Kaikki aineistot
Lisää
Methods for nowcasting lightning using weather radar data were developed using machine learning models. Reflectivity was selected as the main feature for the prediction. The purpose was to examine if machine learning applications could be used to nowcast thunderstorms with minimal data sets. The emphasis was to find out a model which is based on binary image classification and doesn’t require large sets of training data to work sufficiently. Convolutional neural network was the first choice. Accuracy for the model was 0.83. Another approach was made using random forest model. Precision for class 0 (no lightning) was 0.52, and for class (recorded lightning) 1, 0.90 and with total accuracy of 0.88 To improve the sets more features should be used and possibly larger data sets.
This study presents a comparative analysis of two prominent technologies, namely deep learning, and machine learning, in the context of weather forecasting. The main research question is “How can machine learning and deep learning algorithm be implemented to obtain near-accurate weather forecasting”? The objectives of this research are identifying the fundamental differences between deep learning and machine learning algorithms handling weather-related dataset and to ascertain the accuracy of using deep learning as compared to machine learning in weather forecasting. The study begins by providing a detailed overview of deep learning and machine learning techniques, explaining their fundamental principles, and highlighting their respective imple-mentation in weather dataset. In addition, the focus of the research is on the application of technologies such as polynomial regression, gradient boosting, neural prophet, and recurrent neural network models to the process of weather forecasting. The study applied quantitative methodology and used an open-source dataset from Finnish Meteorological Institute which is a weather record collect-ed from the city of Vaasa. The comparative analysis involves employing those techniques to capture nonlinear relationships between weather variables and the pattern within the dataset. Moreover, the study investigates the performance of each technology and evaluates its effectiveness in forecasting weather conditions over different interval of time using performance evaluation matrices. The outcomes of the comparative analysis provide valuable insights into the application of recent machine learning and deep learning methods with regard to the quality and the amount of data applied for the process. This includes proper implementation of data pre-processing techniques, that significantly impact the accuracy of models.
The aim of this thesis is to create a web application which will update weather forecast for Coxsbazar and generate alert. In addition, this thesis covers how weather forecasts are interpreted, how alerts are created and how the users will benefit from these alerts. Before creating this web application, a preliminary design was created and after a thorough analysis technologies were selected. Angular was used as the framework with Highcharts as secure data visualization library. For mapping leaflet with OpenStreetMap was used. After the implementation of this analysis the aim of this thesis was met. The results of this thesis will make life easier for individuals living around the coast by providing five days of weather forecasting alerts. One of the most important benefits of this web application is that users will be aware of inclement weather and will be able to mitigate harm.
The thesis is dedicated to the research of different Python libraries, exploring their functionalities and limitations with the focus on their general use and in meteorological area. As an outcome of this research, the list of the libraries used in meteorology will be presented with comparison by functionalities as well as a small demo of visualized meteorological data using one or more libraries. The theoretical part includes research on Python as a programming language, the history and practices of data visualization in general and of meteorological data in particular. The empirical part consists of exploring the Python’s libraries for visualization in general and in meteorological data. The practical part shows how meteorological data can be visualized using one or more of the libraries. The outcomes of the thesis can be used by those who are generally interested in Python libraries and want to start using them for visualization as well as those who want to learn more about meteorological data visualization.
The present study consisted of the development of a digitalized method for X-band weather radar testing and calibration processes, in the field of electrical engineering. To achieve the outcome the subject was studied to understand the requirements of different stakeholders related to the weather radar customer delivery projects. Due to the nature of this project, which has been developed internally at an internship in Vaisala Oy during the summer of 2022, some names of the programs and key names have been modified, and some confidential information has not been mentioned at all like reference documents. At the end of the project, there was a design process which led to the final execution of the weather radar test and calibration processes with the chosen system configuration. Based on the research and study, there was a conclusion on the most suitable procedure and the creation of an electrical procedure instruction, which was tested in a real production environment and implemented afterwards. Some other phases during the development of the project were collecting feedback and making necessary adjustments and root improvements to the procedure.
Despite the increase in marine traffic in the Polar region, high-quality forecasting has been lackluster in comparison to other areas of the planet. In an effort to improve this situation, the Polar Prediction Project started the multi-year long project named Year of Polar Project (YOPP). FINYOPP is a research project managed by the Finnish Meteorological Institute (FMI) following the objectives of the parent YOPP project. The main aim of this thesis is to research the feasibility of providing forecasts and observations data to vessels operating in the Arctic region. This is carried out by cooperation among Navidium Plc, FMI and KNL Networks all of whom operate in Finland. The forecast delivery network is provided by KNL, software development is handled by Navidium and FMI provides forecasts as well as satellite observations in addition to supervising this project. This thesis then describes various development stages of the project development which started with analyzing the current product of Navidium and measures that would be required to make it compatible with the project. Moreover, methodologies such as the use of Scrum for agile development are briefly discussed. Then, this thesis provides requirements of the project which is based on the location of components installed which is Vessel and Shore modules. The project implementation fulfilling most of the requirements is described with proper workflows to explain the process. Objectives that could not be achieved are provided with reasoning for that to occur. At the end of the implementation, the thesis summarizes the achievement of most of the requirements. At the end section of this thesis, the results are briefly discussed. The problem that arose after the deployment is mentioned along with the suspected cause of it. Finally, this thesis addresses the need for further improvement in the quality of an already quite complete project.
This thesis aimed to develop a mapping application and website which utilized the environ-mental data (temperature, humidity, light level). The Android Application was developed by Java in Android Studio and the website was written in HTML. Just the same as most of map applications, the map application shown information about the cities, roads, buildings and the current location of user using GPS data. Moreover, the application could connect with a sensor device called SensorTag by Bluetooth low energy protocol. SensorTag would collect information of surrounding temperature, humidity and light level, the application then gathered the data from the sensors and shown the environmental information together with the user’s current location on the map. Furthermore, the environmental data and the exact location of the user collected by the application were sent to a server API, which was written by JavaScript and hosted on Heroku cloud service. The web server gathered all the weather and locations data from every using application from the API and shown the information on a big map on the website. Therefore, the more users of the application, the more weather information of different locations is available on the web server. This means that a user who is using the mapping application helps users in other locations to be aware of weather condition of his/her location, and by accessing the web server, user is able to see the weather information of different places. After a considerable studying about different cloud services, Java and JavaScript programming language, the thesis has achieved the initial aim by successfully developing the Android application, the weather map website and the server API. In addition, plenty of value knowledge about cloud services was also obtained. User could observe weather condition at current location by using the Android application and received latest updates about weather information in another locations by accessing the website.
Prosumer microgrids (PMGs) are considered as active users in smart grids. These units are able to generate and sell electricity to aggregators or neighbor consumers in the prosumer market. Although the optimal scheduling and operation of PMGs have received a great deal of attention in recent studies, thechallengesofPMG’suncertaintiessuchasstochasticbehaviorofloaddataandweatherconditions(solar irradiance, ambient temperature, and wind speed) and corresponding solutions have not been thoroughly investigated.Inthispaper,anewenergymanagementsystems(EMS)basedonweatherandloadforecasting isproposedforPMG’soptimalschedulingandoperation.Developinganovelhybridmachinelearning-based methodusingadaptiveneuro-fuzzyinferencesystem(ANFIS),multilayerperceptron(MLP)artificialneural network (ANN), and radial basis function (RBF) ANN to precisely predict the load and weather data is one of the most important contributions of this article. The performance of the forecasting process is improved by using a hybrid machine learning-based forecasting method instead of conventional ones. The demand response (DR) program based on the forecasted data and considering the degradation cost of the battery storage system (BSS) are other contributions. The comparison of obtained test results with those of other existing approaches illustrates that more appropriate PMG’s operation cost is achievable by applying the proposed DR-based EMS using a new hybrid machine learning forecasting method.
Utilizing renewable energy efficiently to meet the needs of mankind's living demands has become an extremely hot topic since global warming is the most serious global environmental problem that human beings are facing today. Burning of fossil fuels, such as coal and oil directly for generating electricity leads to environment pollution and exacerbates global warming. This research is related to machine learning (ML) applications in wind power forecasting (WPF). The objective is to improve understanding of how artificial intelligence (AI) methods could potentially be used to improve the accuracy of WPF. A pilot conceptual system combining meteorological information and operations management has been formulated as a framework named Meteorological Information Service Decision Support System. This system consists of a meteorological information module, wind power prediction module and operations management module. This conceptual framework has been verified by quantitative analysis in empirical cases. This system has a potential to utilize meteorological information for decision-making based on condition-based maintenance in operations and management for the purpose of optimizing energy management. It aims to analyze and predict the variation of wind power for the next day or the following week to develop scheduling planning services for wind power enterprises (WPEs) based on predicting wind speed for every six hours, which is short-term wind speed prediction, through training, validating, and testing dataset. Accurate prediction of wind speed is crucial for weather forecasting service and WPF. This study presents a carefully designed wind speed prediction model which combines fully-connected neural network (FCNN), long short-term memory (LSTM) algorithm with eXtreme Gradient Boosting (XGBoost) technique, to predict wind speed. The performance of each model is tested by using reanalysis data from European Center for Medium-Range Weather Forecasts (ECMWF) for Meteorological observatory located in Vaasa in Finland. The results show that XGBoost algorithm has similar improved prediction performance as LSTM algorithm based on root mean square error (RMSE), mean absolute error (MAE) and coefficient of determination (R2) compared to the commonly used traditional FCNN model. On the other hand, the XGBoost algorithm has a significant advantage on training time in this case study. Additionally, this sensitivity analysis indicates great potential of the optimized deep learning (DL) method, which is a subset of ML, in improving local weather forecast on the coding platform of Python. The results indicate that, by using Meteorological Information Service Decision Support System, it is possible to support effective decision-making and create timely actions within the WPEs. The main outcome of this research can support decision optimization for an ML based decision support system. As a conclusion, the proposed system is very promising for potential applications in wind (power) energy management.