Tennessee Valley Authority

Text-to-speech desktop software for individual use.

These tools use a massive amount of data and provide intelligence beyond what could be seen from a single vector.

Portal users can find answers more quickly and easily with features like auto-complete search queries, natural language support, and typo handling.

KNN regression to predict hourly unitized impact of an upgraded CWHP in a residential home under historical and forecasted weather conditions.

Use NLP to search through CRs.

DR dispatch simulation to demonstrate hourly unitized impact of an EE/DR smart thermostat in a residential home under historical and forecasted weather & load conditions.

Enterprise Planning can now obtain aggregations of ES&P forecasts and ES&P can gain insight into evaluation methodologies.

This app is (1) an ETL tool for ancillary data (2) a light-weight MLOps tool for the modeling of energy efficiency and electrification measure-level impact shapes, primarily as it relates to system planning assumption...

Extreme load analysis involves estimating the likelihood of peaks in energy demand for different months of the year and hours of the day.

The requested updates will augment the GET application by adding the ability to subset and display SSL Confidence Band plots within the user interface, alongside other planned bug fixes and performance enhancements.

The Groundwater Modeling Evaluation Tool (GMET) is an R Shiny application that uses Phreeqc geochemical models and analysis to evaluate groundwater chemistry data at each of TVA's CCR sites

Use computer vision to detect rate of change of water temperature with thermal image data which is indicative of dam leakage

AI system to efficiently help find information for public inquiries related to land managment, dock permits, and annual reviews.

Use predictive analytics and confidence levels to track materials as they go through the supply chain process and risk on being delivered on time

Real-time visibility into network communications, assets, connections, and protocols.

Linear model to predict hourly unitized charging load from non-fleet EVs in a typical meteorological year (TMY).

A multi-scale graph neural network-based autoregressive model.

Phase1: it would simply include a chat window for users to ask questions of our specs.

Set of ~200 linear models for other EE & BE measures to predict hourly unitized impactunder historical and forecasted weather conditions.

Run TVA PSSE models, train a graph neural net on those results, utilize reinforcement learning on top of the GNN to optimally answer user questions such as "Where are the top 5 optimal places for a new battery?" or "w...

Monitoring of vegetation around transmission line and tower and predictive analytics on when/where to trim.

Improved efficiency and productivity, access to information and insights

Use of AWS Q will help IT be more efficient and increase productivity

Extract key business insights from customer calls, video files, clinical conversations, and more.

Increased efficiency and productivity, faster access to information

Increased efficiency and productivity, faster searching and access to information, improved compliance

Increased efficiency and productivity, faster searching and access to information

This app allows internal TVA users to generate sample bills for directly served and BCD-rate customers based on active wholesale rates, historical load, and expected shape/magnitude of new behind-the-meter solar arrays.

Increased efficiency and productivity, faster searching and access to information

Datum is a virtual knowledge employee for Teams.

This update allows TVA staff to filter, visualize, and export model results on demand, reducing reliance on static reports and improving operational efficiency.

Increased efficiency and productivity, faster searching and access to information

Faster insights, access to information, and reduction of duplicates

Using a combination of ruleset definitions specifically designed to detect malicious activity and augmented by a proprietary machine learning algorithm, this tool protects web applications regardless of location.

It serves multiple functional roles to enhance productivity, accelerate insights, and bridge knowledge gaps for users working with time series data.

Increased efficiency, productivity, and access to information