Overview of electricity markets and the incorporation of renewable energy sources within them
Joas, Daniel (2023)
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe20231103142891
https://urn.fi/URN:NBN:fi-fe20231103142891
Tiivistelmä
The need for an increase in renewable electricity sources brought on by climate targets requires substantial changes in the electricity infrastructure and markets. The Nordic electricity system is an expansive and highly integrated system, with many stakeholders and components. One component, the electricity market is itself made up of several parts, all designed to complement each other and provide stability and electricity security to all stakeholders. An increase in renewable electricity sources will require new innovations, such as energy storage, and new practices in the markets, such as shorter imbalance periods. Furthermore, the predictability of future renewable production will also increase in importance when the renewable share increases, as it is one of the stone pillars in the renewable integration process.
In this thesis data from Denmark is used to perform wind power prediction using proven time series modelling approaches and models, such as Box-Jenkins ARIMA and ARIMAX models in addition to LSTM neural network models. This thesis showed that the more complex LSTM model outperformed the simpler ARIMA and ARIMAX models on several occasions, indicating that model complexity plays a significant part in the ability to model a stochastic process as wind power. Likewise, it was also shown that the autoregressive component in wind power is vital in the wind power forecasting models to produce accurate predictions. The findings in this thesis are supported by similar research on wind prediction.
Models used for wind power prediction are far more complicated than those used in this thesis, consisting of several more modelling steps than I employed, and utilising far more accurate data for the prediction. The consensus in the field of renewable electricity integration is that more flexibility is required by all stakeholders. Through an open and competitive market, stakeholders can be incentivised for their contributions, propelling us towards a greener, more sustainable energy future.
In this thesis data from Denmark is used to perform wind power prediction using proven time series modelling approaches and models, such as Box-Jenkins ARIMA and ARIMAX models in addition to LSTM neural network models. This thesis showed that the more complex LSTM model outperformed the simpler ARIMA and ARIMAX models on several occasions, indicating that model complexity plays a significant part in the ability to model a stochastic process as wind power. Likewise, it was also shown that the autoregressive component in wind power is vital in the wind power forecasting models to produce accurate predictions. The findings in this thesis are supported by similar research on wind prediction.
Models used for wind power prediction are far more complicated than those used in this thesis, consisting of several more modelling steps than I employed, and utilising far more accurate data for the prediction. The consensus in the field of renewable electricity integration is that more flexibility is required by all stakeholders. Through an open and competitive market, stakeholders can be incentivised for their contributions, propelling us towards a greener, more sustainable energy future.
Kokoelmat
- 512 Liiketaloustiede [502]