Modelselection,adaptation,and combination for transfer learning in wind and photovoltaic power forecasts

There is recent interest in using model hubs–a collection of pre-trained models–in computer vision tasks.To employ a model hub,we first select a source model and then adapt the model for the target to compensate for differences.There still needs to be more research on model selection and adaption for renewable power forecasts.In particular,none of the related work examines different model selection and adaptation strategies for neural network architectures.Also,none of the current studies investigates the influence of available training samples and considers seasonality in the evaluation.We close these gaps by conducting the first thorough experiment for model selection and adaptation for transfer learning in renewable power forecast,adopting recent developments from the field of computer vision on 667 wind and photovoltaic parks from six datasets.We simulate different amounts of training samples for each season to calculate informative forecast errors.We examine the marginal likelihood and forecast error for model selection for those amounts.Furthermore,we study four adaption strategies.As an extension of the current state of the art,we utilize a Bayesian linear regression for forecasting the response based on features extracted from a neural network.This approach outperforms the baseline with only seven days of training data and shows that fine-tuning is not beneficial with less than three months of data.We further show how combining multiple models through ensembles can significantly improve the model selection and adaptation approach such that we have a similar mean error with only 30 days of training data which is otherwise only possible with an entire year of training data.We achieve a mean error of 9.8 and 14 percent for the most realistic dataset for PV and wind with only seven days of training data.

Wind and Photovoltaic Power Time Series Data Aggregation Method Based on an Ensemble Clustering and Markov Chain

Reducing the input wind and photovoltaic power time series data can improve the efficiency of time sequential simulations.In this paper,a wind and photovoltaic power time series data aggregation method based on an ensemble clustering and Markov chain(ECMC)is proposed.The ECMC method can effectively reduce redundant information in the data.First,the wind and photovoltaic power time series data were divided into scenarios,and ensemble clustering was used to cluster the divided scenarios.At the same time,the Davies-Bouldin Index(DBI)is adopted to select the optimal number of clusters.Then,according to the temporal correlation between wind and photovoltaic scenarios,the wind and photovoltaic clustering results are merged and reduced to form a set of combined typical day scenarios that can reflect the characteristics of historical data within the calculation period.Finally,based on the Markov Chain,the state transition probability matrix of various combined typical day scenarios is constructed,and the aggregation state sequence of random length is generated,and then,the combined typical day scenarios of wind and photovoltaic were sampled in a sequential one-way sequence according to the state sequence and then are built into a representative wind and photovoltaic power time series aggregation sequence.A provincial power grid was chosen as an example to compare the multiple evaluation indexes of different aggregation methods.The results show that the ECMC aggregation method improves the accuracy and efficiency of time sequential simulations.

Optimal Scheduling for Flexible Regional Integrated Energy Systemwith Soft Open Point

The Regional Integrated Energy System(RIES)has brought new modes of development,utilization,conversion,storage of energy.The introduction of Soft Open Point(SOP)and the application of Power to Gas(P2G)technology will greatly deepen the coupling of the electricity-gas integrated energy system,improve the flexibility and safety of the operation of the power system,and bring a deal of benefits to the power system.On this background,an optimal dispatch model of RIES combined cold,heat,gas and electricity with SOP is proposed.Firstly,RIES architecture with SOP and P2G is designed and its mathematical model also is built.Secondly,on the basis of considering the optimal scheduling of combined cold,heat,gas and electricity,the optimal scheduling model for RIES was established.After that,the original model is transformed into a mixed-integer second-order cone programming model by using linearization and second-order cone relaxation techniques,and the CPLEX solver is invoked to solve the optimization problem.Finally,the modified IEEE 33-bus systemis used to analyze the benefits of SOP,P2G technology and lithium bromide absorption chillers in reducing systemnetwork loss and cost,as well as improving the system’s ability to absorb wind and solar and operating safety.