Key Point
GS Wind, a leading domestic wind power company, adopted an AWS cloud-based AI solution, achieving a demanding 8% error rate, securing additional settlement funds and dramatically improving operational efficiency.
The Client
GS Wind, a renewable energy subsidiary of GS Group, is a leading green energy company pioneering and developing the domestic wind power market.
1. A Leader in the Eco-Friendly Energy Transition
In step with the era of eco-friendly energy transition, GS Wind is leading the way in building a sustainable energy ecosystem based on its expertise and technological prowess across the entire wind power business.
2. A Complete Solution for Wind Power
GS Wind provides integrated solutions covering the entire wind power business, from development to construction and operation. For onshore wind power, we precisely analyze wind conditions in key regions across Korea to select optimal sites and develop and construct efficient wind power plants.
We maximize power generation efficiency with advanced monitoring systems and predictive maintenance technologies, and we also provide specialized maintenance services to optimize wind turbine performance and extend their lifespan. Our ongoing research and development of wind power technology is also playing a significant role in achieving domestic wind power self-sufficiency.
The Challenge
1. A Tricky Game of 8% or Less
GS Wind participates in the Korea Power Exchange's renewable energy power generation forecasting system. This system allows users to submit their forecasts for the next day's power generation the day beforehand, and if the error rate between the forecast and actual power generation is within 8%, they receive an additional payment of 3-4 won per kWh.
2. The Fickle Nature of Wind
However, wind power generation fluctuates like a roller coaster depending on weather conditions. Various meteorological factors, such as wind speed, wind direction, air pressure, and temperature, intertwine like a complex puzzle to influence wind power generation. Sudden weather changes or unexpected weather phenomena, in particular, make forecasting as difficult as a hairpin bend.
3. The Limits of Simple Forecasting
Consistently achieving an error rate of 8% or less using existing, simple weather forecast data was as difficult as passing the eye of a needle. As a result, we often missed out on additional payments, which felt like losing money we had in our hands.
4. Impact on Power System Stability
The lack of an accurate power generation forecasting system is a critical issue that goes beyond simply incurring incentive losses, impacting the stability and efficiency of power system operation.
The Solution
1. A Solid Foundation with the AWS Cloud
GS Wind solved all these problems at once with an integrated AI solution based on the AWS cloud!
2. A 24/7 Data Collection Pipeline
Using AWS Lambda, we built an automated data pipeline that collects weather forecast data, real-time weather observation data, and wind turbine operation data 24/7. Like a worker who never rests, we continuously monitor changing weather conditions!
3. A Preprocessing System that Prepares Data
We preprocess all collected data into a standardized format to create a dataset optimized for machine learning models. By comprehensively analyzing meteorological factors such as wind speed, wind direction, pressure, temperature, and humidity, as well as turbine performance data, we pinpoint key variables that impact power generation.
4. A Smart Prediction Model with Amazon SageMaker
We built an advanced machine learning environment using Amazon SageMaker. This platform periodically learns new data, continually improving the model's prediction accuracy. A sophisticated prediction model that even reflects seasonal weather patterns and turbine performance changes has been completed!
5. Completion of a Fully Automated System
The completed model uses an automated process to predict the next day's power generation and automatically submits the results to the Korea Power Exchange at the correct time. Now, a smart system requiring virtually no human intervention has been completed!
The Result
1. A Single, Efficient System
Thanks to the AWS-based integrated AI solution, we can now seamlessly manage the entire process—from weather data collection to machine learning model training, to forecast generation and submission—within a single, integrated system. With previously scattered systems integrated into a single, robust platform, operational efficiency has skyrocketed!
2. People Can Focus on More Important Tasks
The automated system freed our staff from repetitive tasks, allowing them to focus more on improving and optimizing the performance of our machine learning forecasting models. It's as if we've been freed from the shackles of routine work, freeing them to focus on more creative and specialized tasks!
3. Goal Achieved! Stable Error Rate Within 8%
The most important achievement is that we've consistently achieved an error rate of less than 8%! This has allowed us to consistently secure an additional 3-4 won per kWh in settlement fees, significantly improving power generation profitability.
4. Contributing to the stability of the national power grid
By accurately predicting power generation, it contributes to improving the stability of power grid operation and plays a crucial role in the expansion of renewable energy in Korea. This has created a wonderful result, where individual success benefits society as a whole!
Key Point
GS Wind, a leading domestic wind power company, adopted an AWS cloud-based AI solution, achieving a demanding 8% error rate, securing additional settlement funds and dramatically improving operational efficiency.
The Client
GS Wind, a renewable energy subsidiary of GS Group, is a leading green energy company pioneering and developing the domestic wind power market.
1. A Leader in the Eco-Friendly Energy Transition
In step with the era of eco-friendly energy transition, GS Wind is leading the way in building a sustainable energy ecosystem based on its expertise and technological prowess across the entire wind power business.
2. A Complete Solution for Wind Power
GS Wind provides integrated solutions covering the entire wind power business, from development to construction and operation. For onshore wind power, we precisely analyze wind conditions in key regions across Korea to select optimal sites and develop and construct efficient wind power plants.
We maximize power generation efficiency with advanced monitoring systems and predictive maintenance technologies, and we also provide specialized maintenance services to optimize wind turbine performance and extend their lifespan. Our ongoing research and development of wind power technology is also playing a significant role in achieving domestic wind power self-sufficiency.
The Challenge
1. A Tricky Game of 8% or Less
GS Wind participates in the Korea Power Exchange's renewable energy power generation forecasting system. This system allows users to submit their forecasts for the next day's power generation the day beforehand, and if the error rate between the forecast and actual power generation is within 8%, they receive an additional payment of 3-4 won per kWh.
2. The Fickle Nature of Wind
However, wind power generation fluctuates like a roller coaster depending on weather conditions. Various meteorological factors, such as wind speed, wind direction, air pressure, and temperature, intertwine like a complex puzzle to influence wind power generation. Sudden weather changes or unexpected weather phenomena, in particular, make forecasting as difficult as a hairpin bend.
3. The Limits of Simple Forecasting
Consistently achieving an error rate of 8% or less using existing, simple weather forecast data was as difficult as passing the eye of a needle. As a result, we often missed out on additional payments, which felt like losing money we had in our hands.
4. Impact on Power System Stability
The lack of an accurate power generation forecasting system is a critical issue that goes beyond simply incurring incentive losses, impacting the stability and efficiency of power system operation.
The Solution
1. A Solid Foundation with the AWS Cloud
GS Wind solved all these problems at once with an integrated AI solution based on the AWS cloud!
2. A 24/7 Data Collection Pipeline
Using AWS Lambda, we built an automated data pipeline that collects weather forecast data, real-time weather observation data, and wind turbine operation data 24/7. Like a worker who never rests, we continuously monitor changing weather conditions!
3. A Preprocessing System that Prepares Data
We preprocess all collected data into a standardized format to create a dataset optimized for machine learning models. By comprehensively analyzing meteorological factors such as wind speed, wind direction, pressure, temperature, and humidity, as well as turbine performance data, we pinpoint key variables that impact power generation.
4. A Smart Prediction Model with Amazon SageMaker
We built an advanced machine learning environment using Amazon SageMaker. This platform periodically learns new data, continually improving the model's prediction accuracy. A sophisticated prediction model that even reflects seasonal weather patterns and turbine performance changes has been completed!
5. Completion of a Fully Automated System
The completed model uses an automated process to predict the next day's power generation and automatically submits the results to the Korea Power Exchange at the correct time. Now, a smart system requiring virtually no human intervention has been completed!
The Result
1. A Single, Efficient System
Thanks to the AWS-based integrated AI solution, we can now seamlessly manage the entire process—from weather data collection to machine learning model training, to forecast generation and submission—within a single, integrated system. With previously scattered systems integrated into a single, robust platform, operational efficiency has skyrocketed!
2. People Can Focus on More Important Tasks
The automated system freed our staff from repetitive tasks, allowing them to focus more on improving and optimizing the performance of our machine learning forecasting models. It's as if we've been freed from the shackles of routine work, freeing them to focus on more creative and specialized tasks!
3. Goal Achieved! Stable Error Rate Within 8%
The most important achievement is that we've consistently achieved an error rate of less than 8%! This has allowed us to consistently secure an additional 3-4 won per kWh in settlement fees, significantly improving power generation profitability.
4. Contributing to the stability of the national power grid
By accurately predicting power generation, it contributes to improving the stability of power grid operation and plays a crucial role in the expansion of renewable energy in Korea. This has created a wonderful result, where individual success benefits society as a whole!