the color temperature of the lights also can be controlled according to the requirements for ease of public. The proposed smart streetlamp is also equipped with a defect detection program that notifies the management dashboard of any defects. The smart streetlamp which will be introduced will communicate via the network and connect to the cloud through metropolitan Wi-Fi. Also, this system will ensure public safety by controlling the behavior of the streetlamp according to the environment surrounding conditions.

Keywords—smart streetlight, remotely, IOT, energy, brightness, Municipal Wi-Fi, cloud.

Authors —Thiwanka Cholitha Hettiarachchi, Lakisuru Sathyajith Semasinghe, Shashika Lokuliyana, Anuradha Jayakody

Abstract -— For the visually impaired, objects such as texts and symbols are difficult to be recognized and understood without additional help. Though the Braille method has been the solution for this for a long time, there are numerous instances and applications where it has proven to be slow or impossible since it incorporates the person’s sense of touch. This project to design and develop a wearable text reader for visually impaired people is suggested as a solution to this problem. The system is intended to read any text using a camera-based Raspberry pi device to provide an audio output for the visually impaired person. The system, which will consist of mainly two parts, will first identify the text in a particular image using an OCR engine to provide an audio output of the text finally. The Optical Character Recognition (OCR) algorithms and the image processing function of the system were given priority in this project. To derive a suitable algorithm, many text recognition algorithms were analyzed and implemented on the python IDE against different kinds of images such as invoices, digital images, and handwritten characters. Pre-trained modelswere then used with these algorithms to increase the system’s efficiency. Some image pre-processing techniques were adopted to improve the accuracy of the OCR. An image data set of 62992 images with 128x128px resolution containing characters, numbers, and symbols with four kinds of font styles was created to train a unique model for the OCR. The data set was trained using the Tensorflow machine learning framework, and technologies such as deep learning and image processing were mainly incorporated in the development of thesystem. The focus of this project is character recognition and conversion. The accuracy of the OCR will be effectively evaluated using proper and accepted testing methods. The Tesseract algorithm was tested at 94.93% accuracy, and the results are presented in this paper.

Keywords—Machine Learning, Deep Learning, Image processing, Wearable devices, Optical Character Recognition

Authors —MWP Maduranga, NT Jayatilake, PADV Pannala, SPARS Jayathilake

Abstract -— In this paper, an effective way of feeding solar power to the grid is discussed. Voltage is kept constant at grid voltage when the irradiance is varying and the current fed to the grid is varied accordingly. Maximum Power Point Tracking (MPPT) is used to obtain the maximum power point and the boost converter works according to the output given by the MPPT. The perturb and observe algorithm is used for the MPPT. There are six Insulated-Gate Bipolar Transistors (IGBTs) in the inverter and one PWM signal input for each IGBT to switch those. The PWM signals are generated by the control circuit to obtain a sinusoidal output. The inverter output is fed to a LCL filter to deplete the harmonics. To synchronize the frequency and the phase of the system with the grid, the voltage and current are measured from the grid side and they are fed to a phase locked loop. For the control purpose, the d-q frame and αβ0 frame are used. Signals of d and q axis currents and voltages are fed to the current control blocks from where the PWM signals required for the inverter are produced.

Keywords—Photovoltaic; Maximum power point tracking; Perturb and Observe, Two-stage; Grid-connected.

Authors—N.S. Hasaranga, W.Y.U.N. Botheju, N.S.A.D.S. Nanayakkara, J.A.R.R. Jayasinghe

Abstract -— Today automation technology is a trend that is easy to access, user friendly and easy to observe. Having pets can be a joy in our lives. Feeding pets is a major task and the purpose of this article is to automatically feed the pet when the owner is away from home. Over the past few years, many researchers have developed an automatic feeding machine, but their functionality is limited to the home. This advanced machine can provide real time food to the animal and monitor the animal’s behavior. The IoT Based Automatic pet food feeder uses ESP32 CAM camara module, Ultrasonic sensor, Motor control module, and consists of an interface with DC servo motor, and other hardware equipment. A software code is dumped into the micro controller to perform operations of ultrasonic sensor, rotating motors. The web base application developed by the PHP and HTML programming language and the real time data link with firebase database. The whole feeder system is controlled using a mobile phone. The user sends signals to the micro controller using mobile application through cloud. When the DC servo motor runs, the motor rotates the propeller which is in the feeding device. Using of ultrasonic sensor to get pet foods scale on real time. ESP32- cam to get video input to web application and could watch the behavior of the pet.

Keywords—Internet of Things; Pet Food Feeder; Real time monitoring

Authors—D.L.S.T. Jayarathne, H.K.A. Jayasinghe, H.M.R.G. Herath