能源是人类社会赖以生存和发展的物质基础,在国民经济中具有重要的战略地位。随着社会经济的发展,能源的需求越来越大,因此需要对能源的使用进行有效的管理。能源消费预测是能源供需管理的理论前提,建立可靠的能源消费预测模型显得尤为...能源是人类社会赖以生存和发展的物质基础,在国民经济中具有重要的战略地位。随着社会经济的发展,能源的需求越来越大,因此需要对能源的使用进行有效的管理。能源消费预测是能源供需管理的理论前提,建立可靠的能源消费预测模型显得尤为重要。目前,已有的能源消费量预测模型主要包括单一模型和混合模型两大类。本研究提出了基于数据分组处理(group method of data handling,GMDH)的混合预测模型GHFM。该模型首先使用基于GMDH的自回归模型在原始能源消费时间序列上建模,预测其线性趋势,并得到残差序列(非线性子序列)。考虑到非线性子序列预测的复杂性,分别建立BP神经网络、支持向量回归机、遗传规划和RBF神经网络模型,再运用GMDH在非线性子序列上建立选择性组合预测模型,得到非线性子序列的组合预测值。最后,将两个部分的预测值进行整合得到总的能源消费量预测值。选取中国统计年鉴2014能源统计数据中的中国能源消费总量和石油消费总量数据进行实证分析,结果表明,GHFM模型与其他模型相比具有更好的预测效果。最后,给出了使用GHFM模型对2015-2020年中国能源消费总量的样本外预测值。展开更多
The creation of national energy strategy cannot proceed without accurate projections of future electricity consumption;this is because EC is intimately tied to other forms of energy,such as oil and natural gas.For the...The creation of national energy strategy cannot proceed without accurate projections of future electricity consumption;this is because EC is intimately tied to other forms of energy,such as oil and natural gas.For the purpose of determining and bettering overall energy consumption,there is an urgent requirement for accurate monitoring and calculation of EC at the building level using cutting-edge technology such as data analytics and the internet of things(IoT).Soft computing is a subset of AI that tries to design procedures that are more accurate and reliable,and it has proven to be an effective tool for solving a number of issues that are associated with the use of energy.The use of soft computing for energy prediction is an essential part of the solution to these kinds of challenges.This study presents an improved version of the Harris Hawks Optimization model by combining it with the IHHODL-ECP algorithm for use in Internet of Things settings.The IHHODL-ECP model that has been supplied acts as a useful instrument for the prediction of integrated energy consumption.In order for the raw electrical data to be compatible with the subsequent processing in the IHHODL-ECP model,it is necessary to perform a preprocessing step.The technique of prediction uses a combination of three different kinds of deep learning models,namely DNN,GRU,and DBN.In addition to this,the IHHO algorithm is used as a technique for making adjustments to the hyperparameters.The experimental result analysis of the IHHODL-ECP model is carried out under a variety of different aspects,and the comparison inquiry highlighted the advantages of the IHHODL-ECP model over other present approaches.According to the findings of the experiments conducted with an hourly time resolution,the IHHODL-ECP model obtained a MAPE value of 33.85,which was lower than those produced by the LR,LSTM,and CNN-LSTM models,which had MAPE values of 83.22,44.57,and 34.62 respectively.These findings provided evidence of the IHHODL-ECP model’s improved ability to provide accu展开更多
Most studies have conducted experiments on predicting energy consumption by integrating data formodel training.However, the process of centralizing data can cause problems of data leakage.Meanwhile,many laws and regul...Most studies have conducted experiments on predicting energy consumption by integrating data formodel training.However, the process of centralizing data can cause problems of data leakage.Meanwhile,many laws and regulationson data security and privacy have been enacted, making it difficult to centralize data, which can lead to a datasilo problem. Thus, to train the model while maintaining user privacy, we adopt a federated learning framework.However, in all classical federated learning frameworks secure aggregation, the Federated Averaging (FedAvg)method is used to directly weight the model parameters on average, which may have an adverse effect on te model.Therefore, we propose the Federated Reinforcement Learning (FedRL) model, which consists of multiple userscollaboratively training the model. Each household trains a local model on local data. These local data neverleave the local area, and only the encrypted parameters are uploaded to the central server to participate in thesecure aggregation of the global model. We improve FedAvg by incorporating a Q-learning algorithm to assignweights to each locally uploaded local model. And the model has improved predictive performance. We validatethe performance of the FedRL model by testing it on a real-world dataset and compare the experimental results withother models. The performance of our proposed method in most of the evaluation metrics is improved comparedto both the centralized and distributed models.展开更多
文摘能源是人类社会赖以生存和发展的物质基础,在国民经济中具有重要的战略地位。随着社会经济的发展,能源的需求越来越大,因此需要对能源的使用进行有效的管理。能源消费预测是能源供需管理的理论前提,建立可靠的能源消费预测模型显得尤为重要。目前,已有的能源消费量预测模型主要包括单一模型和混合模型两大类。本研究提出了基于数据分组处理(group method of data handling,GMDH)的混合预测模型GHFM。该模型首先使用基于GMDH的自回归模型在原始能源消费时间序列上建模,预测其线性趋势,并得到残差序列(非线性子序列)。考虑到非线性子序列预测的复杂性,分别建立BP神经网络、支持向量回归机、遗传规划和RBF神经网络模型,再运用GMDH在非线性子序列上建立选择性组合预测模型,得到非线性子序列的组合预测值。最后,将两个部分的预测值进行整合得到总的能源消费量预测值。选取中国统计年鉴2014能源统计数据中的中国能源消费总量和石油消费总量数据进行实证分析,结果表明,GHFM模型与其他模型相比具有更好的预测效果。最后,给出了使用GHFM模型对2015-2020年中国能源消费总量的样本外预测值。
文摘The creation of national energy strategy cannot proceed without accurate projections of future electricity consumption;this is because EC is intimately tied to other forms of energy,such as oil and natural gas.For the purpose of determining and bettering overall energy consumption,there is an urgent requirement for accurate monitoring and calculation of EC at the building level using cutting-edge technology such as data analytics and the internet of things(IoT).Soft computing is a subset of AI that tries to design procedures that are more accurate and reliable,and it has proven to be an effective tool for solving a number of issues that are associated with the use of energy.The use of soft computing for energy prediction is an essential part of the solution to these kinds of challenges.This study presents an improved version of the Harris Hawks Optimization model by combining it with the IHHODL-ECP algorithm for use in Internet of Things settings.The IHHODL-ECP model that has been supplied acts as a useful instrument for the prediction of integrated energy consumption.In order for the raw electrical data to be compatible with the subsequent processing in the IHHODL-ECP model,it is necessary to perform a preprocessing step.The technique of prediction uses a combination of three different kinds of deep learning models,namely DNN,GRU,and DBN.In addition to this,the IHHO algorithm is used as a technique for making adjustments to the hyperparameters.The experimental result analysis of the IHHODL-ECP model is carried out under a variety of different aspects,and the comparison inquiry highlighted the advantages of the IHHODL-ECP model over other present approaches.According to the findings of the experiments conducted with an hourly time resolution,the IHHODL-ECP model obtained a MAPE value of 33.85,which was lower than those produced by the LR,LSTM,and CNN-LSTM models,which had MAPE values of 83.22,44.57,and 34.62 respectively.These findings provided evidence of the IHHODL-ECP model’s improved ability to provide accu
基金supported by National Key R&D Program of China(No.2020YFC2006602)National Natural Science Foundation of China(Nos.62172324,62072324,61876217,6187612)+2 种基金University Natural Science Foundation of Jiangsu Province(No.21KJA520005)Primary Research and Development Plan of Jiangsu Province(No.BE2020026)Natural Science Foundation of Jiangsu Province(No.BK20190942).
文摘Most studies have conducted experiments on predicting energy consumption by integrating data formodel training.However, the process of centralizing data can cause problems of data leakage.Meanwhile,many laws and regulationson data security and privacy have been enacted, making it difficult to centralize data, which can lead to a datasilo problem. Thus, to train the model while maintaining user privacy, we adopt a federated learning framework.However, in all classical federated learning frameworks secure aggregation, the Federated Averaging (FedAvg)method is used to directly weight the model parameters on average, which may have an adverse effect on te model.Therefore, we propose the Federated Reinforcement Learning (FedRL) model, which consists of multiple userscollaboratively training the model. Each household trains a local model on local data. These local data neverleave the local area, and only the encrypted parameters are uploaded to the central server to participate in thesecure aggregation of the global model. We improve FedAvg by incorporating a Q-learning algorithm to assignweights to each locally uploaded local model. And the model has improved predictive performance. We validatethe performance of the FedRL model by testing it on a real-world dataset and compare the experimental results withother models. The performance of our proposed method in most of the evaluation metrics is improved comparedto both the centralized and distributed models.
