Generating Synthetic Data to Reduce Prediction Error of Energy Consumption

Renewable and nonrenewable energy sources are widely incorporated for solar and wind energy that produces electricity without increasing carbon dioxide emissions.Energy industries worldwide are trying hard to predict future energy consumption that could eliminate over or under contracting energy resources and unnecessary financing.Machine learning techniques for predicting energy are the trending solution to overcome the challenges faced by energy companies.The basic need for machine learning algorithms to be trained for accurate prediction requires a considerable amount of data.Another critical factor is balancing the data for enhanced prediction.Data Augmentation is a technique used for increasing the data available for training.Synthetic data are the generation of new data which can be trained to improve the accuracy of prediction models.In this paper,we propose a model that takes time series energy consumption data as input,pre-processes the data,and then uses multiple augmentation techniques and generative adversarial networks to generate synthetic data which when combined with the original data,reduces energy consumption prediction error.We propose TGAN-skip-Improved-WGAN-GP to generate synthetic energy consumption time series tabular data.We modify TGANwith skip connections,then improveWGANGPby defining a consistency term,and finally use the architecture of improved WGAN-GP for training TGAN-skip.We used various evaluation metrics and visual representation to compare the performance of our proposed model.We also measured prediction accuracy along with mean and maximum error generated while predicting with different variations of augmented and synthetic data with original data.The mode collapse problemcould be handled by TGAN-skip-Improved-WGAN-GP model and it also converged faster than existing GAN models for synthetic data generation.The experiment result shows that our proposed technique of combining synthetic data with original data could significantly reduce the prediction error rate and increase the prediction accuracy of energy consumption.

Long-Term Energy Forecasting System Based on LSTM and Deep Extreme Machine Learning

Due to the development of diversified and flexible building energy resources,the balancing energy supply and demand especially in smart build-ings caused an increasing problem.Energy forecasting is necessary to address building energy issues and comfort challenges that drive urbanization and consequent increases in energy consumption.Recently,their management has a great significance as resources become scarcer and their emissions increase.In this article,we propose an intelligent energy forecasting method based on hybrid deep learning,in which the data collected by the smart home through meters is put into the pre-evaluation step.Next,the refined data is the input of a Long Short-Term Memory(LSTM)network,which captures the spatio-temporal correlations from the sequence and generates the feature maps.The output feature map is passed into a Deep Extreme Machine Learning network(with seven hidden layers)for learning,which provides the final prediction.Compared to existing techniques,the LSTM-DELM model offers better prediction results.The simulation values demonstrate the superior performance of the proposed model.

A Water Level Forecast of Pattani River in the Southern of Thailand by Deep Learning

Nowadays, the deep learning methods are widely applied to analyze and predict the trend of various disaster events and offer the alternatives to make the appropriate decisions. These support the water resource management and the short-term planning. In this paper, the water levels of the Pattani River in the Southern of Thailand have been predicted every hour of 7 days forecast. Time Series Transformer and Linear Regression were applied in this work. The results of both were the water levels forecast that had the high accuracy. Moreover, the water levels forecasting dashboard was developed for using to monitor the water levels at the Pattani River as well.

市场环境下考虑全周期经济效益的工业园区共享储能优化配置

为提升用户侧储能运行效率、改善投资成效,提出一种在市场环境下考虑全周期经济效益的工业园区共享储能(SES)优化配置方法。一方面,通过协调不同用户间的差异化调节需求,减少储能容量要求;另一方面,通过整合用户与SES的灵活调节能力,参与需求响应市场拓宽盈利渠道,并且考虑了SES全运行周期经济效益测算以降低投资风险。首先,结合电力市场交易规则,提出了多工业用户组建合作联盟的园区SES运营模式。其次,以运营周期内联盟总成本最小为目标,建立SES双层优化配置模型,其中,上层模型旨在形成最大化投资成效的SES规划方案,而下层模型则综合考虑分时电价、需求响应违约风险等因素形成储能的最优投标调度方式,并结合市场时序演变规律精准量化SES在全运行周期内的运营收益,对上层结果进行修正。接着,利用近似KKT(Karush-Kuhn-Tucker)条件将该模型转化为单层模型进行求解,结合雨流计数法与迭代法量化SES容量衰减对其配置方案的影响,并利用双边Shapley值法分摊各工业用户的投资成本。最后,算例仿真验证了所提方法的有效性,并且分析了储能盈利模式、SES容量衰减以及需求响应违约风险等因素对SES投资经济效益的影响。

DFA-ODENets:面向周期多阶段复杂系统的预测仿真框架

部分复杂系统受内外部因素影响在运行时会呈现出周期性的阶段变化,且在不同阶段具有完全不同的动态特性.因此在使用数据驱动方法解决此类系统的预测和仿真问题时,使用单一结构模型难以准确地学习系统在不同阶段的动态特性.本研究提出了基于确定性有限状态机-常微分方程网络的预测仿真框架(DFA-ODENets),以建模周期多阶段系统.该模型由多个ODENet组成,每个ODENet能够从不规则采样的序列数据中学习系统在各个阶段内的动态特性.同时模型集成了基于确定性有限状态自动机思想的阶段转换预测器以实现模型预测时在不同阶段之间自动转换.最后,将DFA-ODENet框架应用于某计算中心制冷系统的预测仿真场景中.模型能够在给定系统运行过程中的服务器负载和环境温度下模拟系统运行过程,并对系统的制冷功率、进气口温度等主要输出变量进行预测.其中,对于制冷系统能耗预测的平均相对误差在5%以内.同时,利用制冷系统仿真模型优化了系统停止制冷时的温度设定值,通过仿真实验表明该优化最高可以节省18%的制冷能耗.

基于模态分解和深度学习的煤矿微震时序预测方法

微震监测数据对煤矿冲击地压等动力灾害预测和预警具有重要作用,利用历史微震监测数据来预测未来微震事件的演化特征是提高冲击灾害预测和预警时效性与准确性的有效方法。然而,由于微震数据是典型的非平稳时间序列,一般的时序预测方法很难进行准确的预测。为此,提出了一种基于模态分解技术和深度学习方法相结合的微震时序预测方法。该方法首先使用自适应噪声完备集合经验模态分解(CEEMDAN)将微震数据分解为多个本征模态序列。然后,通过样本熵将分解后的序列重构为高频、低频和趋势序列。接着,利用变分模态分解(VMD)将高频和低频序列再次分解为多个新的本征模态序列,用作微震数据的多元特征时间序列。最后,将一定时窗内的微震多元特征时间序列和历史微震数据作为输入,下一时刻的微震监测数据作为输出,建立了卷积神经网络(CNN)和长短期记忆网络(LSTM)相结合的微震时序预测模型,并采用贝叶斯优化方法对模型的超参数进行寻优,提高微震预测的准确性。依托新巨龙煤矿多个工作面微震监测数据,使用该方法对微震日最大能量、日平均能量和日频次进行了预测,并针对微震日最大能量数据进行了系列对比试验研究。结果表明,提出的方法能够较好地预测微震事件演化趋势,预测结果和实际监测值误差较小,具有良好的预测性能和泛化性能。

基于自注意力的电力数据能耗预测算法Dual-channel LSTM

电力能源需求随不可预测事件波动性加剧,为实现准确预测建筑用电能耗,提出一种电力数据能耗预测算法Dual-channel LSTM。算法框架主要分为3个模块,首先通过T-LSTM模块对时序存在缺失值的电力数据预处理,使其时序排列规则,随后输入双通道注意力特征模块,该模块引入自注意力机制来引导权重分配,最后由LSTM和FCN结合的预测模块输出结果。在AEP和IHEPC公开电力数据集上评估了该算法,Dual-channel的损失函数MAE/MSE分别仅有0.12/0.06和0.06/0.04,通过消融实验发现,引入T-LSTM对算法性能起决定性影响,引入自注意力机制对算法起辅助性作用,所提方法提高了预测精度和鲁棒性。

隐私保护下融合联邦学习和LSTM的少数据综合能源多元负荷预测

对于用能数据不足的综合能源系统,借助相似系统的丰富数据可以为其建立高精度的多元负荷预测模型,然而,受数据安全等因素的限制,很多系统并不愿意共享自身数据。联邦学习为处理隐私保护下的少数据综合能源多元负荷预测问题提供了一个重要的思路,但是现有方法依然存在相似参与方识别精度不高等不足。鉴于此,本文提出一种融合联邦学习和长短期记忆网络(long short-term memory,LSTM)的少数据综合能源多元负荷预测方法(multitask learning based on shared dot product confidentiality under federated learning,MT-SDPFL)。首先,给出一种基于共享向量点积保密协议的相似参与方识别方法,用来从诸多可用的综合能源系统中选出最为相似的参与方;接着,使用参数共享联邦学习算法对选中的各参与方联合训练,结合LSTM和finetune技术建立每个参与方的多元负荷预测模型。将所提方法应用于多个实际能源系统,实验结果表明,该方法可以在数据稀疏的情况下取得高精度的多源负荷预测结果。

基于大语言模型绿电预测和绿电交易的园区综合能源系统集群多目标协同运行方法

为实现传统工业园区数字化和智能化升级,助力区域高质量发展,亟需园区智能化调度模型。为此,该文结合智慧园区管理系统和园区综合能源系统物理模型建立园区综合能源系统集群架构,提出了园区综合能源系统集群绿电交易三阶段协同运行方法,以解决多园区综合能源系统绿电交易的问题,实现分布式绿电的精准预测以及就地消纳。首先,基于大语言模型LLAMA-7B实现绿电预测,进一步以绿电功率划分购售电园区。其次,基于绿电价格配额曲线预测模型和动态绿电定价策略,制定园区间绿电交易差异价格。在此基础上,建立了多目标低碳经济优化运行模型,从而解决绿电交易所带来的经济因素和环境因素的矛盾问题。算例分析表明:所提出的模型可以统筹规划园区综合能源系统集群的经济成本、实际碳排放量和新能源利用率,对多园区综合能源系统的智能调度具有积极的推动作用。

基于Informer的负荷及光伏出力系数预测

为了支持可再生能源的布局及装机容量规划,长时间序列的电力负荷和光伏出力系数预测至关重要。光伏出力系数反映了光伏发电系统实际运行中的发电效率,但由于难以准确预测来年的气象信息,每日最大光伏出力系数的预测具有挑战性。为了克服这个限制,提出了利用每7 d计算1次的每日最大光伏出力系数的最大值和最小值构建一个包络线,通过预测该包络线的上限和下限,提供每日最大光伏出力系数可能的区间。这种包络线建模方式有助于在克服气象信息不确定性的同时提供更为鲁棒和可靠的预测结果。选用Informer模型作为预测框架,并与Transformer,LSTM和RNN模型进行了比较。基于实际电力负荷数据序列和光伏出力系数包络线上、下限数据序列进行仿真试验,验证了Informer模型的可行性和良好的预测精度。

基于灰色-时序组合模型改进的新能源企业价值评估研究

近年来,新能源行业飞速发展,对新能源企业采用适当方法进行评估可为投资者提供可靠的价值参考,正确引导市场和投资者决策。文章基于已有评估模型,借助灰色-时序组合模型预测企业未来营业收入,对自由现金流量模型进行改进,并以代表性新能源企业——金风科技作为实证分析对象,检验该种组合预测方法在企业自由现金流量评估模型中的适用性,以此来丰富现有企业价值评估方法体系。

A New Energy Efficient Data Gathering Approach in Wireless Sensor Networks

Data gathering in wireless sensor networks is one of the important operations in these networks. These operations require energy consumption. Due to the limited energy of nodes, the energy productivity should be considered as a key objective in design of sensor networks. Therefore the clustering is a suitable method that used in energy consumption management. For this purpose many methods have been proposed. Between these methods the LEACH algorithm has been attend as a basic method. This algorithm uses distributed clustering method for data gathering and aggregation. The LEACH-C method that is the improvement of LEACH, which performs the clustering in centralized mode. In this method, collecting the energy level of information of every node directly in each period increases the energy cost. Also the phenomenon that is seen by sensor nodes continually change over time. Thereby the information received by nodes is correlated. Sending time correlated data in the network cause to energy dissipation. TINA method and its improvement have been proposed in order to not sending correlated data. These approaches have reported errors. In this paper, the idea of not sending time correlated data of nodes has been considered by using the time series function. Also, a model to estimate the remaining energy of nodes by the base station has been presented. Finally, a method has been proposed to aware the base station from the number of correlated data in each node as the estimation of energy will be more precise. The proposed ideas have been implemented over the LEACH-C protocol. Evaluation results showed that the proposed methods had a better performance in energy consumption and the lifetime of the network in comparison with similar methods.

基于自注意力机制和CNN-LSTM的空战目标机动轨迹预测

空战目标机动轨迹是有丰富时空特征的多维时间序列,具有高度复杂性和不确定性。针对现阶段轨迹预测运动学模型建立困难、时序预测的方法难以提取时空特征且只能单一的从T到T+1时刻的顺序式训练的问题,文中提出了一种自注意力机制(self-attention,ATT)和卷积神经网络(convolutional neural network,CNN)-长短期记忆网络(long short-term memory,LSTM)结合的模型(CNN-LSTM-ATT)。离线状态下训练模型,获得的最优模型可以实现目标机动轨迹的高精度预测。文中模型与CNN-LSTM、LSTM模型进行单步预测对比分析,具有良好的单步预测和不同过载机动预测的能力。考虑到电磁干扰和复杂环境导致传输数据的误差和缺失,进行了目标轨迹的5步预测,预测结果和评价指标均优于CNN-LSTM、LSTM模型。

“小批量物料的生产安排”问题解析

给出2022年“高教社杯”全国大学生数学建模竞赛E题“小批量物料的生产安排”的求解方法,并针对学生在参赛论文中出现的问题作了简要的说明与点评.

时间序列分析在低渗透基岩地下水位预测中的应用

为深入掌握低渗透基岩地下水水位波动的宏观和微观特征,需要开展对地下水水位数据的拟合预测研究。以高放废物地质处置北山预选区新场地段低渗透基岩为研究对象,采用二次指数平滑法、ARMA模型和BP神经网络模型对地下水径流路径上的3口深部和3口浅部监测孔水位数据开展研究工作。对于地下水位宏观波动特征的解译,BP神经网络模型拟合精度要高于另外两种模型,且更加适用于具有明显非线性特征的水位序列;ARMA模型和二次指数平滑法模型拟合效果相当,但更适用于预测变化相对平缓的监测水位数据,且对于水位突变特征的拟合ARMA模型要优于二次指数平滑法模型。对于浅部监测孔地下水水位微动态的解译,BP神经网络模型优于二次指数平滑法模型和ARMA模型,预测的地下水位微动态变化周期和振幅基本与实测值一致。

基于CNN-Transformer的城区地下水位预测

提出了一种将Transformer与卷积神经网络(CNN)相结合的城区地下水位预测模型。Transformer模型能够提取地下水位在时间序列上包含的关键信息,有效提升了模型的长时间预测能力;CNN能获取相邻监测站点地下水位数据之间的空间关联信息,使信息的提取更加丰富。使用开源地下水位数据集对模型进行训练,并进行仿真验证。仿真结果表明,在预测未来12个时刻的地下水位值时,CNN-Transformer模型预测结果整体的均方根误差值相比于循环神经网络(RNN)系列模型从0.2507米降到0.1427米,在未来第12个时刻的均方根误差也仅为0.2309米,验证了上述模型能实现长时间高精度的地下水位预测。

基于ARIMA-CNN-LSTM模型的黄河开封段水位预测研究

为准确预测水文条件复杂的黄河开封段水位变化,提出一种基于ARIMA-CNN-LSTM的多变量水位预测模型。该模型通过综合考虑水位的多重影响因素,结合卷积神经网络(CNN)和长短时记忆网络(LSTM)来学习数据中的非线性特征,同时应用ARIMA模型进行参数校正,从而实现对黄河开封段水位未来一段时间的预测。结果表明:相较于LSTM模型、CNN-LSTM模型、ARIMA模型以及BP神经网络模型,ARIMA-CNN-LSTM模型的预测精度更高,对峰值反应更灵敏,可以更加精准地预测未来一段时间的黄河开封段水位变化。

Holt-Winters Algorithm to Predict the Stock Value Using Recurrent Neural Network

Prediction of stock market value is highly risky because it is based on the concept of Time Series forecasting system that can be used for investments in a safe environment with minimized chances of loss.The proposed model uses a real time dataset offifteen Stocks as input into the system and based on the data,predicts or forecast future stock prices of different companies belonging to different sectors.The dataset includes approximatelyfifteen companies from different sectors and forecasts their results based on which the user can decide whether to invest in the particular company or not;the forecasting is done for the next quarter.Our model uses 3 main concepts for forecasting results.Thefirst one is for stocks that show periodic change throughout the season,the‘Holt-Winters Triple Exponential Smoothing’.3 basic things taken into conclusion by this algorithm are Base Level,Trend Level and Seasoning Factor.The value of all these are calculated by us and then decomposition of all these factors is done by the Holt-Winters Algorithm.The second concept is‘Recurrent Neural Network’.The specific model of recurrent neural network that is being used is Long-Short Term Memory and it’s the same as the Normal Neural Network,the only difference is that each intermediate cell is a memory cell and retails its value till the next feedback loop.The third concept is Recommendation System whichfilters and predict the rating based on the different factors.

基于内核时变回归模型的电能预测分析与研究

为实现“双碳”发展目标和满足新型电力系统应用需求,亟需对用电进行精准预测。为了应对周期长、变化幅度大的数据,将KTR模型应用于电能负荷预测的实际场景中。该模型在时变系数回归的方法上进行改进,能够应对较长的时间序列,避免出现过拟合的情况;以及根据不同数据变化情况自适应地使用不同的核函数,保证模型学习与数据特征匹配。实验结果表明,使用通过最佳参数构建的KTR模型进行预测,其总体的电能负荷数据预测值和原始值的SMAPE为8.46%。此外,将文中方法与Prophet和SARIMA模型预测结果进行了对比,结果表明,文中方法的预测精度比另外两种模型分别高2.57%和9.23%,验证了该方法电能预测的准确性。

基于N-BEATS算法的峡江水利枢纽工程水库水位预测

基于N-BEATS算法,针对峡江水利枢纽工程的水库水位预测问题进行研究。首先介绍峡江水利枢纽工程的重要性和水库水位预测的挑战,以及现有方法的优缺点。然后详细介绍N-BEATS算法的原理和特点。最后通过实验部分,展示基于N-BEATS算法的水库水位预测结果,并对算法的优势和局限性进行分析。同时,与其他方法进行对比和分析,并探讨N-BEATS算法对时间序列分解的可解释性和意义。结果显示N-BEATS算法在峡江水库水位预测方面具有优越性能,可为水文预测和管理决策提供有价值的参考。