Registration Number
0225U000422,
(0123U102960)
, R & D reports
Title
Development of an information-measuring system for monitoring the technical condition of solar power plants
popup.stage_title
Розроблення інформаційно-вимірювальної системи моніторингу технічного стану сонячних електростанцій
Head
Sverdlova Anastasiia D., Доктор філософії
Registration Date
11-01-2025
Organization
Institute of General Energy National Academy of Sciences of Ukraine
popup.description1
Improve the monitoring system of solar power plants by utilizing hardware and software tools for collecting and analyzing big data in remote locations of solar power plant sites based on distributed computing infrastructures.
popup.description2
The purpose of the study is to improve the SPP monitoring system using hardware and software tools for collecting and analyzing big data in conditions of remoteness of solar power plant sites based on distributed computing infrastructures; to develop and test models for forecasting the generation of solar power plants, taking into account the influence of external factors, using various methods of time series analysis, including one-dimensional and multidimensional models, to achieve maximum accuracy of long-term forecasting.
Research methods: analytical review of technical solutions covered in the literature; univariate forecasting methods, multivariate analysis using neural networks and machine learning algorithms.
The paper reviews the approaches to monitoring SPPs, identifies their advantages and disadvantages, which makes it possible to choose the best practices and understand the shortcomings that need to be avoided.
The paper considers advanced methods of big data analysis that allow efficient processing of information from a large number of sources, increasing the flexibility and breadth of monitoring.
To ensure the effectiveness and reliability of the monitoring system, the author has determined what equipment is needed to monitor SPPs and what are the requirements for its maintenance.
A functional diagram has been formed that defines the key stages and components of the monitoring process, which is the basis for further system design.
The paper reviews univariate and multivariate methods for forecasting SPP generation. The multidimensional LSTM model was used to achieve a MAPE of 6-8%.
A new approach to forecasting energy loads using LSTM neural networks is proposed. To improve the prediction accuracy, the degradation of solar panels was taken into account by adding a decay factor to the state of the LSTM cell, which explicitly models the gradual decline in performance over time.
Product Description
popup.authors
Romanenko Vladyslav V.
Sverdlova Anastasiia D.
Haidurov Vladyslav V.
popup.nrat_date
2025-01-11
