Loading ...
QR AI Forecaster offers a wide range of ready-for-use, advanced AI forecasting solutions that are tailored for intraday, 7 days ahead, and long term load forecasting requirements of utilities, munis, electric co-ops, retail marketers & generators
Challenges
Clients’ profile. Load serving entities such as utilities, electric co-ops and munis, retail marketeers, and generators operating in real-time intraday and day-ahead (5-60 mins resolution) markets.
Target.
1. Accurate and flexible load forecasts, aggregating or drilling load by any desired criteria, i.e., geographic regions, cities, distribution nodes, or by consumer type (residential, commercial, industrial).
2. Net load forecast, i.e., demand net of behind the meter solar generation, needed to optimize the integration of microgrid fleets with utilities’ macrogrid.
3. These AI demand forecast models can be short-term intraday, up to 7 days ahead, and medium to long term. In the latter case the model should include what-if-scenarios on exogenous factors, e.g., a shift in temperature up or down by 10 degrees, or some global economic or localized indicators such as consumer load growth by 9%, etc.
Implement real-time 24/7 automated, stand-alone and accurate load forecast solutions, as a web cloud solution that can be easily integrated with in-house trading systems, ISOs, weather data services, and internal data sources, e.g., meters, SCADA, etc.
Why QR Load Forecaster Solution
We offer a range of ready-for-use advanced Deep Learning AI and Machine Learning models tailored for short term intraday and (7) day-ahead, and longer term load forecasts.
Long term forecasts include what-if-scenarios on exogenous factors, e.g., a shift in temperature up or down by 10 degrees, some global economic indicators, or localized indicators such as consumer load growth by 9%, etc.
Forecast resolution can be 5-60 minutes, depending on your requirements. Forecasts computation frequency is configurable, e.g., 5-10 minutes, 24/7 for real-time intraday forecasts, and 9-12 am for day-ahead forecast.
Four types of load forecast solutions are offered:
1. Utility Load Forecast is performed either on individual load nodes, or aggregated per business rules such as geographic location, region, city and client type. For greater accuracy, this type of load forecast is adjusted in real-time to match SCADA levels when available, and weather indicators as predictors.
2. Utility Net Load Forecast is a forecast of demand net of behind the meter solar generation, scattered across fleets of individual PV stations. You can provide aggregate, or separate load and PV generation data and the system stores and aggregates them in a modern data hub. We or you can provide multiple weather and irradiance data to add as models’ predictors.
3. Consumer Load Forecast is based on granular or monthly consumption meter data. Such load forecasts can be performed on individual consumer meters, or aggregated per business rules such as geographic location, region, city and client type (residential, commercial, industrial). Weather indicators are used as predictors.
4. ISO and RTO Load Zone, Transmission Authority, Trading Hub Forecast. These load forecasts use as predictors weather and auxiliary data provided by the ISO, e.g., hour-ahead or day-ahead projections.
How it works
Whether we install QR Load Forecaster on your cloud, or you use it as a service, the road map is the same. This solution is fully automated and fetches your load, meter, SCADA, or ISO load data; as well as weather data. It performs load forecasting and publishes the results via API or manual file download. You can always start with a POC Trial.
- Demos and Q&A.
- SOW: Identification of number of load data, and type of forecasts of interests for the POC Trial, and later for full service.
- Execution of POC Trial Agreement.
- Account set-up, access control management.
- Connectivity to client’s sample load data and data fetching for POC.
- Load forecast model set-up and testing.
- Walkthrough and demo for client’s users to use the forecast web-portal for POC Trial.
- Client can access the demo portal 24/7 to see and download the selected price forecasts.
- Support and ongoing Q&A during Trial.
- Client has the option to execute the Subscription Agreement and start using the solution commercially.
- Mapping out the full rollout plan for client’s load forecast project.
Explore a full range of price and load Forecast as a Service solutions
AI Load & Demand Forecast Case Studies
QR Load Forecaster Benefits & Features
We offer flexible delivery options:
a) AI Forecast Service where our expert team and platform do everything and you receive, via API and electronic means, intraday and 5-DA load forecasts, every 5-10 minutes, 24/7.
b) Software as a Service (SaaS) where we implement QR AI Forecaster on a private cloud of your choosing, or on-site, and you control the data and the AI models.
You pay one single annual subscription fee comprising license, maintenance and upgrade releases.
Our expert (MSc and PhD) data science team configures, fine-tunes and deploys your AI forecast models in record time and helps you maintain and calibrate them in time.
QR AI Forecaster is a no-coding AI platform that allows users to drag, drop and combine various AI and Machine Learning methods in a user-friendly dashboard to create their own custom forecast models. The AI platform offers:
a) Deep Learning (Long Short Term Memory Networks (LSTM), 1D/2D Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), and Gated Recurrent Unit (GRU);
b) Machine Learning models, e.g., XGBoost, ADABoost, Random Forest and Catboost;
c) Single Models: MLP (neural networks), decision tree, SVM and KNN.
To ensure the highest degree of accuracy and stability, AI model optimization is automated with specialized built-in tool-boxes that considerably lessen the need and workload of data scientists and analysts.
To perform load forecasting, QR AI Forecaster allows you to aggregate or disaggregate load data on the fly, by any criteria across your operations. E.g., go from individual meters, to cities, regions or by consumer types.
Each forecast can have upper and lower uncertainty bands computed via quantiles or standard deviation.
To reduce project risk you can start with a Proof of Concept (POC) or trial period. The fee is scaled to the implementation efforts needed by the POC. Contact us for more details.
To increase load forecast accuracy QR AI Forecaster allows users to control a host of key parameters: date range for model training and forecast horizons, calendars to replace intra-week holidays’ data with nearest Sunday’s data, model splitting (to create multiple models across off/peak periods and week/week end days, model grouping (create multiple regression/forecast models, and have a classifier merge them into the best forecast, internal and external predictors, e.g.,SCADA, weather indicators, demand and supply side data, generation by fuel type including renewables, outages, ISO published hour and day ahead forecasts, data quantization, scaling, gap filling methods, outlier data remediation.
QR Data Manager is seamlessly integrated to provide end-to-end data management (fetching, validation, insertion) for world ISOs & RTOs, multiple weather data services, meter and SCADA data. The methods supported are web-scraping, web services, API, FTP, or manual upload for csv, Excel, XML and Json file formats.
Our Clients Say
Lets Talk About Your Needs
Our Contact Details:
