Toshiba Develops “Electricity Markets Trading Strategy AI” for Aggregation of Renewable Energy

-Contributions to stable renewable energy supply and realization of a carbon-neutral future-

15 December, 2021
Toshiba Corporation


TOKYO─Toshiba Corporation (TOKYO: 6502) has developed “Electricity markets trading strategy AI” for renewable energy aggregation that supports strategic trading by electricity market operators. The AI bundles renewable energy sources such as solar and wind power and simultaneously makes strategic trading decisions that support profitable trading while avoiding supply imbalances(*1). By realizing stable supply of highly variable renewable energy and promoting its use as a primary power source, the AI will help to advance carbon neutrality.
Next Kraftwerke Toshiba Corporation and Toshiba Energy Systems & Solutions Corporation are participating(*2) in a demonstration experiment using the AI that began on December 1, 2021, as part of the Ministry of Economy, Trade and Industry’s Renewable Energy Aggregation Demonstration Project(*3).

Development Background

Efforts to promote a carbon-neutral economy accelerate worldwide. In Japan, the current feed-in-tariff (FIT) program , which purchases generated power is purchased at a fixed rate, will be replaced in by a feed-in-premium (FIP) program in FY2022, which add a premium (subsidy) to price of electricity sold in wholesale markets(*4). In electricity markets transactions, FIP will increase pressure on renewable energy power producers to secure profits by managing supply imbalances from fluctuations in power generation, and from price fluctuations and other market risk.
Energy aggregators bundle renewable energy sources from multiple small-scale renewable energy generators, and must execute optimal sales plans in the wholesale market to make profits. The scale of the aggregation business is growing fast, and is expected to expand from current levels that is approximately 4.4 billion yen to some 73 billion yen by 2030(*5). However, it is very sensitive market, where power output is directly affected by weather conditions, and market prices vary with the balance of power supply and demand. Aggregators must make profit by avoiding imbalances, and execute complex market transactions guided only by intuition and experience. Toshiba’s AI offers a tool for highly accurate forecasting of renewable energy output and optimal strategic trading that avoids imbalances and secures reasonable profit.

Figure 1: Overview of the renewable energy aggregation business.

Features of the Technology

To address this issue, Toshiba has developed “Electricity markets trading strategy AI” with a unique algorithm that supports strategic trading, allowing aggregators to avoid imbalances and secure revenue. It uses Toshiba’s proprietary high-precision forecasting technology to forecast renewable energy generation and market prices, optimizes strategic transactions for each aggregator, and thereby maximizes profits.
The algorithm calculates optimal ratios of bid volumes sold on the Japan Electric Power Exchange’s (JEPX)  spot market and intraday market when spot market bids are made, on the day before actual supply and demand. Aggregators can use the results to make strategic trading decisions regarding changes to levels of bid volumes proportions sold in each market.
The AI is combine two powerful capabilities: Toshiba’s highly regarded proprietary prediction technology, Grand Prix winner at the PV in Hokkaido 2019 solar power prediction technology contest(*6); and a new algorithm that uses highly accurate forecasting and historical data to automatically and then model multiple future scenarios as optimization problems subject to multiple risks.
After forecasting power generation volumes and market prices, the AI calculates risks that include imbalances and trading opportunity losses caused by combinations of fluctuations in each. Market risk is modeled as CVaR(*7), a measure used in finance, and an optimal solution is derived with imbalance risk as a constraint. Risk modeling using CVaR requires generation of many patterns that could occur in power output and prices on the day of actual supply and demand. This is done with predictive data for the day and actual historical data that is modeled as a formula that combines the data patterns to make a huge number of scenarios. The AI uses this risk modeling to determine ratios of sell bid volumes that maximize revenue in the spot and intraday markets. This successfully realizes risk-aware trading strategies that were previously difficult to achieve.

Figure 2: Overview of “Electricity markets trading strategy AI”.

Future Developments

The test demonstrations of renewable energy aggregation that Next Kraftwerke Toshiba Corporation and Toshiba Energy Systems & Solutions Corporation are participating evaluate the AI’s effectiveness for strategic trading in the spot and intraday markets, in respect of the timing of spot market bidding on the day before actual demand. Toshiba also aim to expand the AI’s functionality to trading in intraday market that is expected to be revitalized in the near future, and to strategic trading in the other markets, such as the recently established capacity and reserve market. Through its comprehensive support for renewable energy aggregators, Toshiba aims to contribute to the promotion of renewable energy as a primary power source, and to long-term stable electric power supply.

*1: Imbalances: The differences between electricity demanded and amounts supply. If renewable energy generation volumes deviate from planned values and that imbalance grows large, the quality of the electricity supply may deteriorate and blackouts may occur. Furthermore, imbalance fees will be imposed as an adjustment cost.

*2: “FY2021 Subsidy for Demonstration Project for Establishing Next-Generation Technologies Using Distributed Energy Resources such as Storage Batteries (Renewable Energy Aggregation Demonstration Project within the Renewable Energy Generation Aggregation Technology Demonstration Project).”

*3: Commencement of Renewable Energy Aggregation Demonstration Experiment (

*4: FIT was introduced in Japan in 2012, with the aim of expanding penetration and reducing costs in early stages of introducing renewable energy. FIT guaranteed electric power companies would purchase electricity generated by renewable energy producers at a set price for pre-determined period. Resulting issues included a lack of incentives to increase supply during peak demand periods, when the market price of electricity is high, because revenues were constant regardless of when the electricity was generated. As FIT has achieved the initial targets for the introduction of renewable energy, FIP that will be introduced in fiscal 2022. It will position renewable energy as a competitive power source, a significant step toward making renewable energy a primary power source. Premiums will be added during peak electricity demand periods, and income will be linked to market prices, as an incentive for power generation companies to increase electricity supplied through use of storage batteries and other means. (Source: Agency for Natural Resources and Energy, “FIT overhaul and restructuring of renewable energy policy,” issued 22 Apr 2019).

*5: Yano Research Institute Market Report, “Energy Resource Aggregation Business 2019”

*6: Grand prix winner at “PV in Hokkaido, a 2019 contest for solar power generation forecasting technologies, sponsored by an electric power company.

*7: CVaR: An abbreviation of “conditional value at risk,” also called an “expected shortfall.” A measure of risk used in the financial sector.