The surge in demand for renewable energy to combat the ever-escalating climate crisis promotes development of the energy-saving,carbon saving and reduction technologies.Shallow ground-source heat pump(GSHP)system is a...The surge in demand for renewable energy to combat the ever-escalating climate crisis promotes development of the energy-saving,carbon saving and reduction technologies.Shallow ground-source heat pump(GSHP)system is a promising carbon reduction technology that can stably and effectively exploit subsurface geothermal energy by taking advantage of load-bearing structural elements as heat transfer medium.However,the transformation of conventional geo-structures(e.g.piles)into heat exchangers between the ground and superstructures can potentially induce variable thermal axial stresses and displacements in piles.Traditional energy pile analysis methods often rely on deterministic and homogeneous soil parameter profiles for investigating thermo-mechanical soil-structure interaction,without consideration of soil spatial variability,model uncertainty or statistical uncertainty associated with interpolation of soil parameter profiles from limited site-specific measurements.In this study,a random finite difference model(FDM)is proposed to investigate the thermo-mechanical load-transfer mechanism of energy piles in granular soils.Spatially varying soil parameter profile is interpreted from limited site-specific measurements using Bayesian compressive sensing(BCS)with proper considering of soil spatial variability and other uncertainties in the framework of Monte Carlo simulation(MCS).Performance of the proposed method is demonstrated using an illustrative example.Results indicate that the proposed method enables an accurate evaluation of thermally induced axial stress/displacement and variation in null point(NP)location with quantified uncertainty.A series of sensitivity analyses are also carried out to assess effects of the pile-superstructure stiffness and measurement data number on the performance of the proposed method,leading to useful insights.展开更多
城市电网在发生N-1故障后,极可能新增运行风险,导致N-1-1时出现大面积停电事故。为管控城市电网N-1后运行风险,该文提出一种改进双智能体竞争双深度Q网络(dueling double deep Q network,D3QN)的城市电网N-1风险管控转供策略。根据风险...城市电网在发生N-1故障后,极可能新增运行风险,导致N-1-1时出现大面积停电事故。为管控城市电网N-1后运行风险,该文提出一种改进双智能体竞争双深度Q网络(dueling double deep Q network,D3QN)的城市电网N-1风险管控转供策略。根据风险管控原则,提出一种无需额外历史数据、考虑备自投装置、单供变电站风险和单供负荷母线风险的N-1场景指标;建立计及动作次序、指标间关系的负荷转供三阶段求解模型。以含预动作-变化探索值选择策略的改进双智能体D3QN方法,将负荷转供分为多个子转供环节学习,使转供思路清晰化,对动作空间进行降维,提高训练寻优效果,得到管控N-1风险的负荷转供策略。通过城市电网多场景算例分析,验证该文模型和方法的有效性。展开更多
基于数据驱动的方法已广泛应用于电力负荷预测领域,以提升预测精度。然而,当售电公司接入新用户时,由于缺乏用户历史用电数据,常规数据驱动方法的适用性会受到一定限制。为解决这一问题,文章提出了一种基于域对抗迁移网络(domain advers...基于数据驱动的方法已广泛应用于电力负荷预测领域,以提升预测精度。然而,当售电公司接入新用户时,由于缺乏用户历史用电数据,常规数据驱动方法的适用性会受到一定限制。为解决这一问题,文章提出了一种基于域对抗迁移网络(domain adversarial transfer network,DATN)的短期电力负荷预测方法。该模型利用Transformer模型作为特征提取器,以捕捉负荷数据中的动态特征和时间依赖性。随后,负荷预测器基于这些特征精准预测未来的负荷情况。通过域判别器与特征提取器的对抗学习,确保模型能够学习到深层域不变特征,同时结合多核最大均值差异(multi-kernel maximum mean discrepancy,MK-MMD)和相关性对齐(correlation alignment,CORAL)进一步减小源域与目标域数据的分布差异。所提模型在南方某省工业用户的用电数据上进行了验证,实验结果表明,在小样本场景下,该方法具备较好的预测精度和场景适应性。展开更多
基金The work described in this paper was supported by grants from the Research Grant Council of Hong Kong Special Administrative Region,China(Grants Nos.CityU 11213119 and CityU 11202121).The financial support is gratefully acknowledged.
文摘The surge in demand for renewable energy to combat the ever-escalating climate crisis promotes development of the energy-saving,carbon saving and reduction technologies.Shallow ground-source heat pump(GSHP)system is a promising carbon reduction technology that can stably and effectively exploit subsurface geothermal energy by taking advantage of load-bearing structural elements as heat transfer medium.However,the transformation of conventional geo-structures(e.g.piles)into heat exchangers between the ground and superstructures can potentially induce variable thermal axial stresses and displacements in piles.Traditional energy pile analysis methods often rely on deterministic and homogeneous soil parameter profiles for investigating thermo-mechanical soil-structure interaction,without consideration of soil spatial variability,model uncertainty or statistical uncertainty associated with interpolation of soil parameter profiles from limited site-specific measurements.In this study,a random finite difference model(FDM)is proposed to investigate the thermo-mechanical load-transfer mechanism of energy piles in granular soils.Spatially varying soil parameter profile is interpreted from limited site-specific measurements using Bayesian compressive sensing(BCS)with proper considering of soil spatial variability and other uncertainties in the framework of Monte Carlo simulation(MCS).Performance of the proposed method is demonstrated using an illustrative example.Results indicate that the proposed method enables an accurate evaluation of thermally induced axial stress/displacement and variation in null point(NP)location with quantified uncertainty.A series of sensitivity analyses are also carried out to assess effects of the pile-superstructure stiffness and measurement data number on the performance of the proposed method,leading to useful insights.
文摘城市电网在发生N-1故障后,极可能新增运行风险,导致N-1-1时出现大面积停电事故。为管控城市电网N-1后运行风险,该文提出一种改进双智能体竞争双深度Q网络(dueling double deep Q network,D3QN)的城市电网N-1风险管控转供策略。根据风险管控原则,提出一种无需额外历史数据、考虑备自投装置、单供变电站风险和单供负荷母线风险的N-1场景指标;建立计及动作次序、指标间关系的负荷转供三阶段求解模型。以含预动作-变化探索值选择策略的改进双智能体D3QN方法,将负荷转供分为多个子转供环节学习,使转供思路清晰化,对动作空间进行降维,提高训练寻优效果,得到管控N-1风险的负荷转供策略。通过城市电网多场景算例分析,验证该文模型和方法的有效性。
文摘基于数据驱动的方法已广泛应用于电力负荷预测领域,以提升预测精度。然而,当售电公司接入新用户时,由于缺乏用户历史用电数据,常规数据驱动方法的适用性会受到一定限制。为解决这一问题,文章提出了一种基于域对抗迁移网络(domain adversarial transfer network,DATN)的短期电力负荷预测方法。该模型利用Transformer模型作为特征提取器,以捕捉负荷数据中的动态特征和时间依赖性。随后,负荷预测器基于这些特征精准预测未来的负荷情况。通过域判别器与特征提取器的对抗学习,确保模型能够学习到深层域不变特征,同时结合多核最大均值差异(multi-kernel maximum mean discrepancy,MK-MMD)和相关性对齐(correlation alignment,CORAL)进一步减小源域与目标域数据的分布差异。所提模型在南方某省工业用户的用电数据上进行了验证,实验结果表明,在小样本场景下,该方法具备较好的预测精度和场景适应性。
