The excessive integration of photovoltaic (PV) in distribution network could result in technical problem, such as overvoltage problem, thermal overloading problem, etc. Consequently, how to evaluate the PV hosting capacity has emerged as a critical technology for the expansion planning of future distribution network. However, assessing the PV hosting capacity usually requires solving a non-convex optimization problem, although several relaxation methods have already been developed to relax the model, the exactness of these methods still remain a concern. To address this problem, a successive second-order cone programming (SOCP) approach with additional linearized error reduction constraints is proposed to reduce the relaxation gap introduced by the use of the SOCP relaxation in the PV hosting capacity problem. Moreover, a penalty item of the relaxation gap is added into the objective function to accelerate the convergence process. Simulations are conducted on an improved IEEE 33-bus system to validate the effectiveness of the successive SOCP approach.

In addressing the issue of exponential increase in solution time with the growing number of simulations when using mixed-integer linear programming methods to solve stochastic models based on Monte Carlo simulation, this paper proposes employing spectral clustering to determine the best-quality classification scheme. By using the spectral clustering method on random samples generated by Monte Carlo simulation, the optimal classification scheme is identified to group similar stochastic scenarios together. Finally, by selecting the center of each class as a representative, the number of scenarios to be considered is reduced, facilitating the solution of problems such as whole-scenario modeling and lowering the difficulty of solving. Experimental results demonstrate that the spectral clustering method proposed in this paper exhibits the best performance among some commonly used methods.

To explore the optimization method of grid-connected voltage support technology in new energy stations, this study first analyzes and discusses this technology. Second, this study describes the deep learning model architecture and feature selection in detail and determines the framework used for the optimization model proposed here. Lastly, the development of optimization and control strategies is investigated, and the optimized model’s effectiveness is verified through experiments. The results reveal that the optimized model's accuracy, precision, recall, and F1 score are higher than those of the comparison model in the performance comparison experiment, reaching the highest values of 0.890, 0.888, 0.878, and 0.883, respectively. This reflects that the optimized model shows high performance on small datasets, and its performance benefits become more pronounced as the data volume increases. This feature is particularly significant because, in practical applications, power systems often need to process large amounts of data to achieve efficient voltage support. In simulation experiments, the optimized model demonstrates excellent performance in terms of response time, stability, robustness, and energy consumption. Moreover, this model effectively addresses various data challenges and uncertainties encountered in grid-connected voltage support technology for power systems, thereby providing robust support for stable and efficient voltage regulation. In light of the findings, this study offers substantial insights for advancing research in the realms of power systems and new energy technologies. The exploration into the application of deep learning and intelligent control strategies within power systems reveals significant potential for transforming grid optimization practices. This study accentuates how data-driven methodologies can revolutionize energy management, paving the way for smarter and more efficient energy systems. By enhancing both the responsiveness and operational efficiency of power grids, the study contributes to the acceleration of digital transformation within the energy sector, fostering innovation and laying a robust foundation for future advancements in energy informatics.

At present, the scale of the integrated energy system is increasing gradually, but the difficulty and time of solving the problem increase geometrically with the increase of the traditional solving method. This brings some challenges to the calculation methods within the integrated energy system (IES). Therefore, in this paper, IES containing electricity, gas, heat and cold and its internal equipment are modeled in the form of energy bus. To solve the difficult problem of large-scale optimal scheduling model, an alternate direction multiplier method (ADMM) using consensus variables to align the boundary conditions is proposed to decompose the problem and solve it distributed. In addition, it is difficult to balance the original residual and dual residual of the original ADMM algorithm, and the phenomenon of oscillation occurs easily in the iteration process. The fixed point iteration form of ADMM is derived, and Anderson accelerated alternate direction multiplier method (AAADMM) is proposed. The feasibility and economy of distributed solution for integrated energy system are demonstrated by numerical examples, and the convergence and computational efficiency of the proposed AAADMM algorithm are verified.

Abstract This article deeply analyses the model and calculation method of distributed energy in reactive power generation and absorption operation modes. It proposes an optimized control model and optimization method based on the segmented power factor of the distribution network voltage level. Simulation calculations were conducted. The results showed the effectiveness, economy, and balance of the reactive power optimization model and algorithm considering voltage level and power factor segmentation.

This study focuses on the optimization of wind-solar storage capacity allocation in intelligent microgrid systems using the Particle Swarm Optimization (PSO) algorithm. The combination of distributed generation and smart grid technology in microgrids demonstrates unique advantages in promoting the utilization of renewable energy and enhancing the intelligence of energy systems. However, the volatility and uncertainty of wind-solar storage systems pose a challenge in capacity allocation. By constructing precise mathematical models for wind and photovoltaic power generation and storage devices, and integrating the particle swarm algorithm for optimization, this paper aims to seek a feasible solution to reduce energy waste and improve the operational efficiency of microgrids. The research results demonstrate the effectiveness of the particle swarm algorithm in determining the optimal capacity allocation, providing a scientific basis for the operational management of practical microgrids.

Currently, China's demand for energy is increasing. In order to reduce China's dependence on traditional mineral energy, some achievements have been made in increasing research and development efforts in new energy. With the continuous increase in the scale of new energy grid connection, the proportion of new energy grid connection is constantly increasing, and problems such as grid stability and power quality degradation are becoming increasingly prominent, which directly affects the widespread application of new energy generation technology. This article explored the research on a remote control system for new energy grid connected power generation based on artificial intelligence. Taking the island detection of photovoltaic grid connected inverters based on Adaboost algorithm as an example, the feasibility of the technology was verified through experiments. The experimental data showed that the accuracy, recall, and F0.5 of island effect prediction based on Adaboost algorithm were 0.98, 0.9523, and 0.7811, respectively, which were higher than other algorithms.

LLC resonant converters have a wide range of applications in industry. However, so far, there is no simple and accurate small signal equivalent circuit model available. This article proposes an equivalent circuit model for LLC resonant converters based on the modified and extended description function method. This model can effectively predict the characteristics of LLC resonant converters at switching frequencies below, close to, or above the resonant frequency. Finally, the effectiveness of the model was verified through simulation. As shown in the research results: the simple third-order equivalent circuit model for LLC resonant converters can effectively predict the dynamic characteristics of the beat frequency and better understand the small signal model and all transfer functions of the model can help engineers design control feedback correctly.

一二次设备实际产生故障时,剩余输出电压会产生波动负载,导致识别时的比差数值较小,功率反送能力较差,针对该问题,设计多信息源的一二次设备故障智能化识别方法.获取一二次设备的多信息源,处理输出电压数值为纵分量,控制一二次设备产生的波动负载,设计故障数据计算模型,联合受限玻尔兹曼机,在一二次设备与故障数据之间构建数值关系,采用二次微分处理数值支持向量机中的超参数,完成一二次设备故障智能化识别.实例测试结果表明,所设计的故障识别方法比差数值最大,功率反送能力最佳.

为了实现主网和配网之间线路运行状态以及相关设备运行状态的协同决策、协同监管,研究基于数据挖掘的主配网一体化协同监视方法.将采集的数据进行格式统一化处理,传送至数据管理层,数据管理层完成数据清洗及校验后,分类存储;协同监视层通过主配网历史数据模型、实时估计以及态势预判感知存储数据,采用随机森林算法构建协同决策规则库,将感知结果与协同决策规则库进行匹配,获取故障控制决策.测试结果表明,该方法的感知性能良好,可快速完成故障和控制策略的匹配,满足主配网一体化协同监视需求.

当前,电气设备异常检测方法主要是单独提取特征点,利用人工规则库进行异常检测,无法挖掘异常特征之间的关联度,导致电气设备运行异常感知准确率低.对此,提出一种基于机器视觉和动态阈值的电气设备运行异常关联感知方法.利用机器视觉中的小波滤波器描述电气设备运行异常兴趣点,并在坐标系中确定基准方向,借助描述符,将兴趣点转换为可识别的参数;优化感知设备结构,增加可感知的状态信息;设计基于动态阈值的电气设备运行异常关联感知流程,完成异常关联感知方法设计.实例测试结果表明,该方法的异常感知准确率、精度、召回率、假阳性均在80％以上,提高了电气设备运行异常感知准确率.

传统电网调度告警算法告警指令运行效率和电网调度故障爆出率较低,告警误报率较高,因此提出基于知识图谱的电网调度告警关联规则与决策树算法.采用关联规则算法优化知识图谱电网数据层,筛选不符合规则的电网调度数据;根据调度告警指令特征,提取电网调度告警事务;基于知识图谱数据层建立决策树矩阵,设计决策树关联规则优化算法,构建电网调度告警模型,确定电网调度故障点位置,实现电网调度的告警.实验结果表明,基于知识图谱的电网调度告警算法的告警指令运行效率为9.55×109条/min,最大告警误报率仅为13.5％,电网调度故障爆出率达到98.06％,告警性能较高.

With the development of new energy sources, the operation of power system needs controllable power loads to balance the fluctuation of the sources. This paper focuses on temperature sensitive load, which has gained many attentions in recent years. Specifically, a method to extract and analyze the summer temperature-sensitive loads of medium voltage classified users is proposed, which uses K-means clustering method to obtain MV classified users and correlation coefficient method to analyze temperature sensitivities of the classified users at different times of the year. The method is applied to field data of a whole year, and the temperature-sensitive loads of MV classified users in July and August are extracted. Their characteristics, including the share of temperature-sensitive load, the maximum daily temperature-sensitive load growth rate and the peak distribution of daily temperature-sensitive load, are analyzed, which are helpful for operation and control of power system.

This paper proposes a rule base construction method of section out of limit disposal strategy for thermal power unit equipment fault, which can solve the problems of abnormal power transmission and grid real-time scheduling caused by section out of limit. Through the improved Apriori algorithm, the section out of limit disposal strategy of thermal power unit equipment fault is designed. In the mining result of thermal power unit equipment fault section out of limit, the association rules between section state and unit and load disposal strategy are established. By using the improved Apriori algorithm, the policy data and section state data are set as the preceding and following constraints of the association rules, The strong association rules are obtained, and the rule base of cross section out of limit disposal strategy is constructed according to the association rules. The experimental results show that: the number of association rules obtained by the improved Apriori algorithm mining each database is less than 1000, and the rule effectiveness and analysis efficiency are significantly improved; The rule base built by this method can effectively deal with the out of limit problems of different sections and units, and the sensitivity of each generator node increases greatly after the treatment.

现有负荷曲线聚类的研究主要基于单日负荷曲线或多日同时刻的负荷分布开展,忽略了负荷在多日间的波动特性和负荷曲线的时间滞后特性,导致聚类结果的准确度和鲁棒性不足.综合考虑负荷的波动特性和时间滞后特性,提出一种快速动态时间弯曲和最小覆盖球相结合的多 日负荷曲线聚类方法.在考虑负荷时间滞后特性的基础上,利用快速动态时间弯曲和多维尺度缩放对负荷曲线进行降维;为每个降维负荷迭代寻找最小覆盖球,并计算不同覆盖球的球间相似度;利用谱聚类算法得到相似度矩阵.算例结果表明,所提方法在准确度和鲁棒性上较传统方法有一定优势.

随着电网规模的不断扩大,电力调度业务规模也在不断增长,为电力调度带来了挑战.为保证电力系统的稳定运行,必须提高电力调度系统的自动化程度.为此,针对配电网故障识别难题,基于小波变换理论和人工神经网络提出了一套故障识别方法;针对传统电力调度系统技术灵活性和智能性不足的问题,设计了一套操作票智能生成系统,该系统在分析电力系统特点的基础上,对构成操作票专家系统的知识库和推理机进行了分析和设计.最后通过案例分析验证了所提出的故障识别方法和操作票生成系统.研究结果表明所提出的故障识别方法可有效识别故障的发生、故障类型以及故障位置,采用Rete算法设计的推理机能够有效针对故障自动生成操作票.

To satisfy the requirements of analyzing and managing large-scale integrated transmission and distribution networks, an integrated optimal power flow algorithm, based on integrated power flow algorithm, is proposed in this paper. Considering the seriously ill-conditioned Jacobian matrix and poor convergence, self-adaptive Levenberg-Marquard method and incomplete LU decomposition method are used to optimize the integrated power flow algorithm. In addition, a modified DC optimal power flow model is built, whose constraints are linear. A self-adjusting interior point method is proposed to solve the model based on branch losses obtained by the integrated power flow algorithm in each iteration. Numerical experiments illustrate that the proposed algorithms have high efficiency. In addition, they are applicable to many issues, such as distribution networks with loops, distribution networks with distributed generations, seriously ill-conditioned systems, multi-distribution-network areas, etc.

Considering the separate management of transmission networks (TNs) and distribution networks (DNs), this article proposes a chance-constrained optimal power flow (CCOPF) formulation for integrated transmission and distribution (I-T&D) networks and its solution algorithm with limited information interaction. The uncertainties of loads and renewable generations are considered in the formulation, which guarantees that generations, power flows, and voltage magnitudes in both TNs and DNs remain within their bounds with a predefined probability. A double-iterative solution algorithm is proposed to solve CCOPF, of which the inner-iteration is applied to solve a deterministic OPF of I-T&D networks while the outer-iteration is to repeatedly update uncertain margins around a forecasted solution to converge finally. Particularly, the heterogeneous decomposition algorithm is applied to solve the deterministic OPF, and an I-T&D-power-flow-based two-point estimation method is proposed to calculate uncertainty margins. The overall solution algorithm is realized based on boundary information exchange between TNs and DNs, where the data and model privacy of TNs and DNs are well-preserved. Numerical experiments demonstrate the accuracy and efficiency of the proposed solution algorithm.

Renewable energies (RES) will receive higher attention in the future, and continuously increase the penetration rate in the power system to achieve low carbon goal. However, the volatility and intermittent nature of RES generation caused by the uncertainty of the weather have brought huge challenges to the safe and stable operation of the power system. IACs have become a substitute for traditional power plants to accommodate RES by changing their operating power. This paper proposes a control method based on the MA filter to smooth photovoltaic generation power fluctuations utilizing inverter air conditioners (IACs). The photovoltaic power data is filtered by the MA filter in real time firstly. An output power with little fluctuations and one power deviation with high frequency fluctuations are obtained. The power deviation with high frequency fluctuations will be smoothed by IACs, while the remaining power deviation will be smoothed by ESSs. The case studies show a smooth power output is obtained for the power system after filtering.

Power grid dispatching decision is a process of making full use of the data of each system in the power grid and comprehensively analyzing the power grid operation state. However, in the current grid dispatching system, the data needed for dispatching decision is stored in different systems, which has the characteristics of decentralized, multi-source and heterogeneous. Existing data architectures are difficult to break down the data barriers between different systems, so they cannot efficiently retrieve the required dispatching data. To meet the needs of integration and aggregation of power grid multi-source heterogeneous dispatching messages, a big data platform for power grid multi-source dispatching messages based on improved ontology aggregation is proposed. The platform is based on Map/Reduce architecture, and the improved ontology clustering method is used to enhance the query and reasoning ability of the platform scheduling data, so as to realize the efficient integration and aggregation of multi-source messages in power grid dispatching.