Machine Learning-based Electric Load Forecasting for Peak Demand Control in Smart Grid

Increasing energy demands due to factors such as population, globalization, and industrialization has led to increased challenges for existing energy infrastructure. Efficient ways of energy generation and energy consumption like smart grids and smart homes are implemented to face these challenges with reliable, cheap, and easily available sources of energy. Grid integration of renewable energy and other clean distributed generation is increasing continuously to reduce carbon and other air pollutants emissions. But the integration of distributed energy sources and increase in electric demand enhance instability in the grid. Short-term electrical load forecasting reduces the grid fluctuation and enhances the robustness and power quality of the grid. Electrical load forecasting in advance on the basic historical data modelling plays a crucial role in peak electrical demand control, reinforcement of the grid demand, and generation balancing with cost reduction. But accurate forecasting of electrical data is a very challenging task due to the nonstationary and nonlinearly nature of the data. Machine learning and artificial intelligence have recognized more accurate and reliable load forecasting methods based on historical load data. The purpose of this study is to model the electrical load of Jajpur, Orissa Grid for forecasting of load using regression type machine learning algorithms Gaussian process regression (GPR). The historical electrical data and whether data of Jajpur is taken for modelling and simulation and the data is decided in such a way that the model will be considered to learn the connection among past, current, and future dependent variables, factors, and the relationship among data. Based on this modelling of data the network will be able to forecast the peak load of the electric grid one day ahead. The study is very helpful in grid stability and peak load control management.     

基于分布式光伏发电的多目标分时电价优化策略

现有电力定价研究大多为峰谷分时定价,时段划分方式单一且大多采用传统非支配排序遗传算法-II求解多目标问题。针对这个问题,提出一种基于分布式光伏发电的多目标分时电价优化策略。建立用电量与电价响应模型,基于等效负荷进行时段划分,以负荷方差最小,等效负荷的峰谷差最小,用户满意度指数最大为目标,建立多目标非线性分布式光伏分时定价模型,并提出基于邻域搜索的多目标遗传算法求解。数值实验结果表明,供电稳定性提高了37.77%,分布式光伏发电的利用率提高了4.51%,用户满意度为74.3%;且提出的求解算法要优于常用的非支配排序遗传算法-II,表明本文提出的定价策略是有效的。     

超临界循环流化床机组全负荷段深度调峰方法研究

随着新能源发电装机容量不断扩大,火力发电市场受到严重挤压,为了满足电网深度调峰的需求,火电厂需要承担起深度调峰的重任,循环流化床机组可以采用停炉不停机的方式参与深度调峰。以山西某电厂超临界350 MW循环流化床机组为研究对象,分析了超临界循环流化床机组2种典型汽水系统全负荷段深度调峰方法,对其中的操作要点进行了详细的阐述,并就操作过程中需要特别关注的问题进行了分析,也提出了相应的防控措施。分析表明:超临界循环流化床机组可以采用2种典型汽水系统实现全负荷段深度调峰;加装炉水循环泵系统投资大、操作复杂,增设贮水箱到除氧器管路投资小、操作简单,但是加装炉水循环泵系统参与全负荷段调峰时,各系统稳定性较好。     

电动汽车智能充放电控制与应用综述

随着电动汽车规模的增大,电动汽车接入电网对电力系统运行与控制的影响不容忽视。电动汽车作为一种可控负荷,对其进行充放电控制可以有效削弱充电负荷带来的不利影响,同时还能起到削峰填谷、促进新能源消纳的作用,这将成为电力系统运行控制的一种重要手段。给出了充电负荷建模需要考虑的因素,总结了建立电动汽车负荷预测模型的方法。归纳了电动汽车参与电网调度的可行方法,并分析了不同方法的特点。同时,为了提高电动汽车参与调度的积极性,介绍了用区块链完成电动汽车电力交易的架构与方法。最后,对尚未解决的问题和可能的研究方向进行了讨论。     

主动配电网规划方法研究

分布式电源接入配电网后对供电可靠性、电压分布、网损均产生一定影响,传统配电网规划方法已无法适用。构建主动配电网（active distribution network,ADN）规划设计体系结构,从电网现状、负荷预测等方面制定主动配电网规划内容,重点研究基于分布式电源风险度出力置信区间的源网荷一体化平衡及网荷协同性规划方法,最后给出主动配电网电气校验方法,验证方案制定的合理性。     

Deep Feature Fusion Model for Sentence Semantic Matching

Sentence semantic matching (SSM) is a fundamental research in solving natural language processing tasks such as question answering and machine translation. The latest SSM research benefits from deep learning techniques by incorporating attention mechanism to semantically match given sentences. However, how to fully capture the semantic context without losing significant features for sentence encoding is still a challenge. To address this challenge, we propose a deep feature fusion model and integrate it into the most popular deep learning architecture for sentence matching task. The integrated architecture mainly consists of embedding layer, deep feature fusion layer, matching layer and prediction layer. In addition, we also compare the commonly used loss function, and propose a novel hybrid loss function integrating MSE and cross entropy together, considering confidence interval and threshold setting to preserve the indistinguishable instances in training process. To evaluate our model performance, we experiment on two real world public data sets: LCQMC and Quora. The experiment results demonstrate that our model outperforms the most existing advanced deep learning models for sentence matching, benefited from our enhanced loss function and deep feature fusion model for capturing semantic context.     

Forecasting Model Based on Information-Granulated GA-SVR and ARIMA for Producer Price Index

The accuracy of predicting the Producer Price Index (PPI) plays an indispensable role in government economic work. However, it is difficult to forecast the PPI. In our research, we first propose an unprecedented hybrid model based on fuzzy information granulation that integrates the GA-SVR and ARIMA (Autoregressive Integrated Moving Average Model) models. The fuzzy-information-granulation-based GA-SVR-ARIMA hybrid model is intended to deal with the problem of imprecision in PPI estimation. The proposed model adopts the fuzzy information-granulation algorithm to pre-classification-process monthly training samples of the PPI, and produced three different sequences of fuzzy information granules, whose Support Vector Regression (SVR) machine forecast models were separately established for their Genetic Algorithm (GA) optimization parameters. Finally, the residual errors of the GA-SVR model were rectified through ARIMA modeling, and the PPI estimate was reached. Research shows that the PPI value predicted by this hybrid model is more accurate than that predicted by other models, including ARIMA, GRNN, and GA-SVR, following several comparative experiments. Research also indicates the precision and validation of the PPI prediction of the hybrid model and demonstrates that the model has consistent ability to leverage the forecasting advantage of GA-SVR in non-linear space and of ARIMA in linear space.     

An Efficient Internet Traffic Classification System Using Deep Learning for IoT

Internet of Things (IoT) defines a network of devices connected to the internet and sharing a massive amount of data between each other and a central location. These IoT devices are connected to a network therefore prone to attacks. Various management tasks and network operations such as security, intrusion detection, Quality-of-Service provisioning, performance monitoring, resource provisioning, and traffic engineering require traffic classification. Due to the ineffectiveness of traditional classification schemes, such as port-based and payload-based methods, researchers proposed machine learning-based traffic classification systems based on shallow neural networks. Furthermore, machine learning-based models incline to misclassify internet traffic due to improper feature selection. In this research, an efficient multilayer deep learning based classification system is presented to overcome these challenges that can classify internet traffic. To examine the performance of the proposed technique, Moore-dataset is used for training the classifier. The proposed scheme takes the pre-processed data and extracts the flow features using a deep neural network (DNN). In particular, the maximum entropy classifier is used to classify the internet traffic. The experimental results show that the proposed hybrid deep learning algorithm is effective and achieved high accuracy for internet traffic classification, i.e., 99.23%. Furthermore, the proposed algorithm achieved the highest accuracy compared to the support vector machine (SVM) based classification technique and k-nearest neighbours (KNNs) based classification technique.     

基于灰色理论的电力系统短期负荷预测

本文主要研究短期电力负荷预报的应用,在分析了目前短期电力负荷预测的现状及各种预报采用的数学模型基础上,结合黑龙江和电网实际负荷数据,在算法上采用了灰色预测中的GM（1,1）改进模型和应用matlab编程,通过对原始数据的平滑处理进行负荷预报,提高了预报准确度,验证了该方法的有效性和可行性。     

ON THE FORECASTING OF ECONOMICALLY OPTIMAL APPROXIMATIONS TO LOAD DURATION CURVES

This paper develops a methodology for forecasting economically optimal approximations to the load duration curve. A dynamic programming algorithm serves as the basis of the optimal approximation, yielding discrete approximations over a known horizon. These approximations are related to a set of economic and weather variables and are, in turn, forecast. This forecast in conjunction with a forecast of the shape of a continuous load duration curve yields the requisite results. The approach is implemented and the results are acceptable.     

Quasi-Phase Equilibrium Prediction of Multi-Element Alloys Based on Machine Learning and Deep Learning

In this study, a phase field model is established to simulate the microstructure formation during the solidification of dendrites by taking the Al-Cu-Mg ternary alloy as an example, and machine learning and deep learning methods are combined with the Kim-Kim-Suzuki (KKS) phase field model to predict the quasi-phase equilibrium. The paper first uses the least squares method to obtain the required data and then applies eight machine learning methods and five deep learning methods to train the quasi-phase equilibrium prediction models. After obtaining different models, this paper compares the reliability of the established models by using the test data and uses two evaluation criteria to analyze the performance of these models. This work find that the performance of the established deep learning models is generally better than that of the machine learning models, and the Multilayer Perceptron (MLP) based quasi-phase equilibrium prediction model achieves the best performance. Meanwhile the Convolutional Neural Network (CNN) based model also achieves competitive results. The experimental results show that the model proposed in this paper can predict the quasi-phase equilibrium of the KKS phase-field model accurately, which proves that it is feasible to combine machine learning and deep learning methods with phase-field model simulation.     

Robust Prediction of the Bandwidth of Metamaterial Antenna Using Deep Learning

The design of microstrip antennas is a complex and time-consuming process, especially the step of searching for the best design parameters. Meanwhile, the performance of microstrip antennas can be improved using metamaterial, which results in a new class of antennas called metamaterial antenna. Several parameters affect the radiation loss and quality factor of this class of antennas, such as the antenna size. Recently, the optimal values of the design parameters of metamaterial antennas can be predicted using machine learning, which presents a better alternative to simulation tools and trial-and-error processes. However, the prediction accuracy depends heavily on the quality of the machine learning model. In this paper, and benefiting from the current advances in deep learning, we propose a deep network architecture to predict the bandwidth of metamaterial antenna. Experimental results show that the proposed deep network could accurately predict the optimal values of the antenna bandwidth with a tiny value of mean-square error (MSE). In addition, the proposed model is compared with current competing approaches that are based on support vector machines, multi-layer perceptron, K-nearest neighbors, and ensemble models. The results show that the proposed model is better than the other approaches and can predict antenna bandwidth more accurately.     

井下移动式破碎机垂直方向随机振动响应的研究简

 为了研究井下移动式破碎机在破碎头受随机载荷作用下垂直方向的振动特性、探究其工作可靠性低的原因,通过假设与简化,根据多体动力学理论,建立了其动力学模型和垂直方向的运动方程,运用虚拟激励法推导了破碎机垂直向振动响应的数学模型．以国产某型井下移动式破碎机为研究对象,分析确定了破碎头冲击岩石时的随机载荷,运用MATLAB语言程序,得到该破碎机在高位时受随机载荷作用下破碎头、支臂与机身在垂直方向的随机振动响应．结果表明：在载荷峰值为375 kN、均值为186 kN、频率为10 Hz的随机激励作用下,该机各部振幅峰值与均值分别为：破碎头：0.0125 m,0.007 25 m;支臂：0.010 m,0.006 57 m;机身：0.006 1 m,0.003 62 m,可见各部振动比较剧烈．破碎头振动和波动最大,支臂次之,机身最小,与实际情况相符．当各部件振幅达最大时,破碎头、支臂与机身的低频与固有频率接近,容易发生共振,这是导致机器振动剧烈,引起零部件、元器件失效,使用寿命下降的主要原因．所得结论为改进移动式破碎机设计、采取合理的减振措施、延长破碎机使用寿命提供参考．     

负载预测更新下的实时电价Stackelberg博弈模型

智能电网环境下,在实时电价Stackelberg博弈模型的基础上引入负载预测,以匹配实时负载和预测负载为目标,设计售电商与用户之间的主从博弈模型以及负载预测更新下的实时定价机制,得到双方的最优实时电力价格和最优用电行为。通过将当日实时电价机制均衡状态下的负载时间序列融入电力供应商电力价格权重时间序列向量,得到进一步优化的日前定价实时电价下的均衡负载时间序列,构成整体不断推进不断优化的闭环。同时,给出实时负载与预测负载序列的匹配程度评价指标与判断标准。通过数值仿真分析,在与未优化的实时定价机制对比以后,发现所提出的负载预测更新下的实时定价机制能够在提高电网运行效率的同时显著降低电力用户用电成本。     

风、光、油互补控制器的设计

本文设计开发的一款基于风、光、油互补的充电控制器,可以实现风能和光能转化为蓄电池电能,在无风、无光照的条件下,且蓄电池的能量不足,可以启动油机进行发电,并对蓄电池进行充电。从而实现对可再生能源的充分利用,在保障设备供电的前提下,最大限度的节省燃油。     

应用神经网络进行短期负荷预测

以某地区购网有功功率的负荷数据为背景,建立了３个BP神经网络负荷预测模型———SDBP,LMBP 及BRBP模型进行短期负荷预测工作,并对其结果进行比较。针对传统的BP算法具有训练速度慢,易陷入局部 最小点的缺点,采用具有较快收敛速度及稳定性的L唱M（Levenberg唱Marquardt）优化算法进行预测,使平均相对误 差有了很大改善,而采用贝叶斯正则化算法可以解决网络过度拟合,提高网络的推广能力。     

Sequential forecast of incident duration using Artificial Neural Network models

This study creates an adaptive procedure for sequential forecasting of incident duration. This adaptive procedure includes two adaptive Artificial Neural Network-based models as well as the data fusion techniques to forecast incident duration. Model A is used to forecast the duration time at the time of incident notification, while Model B provides multi-period updates of duration time after the incident notification. These two models together provide a sequential forecast of incident duration from the point of incident notification to the incident road clearance. Model inputs include incident characteristics, traffic data, time gap, space gap, and geometric characteristics. The model performance of mean absolute percentage error for forecasted incident duration at each time point of forecast are mostly under 40%, which indicates that the proposed models have a reasonable forecast ability. With these two models, the estimated duration time can be provided by plugging in relevant traffic data as soon as an incident is reported. Thereby travelers and traffic management units can better understand the impact of the existing incident. Based on the model effect assessments, this study shows that the proposed models are feasible in the Intelligent Transportation Systems (ITS) context.     

Dynamic loading of overhead lines by adaptive learning techniques and distributed temperature sensing

The need for dynamic loading of overhead lines requires reliable assessment models that should be able to predict both the evolution of the hot-spot temperature and the associated maximum allowed duration, at any load level and on the basis of actual conductor thermal state and forecasted environmental conditions. In order to address this problem, a novel identification semi-physical modelling architecture that combines knowledge coming from expertise with empirical evidence provided by observations is proposed. This is performed by integrating an analytical thermal model, which estimates qualitatively the conductor hot-spot temperature, and an adaptive corrective algorithm, based on a local learning theory and aimed at enhancing the estimation accuracy. The corrective algorithm is continuously adjusted by field data acquired through distributed fibre-optic sensor based on stimulated Brillouin scattering. To assess the performances of the proposed methodology, the main results of experimental studies obtained on a laboratory overhead line are presented and discussed.     

风电场有功功率控制综述

由于风电具有随机性、波动性和反调峰特性,高比例的风电并入电网会对电力系统的稳定性和安全性造成很大的冲击,因此有必要对风电场有功功率输出进行控制,减少风电功率的波动性,提高输出功率的平滑性;同时,随着装机容量的不断增加,造成大量的弃风现象,风电场的控制模式由传统的最大功率点跟踪（maximum power point tracking,MPPT）模式向限功率控制模式转变。基于弃风限电以及风电并网的控制要求这2个背景,分析了控制风电场有功功率的必要性,从单机功率控制和场站级功率控制2个层面出发,以场站级功率控制为侧重点,归纳总结了当前风电有功功率控制的研究现状,总结出不同控制方式的控制特点及不足之处,并对其研究方向进行了展望。     

Load behaviour prediction under blackout conditions using a fuzzy expert system

This paper proposes a method to forecast load behaviour during restoration of power systems after total or partial blackouts. Load behaviour after blackouts can be anomalous and differ significantly from that during normal operation conditions. A heuristic top-down approach was adopted to develop the forecasting method, using an expert system based on linguistic variables and fuzzy logic rules as its central core. The approach addresses the main aspects of load behaviour during re-energisation and is capable of representing physical phenomena. The proposed methodology was tested with data from real electrical power substations and these test results are presented