Machine learning-based technology to forecast solar power generation
POWER & RENEWABLE ENERGY

Machine learning-based technology to forecast solar power generation

Two new machine learning-based models for forecasting the power generated by solar projects have been developed by researchers led by Dr Kalop of Urban Environmental Engineering and Professor Heo Jong-wan of Incheon National University in South Korea.

The advanced models incorporating artificial intelligence, dubbed the adaptive neuro-fuzzy inference system (ANFIS), efficiently forecast the power generated by photovoltaic systems up to a full day ahead of time.

Renewable and Sustainable Energy Reviews published the research paper.

Integrating solar photovoltaic (PV) power into existing power grids is a difficult task because PV systems' power output is heavily influenced by environmental factors. According to the researchers, an accurate forecast of solar PV power generation is required for efficient power integration into existing power grids.

The researchers combined two models with adaptive and time-varying acceleration coefficients: ANFIS and particle swarm optimization (PSO).

According to the researchers, the two models are described as hybrid algorithms because they combine a novel hybrid approach of adaptive swarm intelligence techniques and ANFIS in forecasting solar PV project power generation over time horizons ranging from 0 to 24 hours.

The models were designed and evaluated using climatic variables and historical PV power data from a 960 kW grid-connected PV system in south Italy. To assess the accuracy of the proposed models and the impact of variables on PV power values, several statistical analyses were conducted. At time horizons of 12 hours and 24 hours, the proposed ANFIS-APSO achieved the most accurate PV power forecast with R2 = 0.657 and 0.835, RMSE = 0.081 kW and 0.088 kW, and MAE = 0.079 kW and 0.077 kW , respectively.

The newly constructed ANFIS-APSO outperformed the standard ANFIS-PSO model, as well as other hybrid models, according to the findings. According to the findings, the model could be a promising new tool for engineers to use in forecasting the power generation of solar projects over short and long time horizons.

Renewable energy project developers face a difficult task in accurately forecasting power generation. To integrate renewable generation into the grid efficiently, utilities demand accurate forecasting and scheduling. Renewable energy forecasting technologies are in high demand from utilities as well as renewable energy generators.

Image Source

Two new machine learning-based models for forecasting the power generated by solar projects have been developed by researchers led by Dr Kalop of Urban Environmental Engineering and Professor Heo Jong-wan of Incheon National University in South Korea. The advanced models incorporating artificial intelligence, dubbed the adaptive neuro-fuzzy inference system (ANFIS), efficiently forecast the power generated by photovoltaic systems up to a full day ahead of time. Renewable and Sustainable Energy Reviews published the research paper. Integrating solar photovoltaic (PV) power into existing power grids is a difficult task because PV systems' power output is heavily influenced by environmental factors. According to the researchers, an accurate forecast of solar PV power generation is required for efficient power integration into existing power grids. The researchers combined two models with adaptive and time-varying acceleration coefficients: ANFIS and particle swarm optimization (PSO). According to the researchers, the two models are described as hybrid algorithms because they combine a novel hybrid approach of adaptive swarm intelligence techniques and ANFIS in forecasting solar PV project power generation over time horizons ranging from 0 to 24 hours. The models were designed and evaluated using climatic variables and historical PV power data from a 960 kW grid-connected PV system in south Italy. To assess the accuracy of the proposed models and the impact of variables on PV power values, several statistical analyses were conducted. At time horizons of 12 hours and 24 hours, the proposed ANFIS-APSO achieved the most accurate PV power forecast with R2 = 0.657 and 0.835, RMSE = 0.081 kW and 0.088 kW, and MAE = 0.079 kW and 0.077 kW , respectively. The newly constructed ANFIS-APSO outperformed the standard ANFIS-PSO model, as well as other hybrid models, according to the findings. According to the findings, the model could be a promising new tool for engineers to use in forecasting the power generation of solar projects over short and long time horizons. Renewable energy project developers face a difficult task in accurately forecasting power generation. To integrate renewable generation into the grid efficiently, utilities demand accurate forecasting and scheduling. Renewable energy forecasting technologies are in high demand from utilities as well as renewable energy generators. Image Source

Next Story
Infrastructure Urban

DCPC Prepares for Special Campaign 5.0 with Focus on E-Waste

The Department of Chemicals and Petrochemicals (DCPC), Ministry of Chemicals and Fertilisers, is gearing up for Special Campaign 5.0, to be held from 2nd to 31st October 2025. The initiative will focus on e-waste disposal as per MoEFCC’s E-Waste Management Rules 2022, space optimisation, and enhancing workplace efficiency across field offices.Special Campaign 4.0, conducted between October 2023 and October 2024, delivered notable results in record management, grievance redressal, scrap disposal, and cleanliness drives.Key outcomes of Special Campaign 4.0Records management: 2,443 physical fil..

Next Story
Real Estate

BlackRock India Leases 1.4 Lakh Sq Ft in Bengaluru

BlackRock Services India, the domestic arm of global asset manager BlackRock, has leased 1.4 lakh sq ft of office space at IndiQube Symphony in Bengaluru, according to Propstack data. The 10-year deal is valued at around Rs 4.10 billion.The lease, among the largest transactions in India’s co-working sector, highlights the growing preference of global institutions for flexible office providers. The agreement, commencing October 1, 2025, covers ground plus five floors in KNG Tower 1 at Ashoknagar, MG Road — one of Bengaluru’s prime commercial hubs.As per the lease document, BlackRock will ..

Next Story
Infrastructure Transport

L&T Bags Rs 25–50 Bn Order for Mumbai-Ahmedabad Bullet Train Track Works

Larsen & Toubro’s (L&T) Transportation Infrastructure business has secured an order valued between Rs 25 crore and Rs 50 billion from the National High Speed Rail Corporation Limited (NHSRCL) for the Mumbai-Ahmedabad High Speed Rail (MAHSR) corridor.The contract, Package T1, involves the design, supply, construction, testing, and commissioning of 156 route km of high-speed ballastless track on a Design-Build Lump Sum Price basis. The stretch runs from Mumbai’s Bandra-Kurla Complex to Zaroli village in Gujarat and includes 21 km of underground track and 135 km of elevated viaduct.Se..

Advertisement

Subscribe to Our Newsletter

Get daily newsletters around different themes from Construction world.

STAY CONNECTED

Advertisement

Advertisement

Advertisement

Talk to us?