EV-Insights: open source framework for electric vehicle charging data processing, analysis, and forecasting

The worldwide usage of Internet has been generating data exponentially. Internet has re-evolved business operations and its number of consumers. The data generation begins with the fact that there is vast information to capture and store. The rate of mounting of data on the Internet was one of the important factors in giving rise to the concept of big data. However, it is related to Internet but its existence is due to growing unstructured data which requires management. Organization stores this data in warehouses for future analysis. Besides storage, the organization also needs to clean, re-format and then use some data processing frameworks for data analysis and visualization. Hadoop MapReduce and Apache Spark are among various data processing and analysis frameworks. In this chapter, data processing frameworks Hadoop MapReduce and Apache Spark are used and the comparison between them is shown in terms of data processing parameters as memory, CPU, latency, and query performance.

In the era of big data, various forms of massive data are generated, and the way of dealing with them is also diversified. Common data processing methods are batch processing, stream processing and hybrid processing. In the face of multitudinous data processing framework, it is of practical significance to summarize the types and characteristics of data processing framework and study how to choose the appropriate framework. This paper first introduces several mainstream data processing frameworks (Batch Processing Framework, Stream Processing Framework and Multi Framework Hybrid Processing System), and it briefly describes their related concepts and system architecture. Then it compares and selects typical stream processing framework and multi framework hybrid processing system classification, and gives suggestions on how to choose the appropriate framework. Finally it summarizes and looks forward to the development prospects of big data processing technology.

Impact of COVID-19 on private driving behavior: Evidence from electric vehicle charging data

In this paper, five imputation methods namely Constant (zero), Mean, Median, Maximum Likelihood, and Multiple Imputation methods have been applied to compensate for missing values in Electric Vehicle (EV) charging data. The outcome of each of these methods have been used as the input to a prediction algorithm to forecast the EV load in the next 24 hours at each individual outlet. The data is real world data at the outlet level from the UCLA campus parking lots. Given the sparsity of the data, both Median and Constant (=zero) imputations improved the prediction results. Since in most missing value cases in our database, all values of that instance are missing, the multivariate imputation methods did not improve the results significantly compared to univariate approaches.

In recent years, integration of sustainable energy sources integration into power grids has significantly increased data influx, presenting opportunities and challenges for power system management. The intermittent nature of photovoltaic power output, coupled with stochastic charging patterns and high demands of electric vehicles, places considerable strain on system resources. Consequently, short-term forecasting of photovoltaic power output and electric vehicle charging load becomes crucial to ensuring stability and enhancing unit commitment and economic dispatch. The trends of energy transition accumulate vast data through sensors, wireless transmission, network communication, and cloud computing technologies. This paper addresses these challenges through a comprehensive framework focused on big data analytics, employing Apache Spark that is developed. Datasets from Yulara solar park and Palo Alto's electric vehicle charging data have been utilized for this research. The paper focuses on two primary aspects: short-term forecasting of photovoltaic power generation and the exploration of electric vehicle user clustering addressed using artificial intelligence. Leveraging the supervised regression and unsupervised clustering algorithms available within the PySpark library enables the execution of data visualization, analysis, and trend identification methodologies for both photovoltaic power and electric vehicle charging behaviors. The proposed analysis offers significant insights into the resilience and effectiveness of these algorithms, so enabling informed decision-making in the area of power system management.

A short-term load forecasting method considering meteorological factors and electric vehicles is essential to the successful operation of the power system. This paper proposes a unique short-term load forecasting method based on neural network. First, through the analysis of typical daily load data, it is demonstrated that the short-term load data changes with the daily, weekly, weather type and the charging of electric vehicles. Then, the load forecasting model based on the neural network is set up with historical data, meteorological data and electric vehicle charging data as input. Finally, the prediction model is simulated to improve the accuracy of load forecasting.

With the emergence of electric vehicle as an emerging industry, electric vehicle charging stations, electric vehicle supporting charging technologies and electric vehicle surrounding industries have also seen great development and changes. However, there are still problems such as difficulties in finding piles for electric vehicle users, low utilization rate of charging piles, and difficulty in making profits for operating enterprises. How can an electric car battery technology, the reasonable solution to charging pile enough information circulation and mutual trust between operators and electric car owners security issues, on this basis, considering a reasonable interest with the user of electric vehicle charging pile match, improve the utilization rate of resources, guarantee the user experience of charging at the same time, this is the need to study and solve the problem. Firstly, this topic proposes to establish an electric vehicle charging chain by using block chain technology, so as to reasonably solve the problem of insufficient circulation of charging pile information and mutual trust and safety between operators and electric vehicle owners. Secondly, in order to power grid security, user experience and the interests of the operators make a comprehensive consideration, assignment model is proposed based on distributed ordering of electric vehicle charging model will charge directly link between cars and charging pile, put forward considering ordering charging scheme of interests, and by solving the model to get the most reasonable ordering charging scheme. Thirdly, in order to better allocate the resources in the station, the waiting time of electric vehicle charging is reduced, and the orderly charging model is optimized. This topic proposes the power distribution strategy within the station, through a simple algorithm, to achieve the priority of using high power charging in the charging station with low power, and fully balance the load resources of each station. Finally, in order to verify the effectiveness of the model, the simulation platform is built and verified. Collected by charging and the electric car GPS data, set up the simulation object model by using ArcGIS software, and set up complete simulation platform of distributed electric vehicle charging ordering model for simulation, the simulation results show that the proposed ordering of electric vehicle charging model based on block chain technology can improve the utilization rate of resources, charging at the same time guarantee the user experience.

Identification of potential locations of electric vehicle supply equipment

This paper examines AI concerning cloud computing to establish how it can be used to build a framework for intelligent data processing on a large scale in distributed systems. As the size and density of data continue to extend across the globe, industries require novel and efficient ways of processing the data intensively and in a real fashion. Cloud computing provides more open architecture and scalability, and AI makes it possible for organizations to process big data, make reasonable conclusions, and take the proper actions faster. Instead, the problem is practical planning to integrate these technologies into a seamless, highly performing, affordable, and scalable solution. The proposed framework features a three-layered architecture: a data layer for data delivery, a Processing Layer for computing, and an Outcome Layer for application execution and insight rendering to the user. This architecture is designed to utilize various cloud services, including Amazon Web Services (AWS), Microsoft Azure, or Google Cloud, to allocate resources dynamically and efficiently manage workloads. The research establishes that the framework enhances data processing in terms of throughput and decreased latency, resource utilization, and reduced operational expenses. The results show that not only does the application of AI improve scalability alongside cloud computing, but it also assists businesses in making better decisions using data. This thesis brings a practical solution in cloud computing and artificial intelligence to narrow the present-day data processing problem in a distributed environment.

Data processing is the basis and kernel for implementing terminal applications in Internet of Things(IoTs) services. As a large amount of sensing devices need to be connected to the IoTs service, it will bring an enormous number of heterogeneous data, and the data is difficult to be directly used by terminal applications in upper layer. So we need a data processing framework to achieve data integration and processing in IoTs services. In this paper, we introduce a data processing framework, LambDP, which includes a protocol stack system, a publish/subscribe system and a data processing platform. Protocol stack system offers dynamical protocol adaptation. Publish/subscribe center is a message queue, which can provide transfer and distribution for data. Data processing platform is based on lambda theory to implement parallel computing streaming mode and batch mode of data. In addition, we use the workflow pattern to encapsulate the business logic in LambDP into components, which can provide data processing services to upper layers applications. As for the performance of the data processing framework based on Lambda, we will deploy the data processing platform to give the verification.

A machine learning framework for low-field NMR data processing

From the perspective of medical big data utilization, analyze the current situation of China's medical informatization development, build a medical big data processing and service framework for digital and intelligent transformation, and provide new ideas and methods for medical informatization and intelligent medical development. Based on information life cycle management theory, combined with a large number of literature research, a medical big data processing and service framework is constructed. The framework discusses the functions of the original data module, data collection module and integrated system module, and further expounds the content and application of digital intelligent medical services.

A GTFS data acquisition and processing framework and its application to train delay prediction

This paper puts forward the support technology of fast charging supply and demand matching in charging stations and analyzes the common large-capacity electrochemical energy storage technical parameters in charging stations. For the safety of electric vehicle charging, the thermal reaction and thermal runaway processes of power batteries are introduced. Design the electric vehicle charging state monitoring and safety warning methods, and select the multi-timescale ARIMA algorithm to build the electric vehicle charging safety warning model. The sliding window method is used to process the residual mean and residual standard deviation of electric vehicle charging data to improve prediction data and decrease the chance of misjudging pre- and alarms. Combined with the evaluation standard of the safety early warning model, set reasonable pre- and alarm thresholds using the residual analysis method. The safety warning model designed in this paper is verified by different charging fault warnings. Different charging fault warning examples show that the ARIMA-based charging safety early warning model proposed in this paper can be good for the charging facility’s output voltage, output current, and charging module temperature faults for early warning to ensure that the warning is carried out before the alarm of the actual fault information, to protect the charging safety of electric vehicles.

Aims and Background: A re-evaluation of distribution transformer ratings is necessary to ensure efficient and reliable operation due to the significant impact of the growing number of electric vehicles (EVs) on load dynamics. The objective of this study is to optimize the rating of the distribution transformers to accommodate the year-round demand for electric vehicle charging while minimizing expenses and no-load losses. The performance and lifespan of transformers under dynamic hourly loads are evaluated by analysing real-world EV charging data, load profiles, and transformer parameters. Using state-of-the-art simulation, real-world data analysis, and optimization algorithms, this study optimizes the transformer distribution ratings under the integration of EV charging and renewable energy. Objectives and Methodology: Dynamic hourly load changes can be captured by analysing realworld electric vehicle charging data in conjunction with seasonal load profiles. In order to maximise transformer lifetime and minimise losses, convex optimisation is subjected to Karush-Kuhn- Tucker (KKT) conditions. Transformer performance is also assessed by using energy storage devices and solar energy simulations. To determine the impact of various electric vehicle charging scenarios on thermal stress and transformer ageing, sensitivity analyses are performed. Results and Discussion: Transformers with a 10% higher rating can manage maximum electric vehicle charging loads with a 15% slower loss of life acceleration, according to our models. Transformer performance is further optimised with the integration of solar energy and energy storage systems, which improves load control and reduces operational costs by up to 20%. Conclusion: Using a convex optimisation framework and the Karush-Kuhn-Tucker (KKT) criteria, the study achieves a 95% accuracy rate in predicting hourly load fluctuations.