Real Time Forecasting Solution
EFS enable utilities to operate more efficiently and effectively at all levels.
This Energy Forecasting Solution allows utilities to own their forecasting process from beginning to end, and use data-driven analytical insights for prudent financial decisions.
Our Energy Forecasting Solution produces repeatable, scalable, traceable and defensible energy forecasts. Utilities can respond quickly and confidently to forecasts and can improve forecasting performance for planning horizons that range from very short-term to very long-term. Improved forecasting generates tangible value in areas such as planning, operations, energy trading, contract power purchases and demand-side management. This solution helps utilities of all sizes operate more efficiently by supporting the entire forecasting process, from data management and model development to forecasting and business reporting. The solution is designed for forecasters, analysts and planners at utility companies. In the highly competitive and regulated utility business, there’s a clear link between the company’s bottom line and forecast accuracy and reliability. But most legacy load forecasting solutions were not designed to handle the variability, complexity and volumes of data emerging in the utility landscape today. These solutions also lack the automation needed to permit appropriate model selection for each type of forecast required. Because legacy forecasting software tends to provide forecasts of questionable accuracy, many organizations are unwilling to rely on statistically generated forecasts. Yet they know that forecasting errors result in suboptimal decisions. This Energy Forecasting Solution allows utilities to own their forecasting process from beginning to end, and use data-driven analytical insights for prudent financial decisions. From advanced data management through best in-class analytics, forecasters can now handle the complex questions that arise from utilities’ dynamic business models. Using this solution will ensure that forecasts are accurate which in turn will allow for optimized gas distribution strategies and improved quality of information provided to all stakeholders. Amongst others, this solution brings all the following state of the art capabilities built-in: 1. Machine Learning 2. Advanced Analytics 3. Artificial Intelligence 4. Time Series Analysis In addition, the solution brings all required tools to deliver the following applications: 1. Fraud Detection 2. Predictive Maintenance 3. Data Quality Processes 4. Advanced Reporting KEY FEATURES Energy forecasting • Multiple regression models form the foundation for load forecasting, where additional variables and combinations are tested sequentially for model improvement; models are tested at each iteration to prevent over-fitting. • Second-stage models are developed using UCM, ARIMAX, exponential smoothing and neural nets. • Users can choose the level of automation for the forecasting process. (Default is fully automated, hands-off forecast generation.) • Users can set automatic selection criteria in the model. • Custom models can be built within the software, with any structure or frequency. • Existing externally developed models can be imported to use in the software, automating parameter estimation. • Users can choose to re-estimate any model at any time, as often as needed. • Model results and statistics are available at each step of the process. • Outlier files are automatically generated and can be reviewed and visualized with a GUI. User-driven hierarchical forecasting • Includes a graphical user interface for energy forecasting – from model development through testing and publication. • Lets you organize projects by time horizon, execution date or member companies. • Employs a common hour-by-hour forecasting methodology across all time horizons. • Uses the same “diagnose” process to look at errors and outliers, regardless of time horizon. • Common search, drill-down and filter options across all time horizons. • User can select top-down or bottom-up reconciliation. Data management • An easy-to-use, point-and-click GUI has an intuitive set of configurable windows for managing data integration development processes. Drag-and-drop functionality eliminates programming. • Dedicated GUI for profiling data quality, to identify and repair source system issues while retaining the business rules for use in other ETL processes. • Support for unstructured and semi structured data to parse and process files. • Access to static and streaming data for sending and receiving via web services. • Metadata is captured and documented throughout transformations and data integration processes and is available for immediate reuse. • Capability to determine the path, processes and transformations taken to produce the resulting information. • Data lineage (reverse impact analysis) is available – critical for both validating processes and building user confidence in data. What-if analysis and scenario planning • What-if planning capabilities using economic and weather scenarios provided by SAS. • Ability to change model parameter estimates to determine effects on forecast scenarios. SAS Visual Analytics Administration and Reporting • Design and distribute BI reports and dashboards through an interactive user interface, no programming required. • Present data and results in the most insightful way through a broad array of visualization techniques. • Enable viewing, interaction and collaboration via dynamic reports and dashboards through native mobile BI apps for Apple® iOS and Android devices. • Native integration with ESRI mapping technology. • Recipients can add consolidated comments to a report. Mobile users, Microsoft Office users and web users can all share and view comments in a central location. Comments stay with the reports so everyone can see them. • Data can be interactively prepared for analysis – this includes including joining tables, defining custom calculated columns and creating custom expressions. Flexibility and scalability beyond the GUI • Includes data mining, linear and logistic regression, decision trees and neural networks. • Lets you re-diagnose and identify candidate models, re-estimate existing model parameters, or generate forecasts using existing models and parameters. • Provides more effective, statistics-based forecasting methods to use throughout the organization. • Offers an extensible model repository, including intermittent demand models, unobserved components models, ARIMAX models, dynamic regression, exponential smoothing models with optimized parameters, and user-defined models – enabling you to create more appropriate forecasting models for a wide range of behaviours.
Our Energy Forecasting solution will provide many benefits to Wales and West Utilities, meeting the key requirements from the details provided to us. We will work with Wales and West to ensure all benefits are maximized. Below details the benefits of our solution. * Get hourly forecasting for all time horizons in a single, comprehensive solution that lets you use existing planning resources and helps everyone work more effectively. * Reduce total cost of ownership with self-service BI. * Make better predictions about energy demand by building accurate predictive models based on more data from more sources, including smart meters. Automatically track model accuracy and easily update the model to reflect changes. * Forecast big – big data, that is. With native access to relevant big data platforms, the size, location and format of your data is not a constraint when it comes to improving forecast performance. * Quickly evaluate more scenarios. Combined economic and weather range scenarios allow our solution to let you evaluate more scenarios faster and with fewer resources. With various scenarios in hand, forecasters can create multiple mid-term and long-term models based on anticipated outcomes, and adjust the models with ease. * Get time-sensitive insights. With automatic re-forecasting based on data updates, your teams can make decisions that mitigate risks, surface new business opportunities and ultimately create a competitive advantage. * Push the limits with flexible, expandable functionality. Get creative as you explore new forecasting challenges, data types and hierarchies, all in one solution. Our Energy Forecasting solution improves results by providing trustworthy, repeatable and defensible energy forecasts for short-term (an hour ahead) to long-term (50 years ahead) planning horizons. It helps utilities operate more efficiently and effectively at all levels of decision making, while providing a broad degree of automation, scalability, statistical sophistication and transparency. Forecasting for all time horizons Without requiring multiple tools, our Energy Forecasting Solution gives you an approach to each forecasting situation that arises. It encompasses hourly forecasting for all time horizons, based on trusted data and advanced forecasting algorithms. Our solution also provides hierarchical forecasting for big data, including smart meter data. If your organization has relied on third parties or untrainable models, our solution allows for more sophisticated model creation – so models are selected on their likelihood to produce good forecasts of the future, not just on how well they fit the past. Integrated data management Data used for forecasting often has erroneous or missing values and does not reside in one location. Our solution helps forecasters manage data integration projects on an enterprise scale in a timely, cost-effective manner while meeting the high data expectations of information consumers. Data governance and data quality capabilities ensure that the data is trustworthy and that its lineage is transparent and auditable. Forecasters can reduce the manual efforts of data transformations, focusing instead on high-value analytical activities. High-performance load forecasting Our Energy Forecasting solution enables utilities to maximize value from existing planning resources and improve forecast performance by operating with greater efficiency. Because the solution is built on a big data analytics platform, forecasters are not constrained by the size of the data or forecasting objectives. As a result, forecasters and planners can continue to play a role in business decisions at all levels of the organization, using new insights to contribute to a return on smart grid investments. A common visual interface for analysts and executives With our Energy Forecasting Solution, business managers have a visual interface for viewing forecast results – powered by, yet separate from, the forecast workbench. Forecasters can launch the solution, then business users can explore the output. Components of included with our Energy Forecasting solution provide drag-and-drop and auto-charting capabilities, so no coding is required. Plus, you can share your reporting results via the web and mobile devices. Combined economic and weather scenarios Our Energy Forecasting solution provides automatic, combined economic and weather scenarios for better long-term energy forecasting. Scenario analysis lets forecasters evaluate exceptions to prior load history and plan for future vents. Forecasters can investigate alternate scenarios using pre-packaged economic forecasts to run what-if analyses. And they can simulate and test forecast rigor as a way to select long-term models that meet criteria for planning and operations. Capabilities beyond the traditional forecast Based on years of forecasting software development, we know that today’s utility analyst is under more pressure than ever. That’s why our Energy Forecasting solution lets analysts explore complex forecasting issues by including additional solutions specific to the task. This allows data scientists to create specific forecasting models for a wider range of behaviours by using an extensible model repository, including decision trees and neural networks. Forecasters can also interact with additional data sets to answer questions from the meter to the system level, even understanding maximum load under certain conditions. Surface key insights and share across the organization Our solution helps utility load forecasters collect, explore, analyse and interpret all types of data, without limitations. With this software’s unique depth and breadth of analytics, you can test alternatives, build models and measure results, for better efficiency and accuracy. Then you can share results across all levels of the business. One-stop shop for forecast process efficiency Our Energy Forecasting solution covers the entire forecasting process, from data integration and model development to forecasting and reporting. It prevents headaches that result from trying to patch together mismatched components – and provides you with a complete forecasting solution that encompasses data management, analysis, visualization and reporting. Amongst others, this solution brings all the following state of the art capabilities built-in: 1. Machine Learning 2. Advanced Analytics 3. Artificial Intelligence 4. Time Series Analysis In addition, the solution brings all required tools to deliver the following applications: 1. Fraud Detection 2. Predictive Maintenance 3. Data Quality Processes 4. Advanced Reporting
The presented solution is extremely flexible and adaptable. It allows for functionality extension as users can elaborate new applications based on the platform. Amongst the many, we can highlight the following: Gas Demand Forecasting Power Demand Forecasting Fraud Detection Predictive Maintenance Data Quality Processes Advanced Reporting A wide range of Analytics techniques and tools are covered, as listed below: Machine Learning Advanced Analytics Artificial Intelligence Time Series Analysis Predictive Modelling Deep diving into more detail, the following applications can be found and successfully tackled by the proposed solution: ETL Data Management Data Quality Modelling Model Governance Output integration Reporting Business Discovery