Solutions

Features and Benefits

Forecastive’s proprietary LoadVISION system offers:

• Load forecasting — Short term hourly, & sub-hourly, for electric power and energy systems

• • “Intelligent-Interactive” Module — for predicting “Abnormal Patterns”
  • advanced Neural Networks engine — for predicting “Normal Patterns”
  • optimally-averaged Zone Temperatures (historical & forecasted) from multiple weather stations
  • Unique ‘Special Days’ module further refines & improves Holiday & ‘Difficult Day’ forecasts
  • Full integration with most customer systems & fully automated to run on specified time intervals

• Electricity market price forecasting (day-ahead & same-day) for major trading hubs, as well as locational or nodal marginal pricing (LMP)
• Modeling, Evaluation, & Tune-Up options based on your custom model, or ours
• Guaranteed lower MAPE’s relative to any other source or vendor

Forecastive’s proprietary technology engine, LoadVISION, provides the most current energy demand forecasts including hourly or sub-hourly electricity load and wind every 5 minutes. It offer fully integrated weather systems, multiple automated updates throughout each day, and provides for easy integration with your other data and software applications. LoadVISION’s learning curve is easy and fast; only a few hours of training are necessary. LoadVISION allows you manually manipulate (‘tweak’) the Neural-Net generated Hourly Forecasts in real-time, to create even further forecasting scenarios and accuracies.

Price Forecasts (LMP’s, Nodals)

LoadVISION offers Electricity Price Forecasting, Day-Ahead & Same Day, for major Trading Hubs & Locational or Nodal Marginal Pricing (LMP). Also, LoadVISION works in conjunction with advanced LMP computational systems.

Output Frequency

Hourly, or sub-Hourly loads (for the next 7-10 days), each hour, down to 5 minute intervals.

Distribution Method

Cloud-based (Saas, web-based), or in-house Server.

Data Format

Forecasts are easily exported in a variety of formats, including Excel spreadsheet, and ASCII text. While the amount of data available from smart grid and other app’s may be large, LoadVISION is easily able to handle high volumes of data without disruption. As some data retrieved from smart meters man be erroneous, we recommend employing faster pre-processing techniques to filter out bad data, before integration into our forecasting system.

Licensing

LoadVISION is generally licensed by company application, not to individual users, so typically other departments within your company may use the output data.

The Science

Instead of employing a single mathematical technique in forecasting electricity demand, we use a combination of statistical and data driven techniques, to leverage the advantages of each technique. Our deeply sophisticated, hybrid model provides the most accurate demand forecasts available, over any other system. Our forecaster is modularly designed, offers a user-friendly desktop interface, with built-in support for remote data sources, an automation service and other features that allow it to be customized according to client needs.

Automated Tuning

LoadVISION synergizes continual improvement (i.e., improved accuracies) by utilizing prior forecasted results as they are auto-accommodated into subsequent time interval forecasting models, producing more accurate results. So, over time, we continue to minimize forecast errors.

Integrated Weather

LoadVISION incorporates full integration of multiple weather models, allowing users to make pertinent load forecast adjustments, in real-time. Our system also auto-recognizes and auto-updates itself, for sudden weather shifts, and does not require the customer to intervene. The need for weather forecasts (to include temperature, humidity, wind speed, cloud cover, etc) have increased in load forecasting, as generation from renewable energy sources like solar cells and wind turbines are highly dependent on weather conditions. LoadVISION also includes an out of the box plugin for the NOAA weather service.

Special Days

LoadVISION’s intelligent, interactive, real-time tool for ‘Special Days’ has received a lot of praise, as it ‘tweaks’ (improves) load forecasts for Holidays, Difficult Days, Extreme Weather events, etc, which offer difficult forecasting challenges.  In other words, we overcome inherent limitations of NN-based systems.

Sudden Weather Changes

LoadVISION auto-recognizes sudden weather shifts, and does not require the customer to intervene. This same auto-monitoring feature applies to various holidays, special events, partial holidays, etc. Our automations and “special day” capability of LoadVISION further set us ahead of our competitors.

Weather & Smart Grid

The role of weather in demand forecasting has increased in the smart grid environment. The weather impact on all customers may not be uniform across all the regions in larger utilities, spread across different regions. Forecastive addresses weather impact on the electricity demand for smart grids, as LoadVISION has already incorporated multiple weather sources into our system.

Intuitive Confidence Bands

Displays associated with forecasts allowing user identification of worst, and best case, scenarios

Confidence Intervals

Provides examine charts and graphs to judge the quality of forecasts

• ex: performance MAPE chart for a period (say, 1 week) for One-day ahead forecasts together with a Weekly chart showing performance figures for each day of the week. The user can specify the period of interest (e.g. 1 week, 1 month, 1 year, etc. in the past to judge the quality)

Modeling, Evaluation, & Tune-Up options (your model, or ours)

• Empirically analyze how your current systems, models, & methods compare to industry-leading successes, and recent new advancements
• Learn no-cost, & low-cost, ways to improve your overall STLF results
• Benchmark your accuracies vs. industry leaders
• Analyze if an ‘aggregation’ (combined weighted sum) of your individual Zones (i.e. micro-Zones) improves your overall performance

We can help your researchers and forecast analysts, to investigate the applicable boundaries of your in-house developed models and algorithms (as there is no single model, or mix, or algorithm, that is superior for all utilities or traders). Utility service areas vary in differing mixtures of industrial, commercial, and residential customers, as well as in geographic, climatic, economic, and social characteristics. We assist you in formulating the most suitable algorithm (or combinations) by testing various algorithms on your real data. Some utilities use several load forecasting methods in parallel. Without professional and objective thorough analysis, analysts cannot ideally propagate which forecasting method is best suited for a given load area. So, we can assist you to investigate the sensitivity of your load forecasting algorithms and models in relation to your customer volume, zone characteristics, energy prices, and other relevant factors.