Decarbonisation policies aim at reducing fossil fuel based generation in favour of cleaner renewable energy sources. Changes in the generation mix to supply future electricity demand will require tools capable to emulate the bidding behaviour of new generation plants. Price forecasting tools lacking this feature and only based on historical data time series might soon become not satisfactory for this scope. This paper presents a methodology that, by considering hourly electricity generation offers (price, volumes) datasets, allows simulating future electricity wholesale's prices. This is done by taking into account new generation units and the dismissing of old (coal-based) units according to the demand and generation forecasts in the European Ten Year Network Development Plan (TYNDP) 2030 scenarios. Machine learning, clustering and distribution sampling techniques are used in this work to finally estimate prices distribution in 2030 in the biggest bidding zone of the Italian market. The results suggest that the prices obtained in the different scenarios do converge to those estimated by the TYNDP. The approach used bypasses the need to have access to all the transactions of a given market. Probability distributions are in fact enough in the proposed methodology to achieve similar results to those based on full knowledge of transaction datasets.

脱碳政策旨在减少以矿物燃料为基础的发电，而转向更清洁的可再生能源。为了满足未来的电力需求，发电方式的变化将需要能够模拟新一代电厂招标行为的工具。缺少此功能且仅基于历史数据时间序列的价格预测工具可能很快就无法满足此范围的需求。本文提出了一种方法，通过考虑每小时发电量（价格，数量）数据集，可以模拟未来的电力批发价格。这是根据2030年欧洲十年网络发展计划（TYNDP）方案中的需求和发电量预测，考虑了新一代机组和淘汰旧（煤基）机组后完成的。机器学习 在这项工作中，使用聚类和分布抽样技术最终估算了意大利市场最大招标区域2030年的价格分布。结果表明，在不同情况下获得的价格确实与TYNDP估算的价格一致。使用的方法绕过了访问给定市场的所有交易的需求。实际上，所提出的方法中的概率分布足以实现与基于交易数据集的全面知识所得出的结果相似的结果。使用的方法绕过了访问给定市场的所有交易的需求。实际上，所提出的方法中的概率分布足以实现与基于交易数据集的全面知识所得出的结果相似的结果。使用的方法绕过了访问给定市场的所有交易的需求。实际上，所提出的方法中的概率分布足以实现与基于交易数据集的全面知识所得出的结果相似的结果。