学术论文

Practicality-enhanced behind-the-meter PV power generation disaggregation based on synchronization and transferability fused LSTM framework
Energy and AI

Chengye Zhang; Huan Long*; Zijun Zhang; Jinde Cao

To facilitate the operation of distribution networks with a large scale of household photovoltaic systems integrated, the availability of community-level behind-the-meter (BTM) PV power generation is crucial. Yet, due to the scarcity of smart meters installed, it is challenging to obtain such information via directly aggregating measured power outputs of individual PV systems, and an effective estimation method needs to be developed. Considering the similarity between household-level and community-level data within the same geographical area, this paper develops a synchronization and model-transfer fused LSTM framework (SAM-LSTM). The core technical contribution lies in the development of the Synchronized Long Short-Term Memory (Syn-LSTM), which separately models the synchronized factors and disaggregated BTM data to capture more generalized representations. The learned household-level representations are then transferred to the community-level. Finally, by explicitly leveraging the complementarity between PV generation and consumption, a dual time-series modeling architecture is developed to refine the initial community-level PV power generation estimates, thereby alleviating potential biases introduced during the model-transfer process. Extensive computational studies are conducted to demonstrate the effectiveness of SAM-LSTM in community-level BTM PV power generation disaggregation in real data from Hebei, China. Compared with the best-performing benchmarks, SAM-LSTM achieves up to 56% lower MSE, significantly demonstrating its strong generalization and robustness capabilities.

LFTL: Lightweight feature transfer learning with channel-independent LSTM for distributed PV forecasting
Energy and AI

Yuanjing Zhuo; Huan Long*; Zhi Wu; Wei Gu

Distributed photovoltaic (PV) power forecasting in newly installed systems faces challenges due to inherent stochastic volatilities and limited historical data. This paper proposes a lightweight feature transfer learning (LFTL) method that enables rapid and accurate forecasting of new distributed PVs. Firstly, the raw fluctuating PV data are preprocessed through decomposition to separate low- and high-frequency components. These components are then multi-scale segmented to capture diverse temporal characteristics. Following feature compression and LSTM temporal modeling, the informative features from the source domain enable lightweight transfer. For the target domain, a channel-independent encoder is designed to prevent negative interactions between heterogeneous frequencies. The frequency-fused segment-independent decoder equipped with positional embeddings enables local temporal analysis and reduces error accumulation of multi-step forecasts. LFTL trains with a joint training strategy to avoid negative transfer caused by domain disparity. LFTL consistently outperforms state-of-the-art time-series forecast models while maintaining a relatively low computational overhead based on real-world distributed PV data.

基于自监督学习的配电网分布式最优潮流求解
中国电机工程学报/Proceedings of the CSEE

龙寰; 蔡辉煌; 张晓; 吴志; 顾伟

随着网架结构的复杂化，配电网分区自治已成为提升调度效率、可靠性与经济性的必要手段。然而，传统分布式凸优化算法逐渐难以适配新型配电网的高效运行需求。在此背景下，机器学习在配电网调度中得到越来越广泛的关注，但大量标注样本的需求限制了机器学习的实际应用。针对这一限制，本文提出一种融合交替方向乘子法(Alternating Direction Method of Multipliers， ADMM)的自监督学习算法，用于求解配电网分布式最优潮流。首先，建立了融合ADMM的自监督学习框架，通过两个相互独立的网络模拟ADMM算法中全局变量与拉格朗日乘子交互迭代，实现网络间的相互监督，原始网络用于估计全局变量，对偶网络用于估计拉格朗日乘子。其次，考虑分布式最优潮流全局变量与边界变量的一致性，设计了考虑一致性约束的损失函数，用于引导原始网络的训练，保证训练过程的收敛性。再次，基于卡罗需-库恩-塔克条件分析证明所提出方法的收敛性，并给出残差的上界。最后，在改进的IEEE 123节点系统及苏州实际276节点配电系统验证了所提出方法的有效性。

A Learning to Optimize Approach to Accelerating Distributed Optimal Power Flow Solving
Journal of Modern Power Systems and Clean Energy

Huihuang Cai; Huan Long; Zhi Wu; Wei Gu; Jingtao Zhao

As power systems continue to scale, a fast and accurate distributed optimal power flow solver becomes crucial for effective power system dispatch. This work proposes a Learning to Optimize (L2O) approach to accelerate the distributed optimal power flow solving. The final convergence values of global variables and Lagrange multipliers of the Alternating Direction Method of Multipliers (ADMM) are estimated, which serve as its warm-start solution. A Long Short-Term Memory-Variational Auto-Encoder (LSTM-VAE) model is developed as the core model to estimate the convergence value. The LSTM part processes high-dimensional temporal characteristics of global variables and Lagrange multipliers, extracting their latent temporal patterns to generate low-dimensional representations. Subsequently, VAE decoder part reconstructs these compressed latent vectors back to the high-dimensional asymptotic convergence values of ADMM variables. A novel loss function is designed in the form of a quadratic sum penalty term to incorporate the constraint violations of the Lagrange multipliers. Additionally, a two-stage training data generation strategy is proposed to efficiently generate substantial data in a limited amount of time. The effectiveness of the proposed approach is evaluated using the modified IEEE 123-bus system, a synthetic 500-bus system, and a 793-bus system.

基于多标签分类与卷积神经网络的配电网拓扑辨识
高电压技术/High Voltage Engineering

龙寰; 石子晴; 赵景涛; 郑舒; 张晓燕; 谢文强

为适应新一代配电网运行特性，配电网开关需频繁动作调整网络结构，难以及时、准确获取配电网的实时拓扑结构，给配电网的态势感知带来一定困难。鉴于传统以状态估计为框架的配电网拓扑识别方法计算复杂度高、在线应用困难，同时大规模配电网拓扑结构多样化，该文提出了基于多标签分类与卷积神经网络的配电网拓扑辨识方法。通过配电网量测电压数据与开关状态间的多映射关系，引入多标签分类机制，对配电网拓扑结构进行多标签编码，将配电网开关与拓扑辨识模型输出进行物理映射，利用卷积神经网络搭建多标签分类器，实现拓扑的准确辨识。基于改进的IEEE 123节点配电网算例对所提方法进行验证，实验结果表明：所提模型具有较高的拓扑识别准确率，且对于在训练样本空间外的未知拓扑结构，其具备更好的推理能力，更适用于实际拓扑识别的场景，证实了所提方法的优越性和鲁棒性。

Wind turbine condition monitoring based on SCADA data-image conversion
IEEE Transactions on Instrumentation and Measurement

Huan Long; Shaohui Xu; Huihuang Cai; Wei Gu

This article investigates a data-image conversion-based condition monitoring algorithm for wind turbines (WTs) using supervisory control and data acquisition (SCADA) data. The traditional condition monitoring problem is converted into an image classification problem in the proposed method. It consists of three parts, feature selection, data-image conversion, and condition monitoring based on image classification. The important features are selected from SCADA data by gradient boosting decision tree (GBDT) and permutation importance. Through the data-image conversion method, the selected numerical features are converted into heatmap images and combined into RGB images. AlexNet is introduced to classify the generated images to detect the operation state of WT. Data augmentation, inspired by symbolic augmentation, is developed to expand the number of fault data images to solve the overfitting caused by uneven data during training. The effectiveness of the proposed condition monitoring method is validated on the dataset collected from Chinese wind farms. The comparison result shows the proposed image-based condition monitoring method has achieved significant improvements.

Short-term load interval prediction with unilateral adaptive update strategy and simplified biased convex cost function
IET Generation, Transmission & Distribution

Shu Zheng; Huan Long; Zhi Wu; Wei Gu; Jingtao Zhao; Runhao Geng

This article proposes a unilateral Adaptive update strategy based Interval Prediction (AIP) model for short-term load prediction, which is developed based on lower and upper bound estimation (LUBE) architecture. In traditional LUBE interval prediction model, the model training is usually trained by heuristic algorithms. In this article, the model training is formulated as a bi-level optimization problem with the help of proposed unilateral adaptive update strategy and cost function. In lower-level problem, a simplified biased convex cost function is developed to supervise the learning direction of basic prediction engines. The basic prediction engine utilizes Gated Recurrent Unit (GRU) to extract features and Full connected Neural Network (FNN) to generate interval boundary. In upper-level problem, a unilateral adaptive update strategy with unilateral coverage rate is put forward. It iteratively tunes hyper-parameters of cost function during training process. Comprehensive experiments based on residential load data are implemented and the proposed interval prediction model outperforms the tested state-of-the-art algorithms, achieving a 15% reduction in prediction error and a 20% decrease in computational time.

A charge-discharge optimization strategy considering the spatiotemporal distribution of electric vehicles and the operational safety of the power distribution network in the power-transportation coupling network
Journal of Cleaner Production

Huan Long; Zhengyang Guo; Chengang Zhou

An electric vehicle (EV) charge-discharge optimization (CDO) strategy that accommodates both grid-side and user-side demands is conducive to mitigating the adverse effects of disordered EV charging on the power distribution network (PDN). To tackle the issue of inaccurate estimation of the schedulable capacity of EVs in existing research, a high temporal resolution dynamic spatiotemporal distribution simulation model for EVs is developed. Furthermore, leveraging the characteristics of the power-transportation coupling network (PTCN), a sub-districted dynamic electricity pricing (DEP) is proposed to assist in improving the PDN node voltage distribution. Subsequently, an incentive coefficient is introduced to incentivize the discharge of EVs during potential peak periods. The proposed CDO strategy, considering the non-cooperative behavior of EV owners (EVOs), can provide personalized charge-discharge plans for them. The temporal sequence experiments reveal a significant decrease in the peak-to-valley disparity of the load, achieving an optimization rate of load mean square error that exceeds 80%. The sub-districted DEP exhibits a significant advantage in maintaining PDN voltage stability compared to traditional time-of-use electricity pricing (TOUEP) and DEP, while also achieving reductions in active power losses of more than 15% and 10% for PDN, respectively. Moreover, in comparison to the conventional scenario, the CDO strategy leads to a maximum reduction of up to 942.5% and 22.3% in the total economic cost for electric private cars and electric taxis respectively. Lastly, the impact of seasonal factors is discussed. Numerical results indicate effective alleviation of load peak-valley differences, load fluctuations, and voltage drop phenomena in both winter and summer. Additionally, the maximum reduction in active power losses reaches 20.1% and 21.9%, respectively, while the total economic cost for EVOs participating in the CDO strategy is reduced by up to 97.7% and 114.7%.

Optimal Operation Control Strategies for Active Distribution Networks Under Multiple States: A Systematic Review
Journal of Modern Power Systems and Clean Energy

Jingtao Zhao; Zhi Wu; Huan Long; Huapeng Sun; Xi Wu; Chingchuen Chan

With the large-scale integration of distributed renewable generation (DRG) and increasing proportion of power electronic equipment, the traditional power distribution network (DN) is evolving into an active distribution network (ADN). The operation state of an ADN, which is equipped with DRGs, could rapidly change among multiple states, which include steady, alert, and fault states. It is essential to manage large-scale DRG and enable the safe and economic operation of ADNs. In this paper, the current operation control strategies of ADNs under multiple states are reviewed with the interpretation of each state and the transition among the three aforementioned states. The multi-state identification indicators and identification methods are summarized in detail. The multi-state regulation capacity quantification methods are analyzed considering controllable resources, quantification indicators, and quantification methods. A detailed survey of optimal operation control strategies, including multiple state operations, is presented, and key problems and outlooks for the expansion of ADN are discussed.

基于数据驱动的风电机组状态监测与故障诊断技术综述
电力系统自动化/Automation of Electric Power Systems

龙寰;杨婷;徐劭辉;顾伟

随着大规模风电场的建设，风电机组的状态监测和故障诊断成为一个重要的研究课题。早期的风电机组状态监测和故障诊断依靠人工巡检，而随着风电机组装机容量的不断增长，人工巡检的成本和难度也随之增加。近年来，基于数据驱动方法的风电机组状态监测和故障诊断逐渐成为热点。文中从运行数据类型出发，对相关研究内容进行综述。首先，针对风电机组数据采集与监控（SCADA）系统，从监测对象角度出发，剖析基于SCADA数据的状态监测与故障诊断方法的研究现状；其次，针对风电机组组件振动数据，分析对比各类振动故障特征提取方法的优点和局限性；然后，针对新兴基于图像数据或数据-图像转换数据的状态监测与故障诊断方法，从单一图像诊断和数据-图像转换评估两方面对现有研究进行论述与总结；最后，对未来状态监测和故障诊断的研究方向进行了展望。

考虑反弹效应的空调负荷可调潜力评估与控制策略研究
电网技术/Power System Technology

龙寰;赵烁;吴志;陈鼎;刘维亮

当前需求侧响应研究以需求侧响应发生时段为研究重点，很少关注需求侧响应后反弹效应对电网的冲击。文章提出了一种考虑反弹效应的空调负荷可调潜力评估与控制策略。首先，从统计学角度出发，选取空调负荷运行功率和室外温度两个特征，基于用户历史数据评估空调负荷的可调潜力区间。然后，以负荷历史功率在不同温度出现的频率衡量负荷可削减的难易程度，建立用户激励成本模型，并以负荷聚合商激励成本与通信成本之和最小为目标函数，建立用户挑选模型。最后，为减轻用电反弹现象对于电网的冲击，轮换选取不同用户参与需求侧响应，提出了一种考虑反弹效应的时序循环控制策略。通过美国Pecan数据集分析不同削减量下全员参与策略、增加延迟单元策略与所提的时序循环控制策略的平抑反弹效应效果，验证了所提策略的有效性。

Small-Sample Solar Power Interval Prediction Based on Instance-Based Transfer Learning
IEEE Transactions on Industry Applications

Huan Long, Runhao Geng, Shengxing Wan, Hui Hui, Rui Li, Wei Gu

In the context of high photovoltaic (PV) penetration, high-quality solar power interval prediction is important for grid system operation. However, in some cases, sufficient amount of data are not available to train a reliable prediction model, especially for new installed PV stations. To tackle this problem, this article proposes a novel small-sample interval prediction model with improved TrAdaBoost (SIPTAB) method for solar power prediction. Sample weight is designed to select the important samples from other data sources. First, Extreme Learning Machine (ELM) with Direct Quantile Regression (DQR) is employed as the base predictor to construct interval boundaries. Second, an improved TrAdaBoost algorithm is proposed to iteratively construct a boosting ensemble interval predictor to enhance the prediction performance with limited amount of data. Third, a two-stage model training strategy is introduced in the architecture to optimize the boosting ensemble interval predictor and further improve prediction quality. Comprehensive experiments based on realistic solar power data are conducted to confirm the superiority of proposed model.

Safety-aware Semi-end-to-end Coordinated Decision Model for Voltage Regulation in Active Distribution Network
IEEE Transactions on Smart Grid

Linwei Sang; Yinliang Xu; Huan Long; Wenchuan Wu

Prediction plays a vital role in the active distribution network voltage regulation under the high penetration of photovoltaics. Current prediction models aim at minimizing individual prediction errors but overlook their collective impacts on downstream decision-making. Hence, this paper proposes a safety-aware semi-end-to-end coordinated decision model to bridge the gap from the downstream voltage regulation to the upstream multiple prediction models in a coordinated differential way. The semi-end-to-end model maps the input features to the optimal var decisions via prediction, decision-making, and decision-evaluating layers. It leverages the neural network and the second-order cone program (SOCP) to formulate the stochastic PV/load predictions and the var decision-making/evaluating separately. Then the var decision quality is evaluated via the weighted sum of the power loss for economy and the voltage violation penalty for safety, denoted by regulation loss. Based on the regulation loss and prediction errors, this paper proposes the hybrid loss and hybrid stochastic gradient descent algorithm to back-propagate the gradients of the hybrid loss with respect to multiple predictions for enhancing decision quality. Case studies verify the effectiveness of the proposed model with lower power loss for economy and lower voltage violation rate for safety awareness.

Electricity price prediction for energy storage system arbitrage: A decision-focused approach
IEEE Transactions on Smart Grid

Linwei Sang; Yinliang Xu; Huan Long; Qinran Hu; Hongbin Sun

Electricity price prediction plays a vital role in energy storage system (ESS) management. Current prediction models focus on reducing prediction errors but overlook their impact on downstream decision-making. So this paper proposes a decision-focused electricity price prediction approach for ESS arbitrage to bridge the gap from the downstream optimization model to the prediction model. The decision-focused approach aims at utilizing the downstream arbitrage model for training prediction models. It measures the difference between actual decisions under the predicted price and oracle decisions under the true price, i.e., decision error, by regret, transforms it into the tractable surrogate regret, and then derives the gradients to predicted price for training prediction models. Based on the prediction and decision errors, this paper proposes the hybrid loss and corresponding stochastic gradient descent learning method to learn prediction models for prediction and decision accuracy. The case study verifies that the proposed approach can efficiently bring more economic benefits and reduce decision errors by flattening the time distribution of prediction errors, compared to prediction models for only minimizing prediction errors.

An abnormal wind turbine data cleaning algorithm based on color space conversion and image feature detection
Applied Energy

Huan Long; Shaohui Xu; Wei Gu

Wind power curve (WPC) is established through data collected from the Supervisory Control and Data Acquisition (SCADA) system of each wind turbine, which can be used to analyze the operation status. However, numerous outliers are contained in SCADA data caused by wind turbine failures, shutdown maintenance or other extreme conditions to deform the wind power curve. This paper proposes a data cleaning algorithm for wind turbine abnormal data based on wind power curve image by color space conversion and image feature detection. Considering wind speed, wind power and data frequency, a three-dimensional (3D) WPC image is constructed. The scattered outliers are cleared by their statistical characteristics. The Canny edge detection and Hough transform are introduced to extract image features of stacked outliers and locate them accurately. The proposed algorithm is compared with three common outlier detection algorithms, including two data-based algorithms and an image-based algorithm. Extensive experiments conducted on the data of 22 wind turbines from two different wind farms in China indicate the efficiency, stability and reliability of the proposed algorithm.

Line Aging Assessment in Distribution Network Based on Topology Verification and Parameter Estimation
Journal of Modern Power Systems and Clean Energy

Zhi Wu; Huan Long; Chang Chen

The aging of lines has a strong impact on the economy and safety of the distribution network. This paper proposes a novel approach to conduct line aging assessment in the distribution network based on topology verification and parameter estimation. In topology verification, the set of alternative topologies is firstly generated based on the switching lines. The best-matched topology is determined by comparing the difference between the actual measurement data and calculated voltage magnitude curves among the alternative topologies. Then, a novel parameter estimation approach is proposed to estimate the actual line parameters based on the measured active power, reactive power, and voltage magnitude data. It includes two stages, i.e., the fixed-step aging parameter (FSAP) iteration, and specialized Newton-Raphson (SNR) iteration. The theoretical line parameters of the best-matched topology are taken as a warm start of FSAP, and the fitted result of FSAP is further renewed by the SNR. Based on the deviation between the renewed and theoretical line parameters, the aging severity risk level of each line is finally quantified through the risk assessment technology. Numerous experiments on the modified IEEE 33-bus and 123-bus systems demonstrate that the proposed approach can effectively conduct line aging assessment in the distribution network.

A collaborative voltage optimization utilizing flexibility of community heating systems for high PV penetration
Energy

Lu Shen; Xiaobo Dou; Huan Long; Chen Li; Kang Chen; Ji Zhou

The increasing penetration of photovoltaic generation exacerbates the risk of voltage violations in distribution networks. One viable tactic for increasing the flexibility of regional distribution networks is to use the thermal inertia of community heating systems (CHSs). However, the slow heat dynamics in the heating network brings difficulties in synchronous joint operation decisions of the integrated system. In this paper, we propose a novel power-to-temperature sensitivity model (PTSM), equating the whole CHS to a flexible electric load with thermodynamic characteristics. It denotes the direct relationship between the average water temperature change and the electric power consumed by ground-source heat pumps (GSHPs). A robust collaborative voltage optimization model is then developed from the perspective of power distribution system operators. The PTSM provides accurate prediction of CHS temperature variations to construct the operation constraints of GSHPs. This mixed integer programming problem is solved by the column-and-constraint generation algorithm. The test results of an actual residential community demonstrate that the PTSM can ensure the prediction errors of the CHS average temperature within ±0.4 °C in case of the optimal time resolution and aggregated heating load location. An IEEE 69-bus distribution network with several CHSs is studied to show that, the flexibility of CHSs can achieve an extra 17.25% reduction of the daily voltage deviation. The proposed method has significant advantages in computation time and robustness.

A combination interval prediction model based on biased convex cost function and auto-encoder in solar power prediction
IEEE Transactions on Sustainable Energy

Huan Long; Chen Zhang; Runhao Geng; Zaijun Wu; Wei Gu

Due to the intermittent and stochastic nature of solar power, solar power interval prediction is of great importance for grid management and power dispatching. A combination interval prediction model based on the lower and upper bound estimation (LUBE) is proposed to efficiently quantify the solar power prediction uncertainty. In the proposed model, the upper and lower bounds are separately predicted by two prediction engines. The extreme learning machine (ELM) is selected as the basic prediction engine. The auto-encoder technique is used to initialize the input weight matrix of ELM for efficient feature learning. A novel biased convex cost function is developed for ELM to predict the interval boundary. The output weight matrix of ELM can be solved via the convex optimization technique instead of the conventional heuristic algorithm. The proposed interval prediction model can be formulated as a bi-level optimization problem. In the lower-level problem, the lower and upper ELMs are trained under different candidate hyper-parameters of the biased cost function. In the upper-level problem, the optimal combination of the lower and upper prediction engines is determined by evaluating the interval prediction performance. Comprehensive experiments based on public data set are conducted to validate the superiority of the proposed interval prediction model.

Wind power curve data cleaning by image thresholding based on class uncertainty and shape dissimilarity
IEEE Transactions on Sustainable Energy

Guoyuan Liang; Yahao Su; Fan Chen; Huan Long; Zhe Song; Yong Gan

With the rapid development of wind farm worldwide, monitoring the status of numerous wind turbines becomes the essential work. Abnormal data in wind power curve (WPC) are quite important for wind farm operations and maintenances because they usually reveal wind turbine failures or some extreme conditions. This paper proposes a new algorithm of WPC abnormal data detection and cleaning by image thresholding based on minimization of dissimilarity-and-uncertainty-based energy (MDUE). The basic idea is to transform the scattered data into a digital image and the problem of data cleaning is turned into an image segmentation problem. For all data pixels, the confidences of being classified as normal class are computed and make up a grey level feature image. Then the optimum threshold is determined by searching through the energy space based on intensity-based class uncertainty and shape dissimilarity. Finally, the normal and three types of abnormal data are marked after applying image thresholding to the feature image. The algorithm is compared with several data-based algorithms and a recently published image-based algorithm. A large number of experiments conducted on real-world WPC data collected from 37 wind turbines in two wind farms verified the superior performance of the proposed method.

A cloud-edge cooperative dispatching method for distribution networks considering photovoltaic generation uncertainty
Journal of Modern Power Systems and Clean Energy

Lu Shen; Xiaobo Dou; Huan Long; Chen Li; Ji Zhou; Kang Chen

With the increasing penetration of renewable energy generation, uncertainty and randomness pose great challenges for optimal dispatching in distribution networks. We propose a cloud-edge cooperative dispatching (CECD) method to exploit the new opportunities offered by Internet of Things (IoT) technology. To alleviate the huge pressure on the modeling and computing of large-scale distribution system, the method deploys edge nodes in small-scale transformer areas in which robust optimization subproblem models are introduced to address the photovoltaic (PV) uncertainty. Considering the limited communication and computing capabilities of the edge nodes, the cloud center in the distribution automation system (DAS) establishes a utility grid master problem model that enforces the consistency between the solution at each edge node with the utility grid based on the alternating direction method of multipliers (AD-MM). Furthermore, the voltage constraint derived from the linear power flow equations is adopted for enhancing the operation security of the distribution network. We perform a cloud-edge system simulation of the proposed CECD method and demonstrate a dispatching application. The case study is carried out on a modified 33-node system to verify the remarkable performance of the proposed model and method.

A data-driven evolutionary algorithm for wind farm layout optimization
Energy

Huan Long; Peikun Li; Wei Gu

The wind farm layout model is to optimize the location of wind turbines to maximize the power output of the wind farm. Due to the complexity of the wind farm layout problem, the computation of objective function costs lots of time. To reduce the high computational cost while maintaining the solution performance, a data-driven evolutionary algorithm is proposed. An adaptive differential evolution algorithm (ADE) is proposed as the solver of the wind farm layout model. The adaption mechanism of ADE benefits the automatic adjustment of parameters in the mutation and crossover operators to achieve the optimal solution. The general regression neural network (GRNN) algorithm builds the data-driven surrogate model. The data-driven surrogate model is trained and updated using the data generated by the evolutionary algorithm throughout the evolution process. Through the data-driven surrogate model, the objective function is fast approximated and the bad candidate solutions are identified. The algorithm efficiency is greatly improved by fast filtering the bad candidate solutions. The ADE-GRNN is compared to other three conventional optimization methods based on two different wind scenarios. The results show the super-performance of ADE-GRNN in complex situations in terms of power output and execution time.

Cyber-attack detection strategy based on distribution system state estimation
Journal of Modern Power Systems and Clean Energy

Huan Long; Zhi Wu; Chen Fang; Wei Gu; Xinchi Wei; Huiyu Zhan

Cyber-attacks that tamper with measurement information threaten the security of state estimation for the current distribution system. This paper proposes a cyber-attack detection strategy based on distribution system state estimation (DSSE). The uncertainty of the distribution network is represented by the interval of each state variable. A three-phase interval DSSE model is proposed to construct the interval of each state variable. An improved iterative algorithm (IIA) is developed to solve the interval DSSE model and to obtain the lower and upper bounds of the interval. A cyber-attack is detected when the value of the state variable estimated by the traditional DSSE is out of the corresponding interval determined by the interval DSSE. To validate the proposed cyber-attack detection strategy, the basic principle of the cyber-attack is studied, and its general model is formulated. The proposed cyber-attack model and detection strategy are conducted on the IEEE 33-bus and 123-bus systems. Comparative experiments of the proposed IIA, Monte Carlo simulation algorithm, and interval Gauss elimination algorithm prove the validation of the proposed method.

Recourse-cost constrained robust optimization for microgrid dispatch with correlated uncertainties
IEEE Transactions on Industrial Electronics

Haifeng Qiu; Huan Long; Wei Gu; Guangsheng Pan

To accomplish more practical scheduling of microgrids under source-load uncertainties, this article first proposes a novel recourse-cost constrained adaptive robust optimization (RC-ARO) model with binary recourse variables. The dispatch plan in the nominal scenario is optimized in the first-stage to get the minimal operation cost, then the adjustment plan in the worst scenario is determined in the second-stage that minimizes the recourse-cost. This model has overcome the defect of conventional adaptive robust optimization (ARO), which can only get the scheduling plans in the worst scenario. Second, a spatiotemporal correlation model of wind power uncertainty is further developed based on the similarities of power time sequences, aimed at avoiding impossible scenarios in reality and reducing the conservativeness of independent uncertainty sets. Third, a new column-and-constraint generation (C&CG) algorithm with alternating optimization procedure (AOP) is developed to directly obtain the binary solution, which helps accelerating the solution of RC-ARO model using traditional nested-C&CG. Finally, case studies demonstrate the effectiveness and superiority of the proposed RC-ARO model, the developed uncertainty sets, and the novel solving algorithm. The solving time of C&CG-AOP reduces by half compared with nested-C&CG, and a larger scale of decision variables under uncertainties brings more significant speedup by the proposed algorithm.

High-precision dynamic modeling of two-staged photovoltaic power station clusters
IEEE Transactions on Power Systems

Peixin Li; Wei Gu; Huan Long; Ge Cao; Zhihuang Cao; Bin Xu

Accurate modeling is an important method for dynamic response analysis and control strategy verification of high photovoltaic (PV) penetration distribution networks. This paper proposes a precise dynamic modeling framework for the two-staged PV station cluster, namely as deep learning clustering hybrid modeling framework. It includes clustering-based equivalent model and error correction model (ECM). A long short-term memory network is used to form the ECM, which models the dynamic response error between the existing equivalent model and the detailed model. The competence of this framework is validated by numerous case studies based on a practical PV cluster construction. The simulation results reveal that the proposed method is featured of low complexity and fast response speed as the equivalent model but has much higher accuracy.

Image-based abnormal data detection and cleaning algorithm via wind power curve
IEEE Transactions on Sustainable Energy

Huan Long; Linwei Sang; Zaijun Wu; Wei Gu

This paper proposes an image-based algorithm for detecting and cleaning the wind turbine abnormal data based on wind power curve (WPC) images. The abnormal data are categorized into three types, negative points, scattered points, and stacked points. The proposed algorithm includes three steps, data pre-cleaning, normal data extraction, and data marking. The negative abnormal points, whose wind speed is greater than cut-in speed and power is below zero, are first filtered in the data precleaning step. The scatter figure of the rest wind power data forms the WPC image and corresponding binary image. In the normal data extraction step, the principle part of the WPC binary image, representing the normal data, is extracted by the mathematical morphology operation (MMO). The optimal parameter setting of MMO is determined by minimizing the dissimilarity between the extracted principle part and the reference WPC image based on Hu moments. In the data mark step, the pixel points of scattered and stacked abnormal data are successively identified. The mapping relationship between the wind power points and image pixel points is built to mark the wind turbine normal and abnormal data. The proposed image-based algorithm is compared with kmeans, local outlier factor, combined algorithm based on change point grouping algorithm and quartile algorithm (CA). Numerous experiments based on 33 wind turbines from two wind farms are conducted to validate the effectiveness, efficiency, and universality of the proposed method.

The data-driven schedule of wind farm power generations and required reserves
Energy

Huan Long; Zijun Zhang; Mu-Xia Sun;Yan-Fu Li

In this paper, a double layer wind farm and power reserve operational framework is introduced to develop the data-driven scheduling model for operating wind power generations and required power reserves. In this framework, data-driven approaches are applied to model real operational characteristics of wind power generation processes of individual wind generators (WGs) based on supervisory control and data acquisition (SCADA) data. The considered optimization objective in scheduling is to minimize the total wind power generation cost under the worst wind power generation scenario. Constraints considered include the wind power supply commitment and power supply reliability. By integrating the data-driven wind power models, multiple uncertainty sets, the cost objective function and reliability constraints, a data-driven scheduling model is formulated. Due to the complexity of the scheduling model, a two-level heuristic solution method is proposed to solve it. Two traditional power reserve approaches are regarded as the benchmark to evaluate the performance of the data-driven scheduling model. A comparative analysis is conducted to study the effectiveness of the solution method as well as the impact of different uncertainty sets and reliability constraints on the scheduling solutions.

Differential evolution with a new encoding mechanism for optimizing wind farm layout
IEEE Transactions on Industrial Informatics

Yong Wang; Hao Liu; Huan Long; Zijun Zhang; Shengxiang Yang

This paper presents a differential evolution algorithm with a new encoding mechanism for efficiently solving the optimal layout of the wind farm, with the aim of maximizing the power output. In the modeling of the wind farm, the wake effects among different wind turbines are considered and the Weibull distribution is employed to estimate the wind speed distribution. In the process of evolution, a new encoding mechanism for the locations of wind turbines is designed based on the characteristics of the wind farm layout. This encoding mechanism is the first attempt to treat the location of each wind turbine as an individual. As a result, the whole population represents a layout. Compared with the traditional encoding, the advantages of this encoding mechanism are twofold: 1) the dimension of the search space is reduced to two, and 2) a crucial parameter (i.e., the population size) is eliminated. In addition, differential evolution serves as the search engine and the caching technique is adopted to enhance the computational efficiency. The comparative analysis between the proposed method and seven other state-of-the-art methods is conducted based on two wind scenarios. The experimental results indicate that the proposed method is able to obtain the best overall performance, in terms of the power output and execution time.

Day-ahead prediction of bihourly solar radiance with a Markov switch approach
IEEE Transactions on Sustainable Energy

Yu Jiang; Huan Long; Zijun Zhang; Zhe Song

A Bayesian inference based Markov regime switching model is introduced to predict the intraday solar radiance. The proposed model utilizes a regime switching process to describe the evolution of the solar radiance time series. The optimal number of regimes and regime-specific parameters are determined by the Bayesian inference. The Markov regime switching model provides both the point and interval prediction of solar radiance based on the posterior distribution derived from historical data by the Bayesian inference. Four solar radiance forecasting models, the persistence model, the autoregressive (AR) model, the Gaussian process regression (GPR) model, and the neural network (NN) model, are considered as baseline models for validating the Markov switching model. The comparative analysis based on numerical experiment results demonstrates that in general the Markov regime switching model performs better than compared models in the day-ahead point and interval prediction of the solar radiance.

Wind turbine gearbox failure identification with deep neural networks
IEEE Transactions on Industrial Informatics

Long Wang; Zijun Zhang; Huan Long; Jia Xu; Ruihua Liu

The feasibility of monitoring the health of wind turbine (WT) gearboxes based on the lubricant pressure data in the supervisory control and data acquisition system is investigated in this paper. A deep neural network (DNN)-based framework is developed to monitor conditions of WT gearboxes and identify their impending failures. Six data-mining algorithms, the k-nearest neighbors, least absolute shrinkage and selection operator, ridge regression (Ridge), support vector machines, shallow neural network, as well as DNN, are applied to model the lubricant pressure. A comparative analysis of developed data-driven models is conducted and the DNN model is the most accurate. To prevent the overfitting of the DNN model, a dropout algorithm is applied into the DNN training process. Computational results show that the prediction error will shift before the occurrences of gearbox failures. An exponentially weighted moving average control chart is deployed to derive criteria for detecting the shifts. The effectiveness of the proposed monitoring approach is demonstrated by examining real cases from wind farms in China and benchmarked against the gearbox monitoring based on the oil temperature data.

Configuration optimization and analysis of a large scale PV/wind system
IEEE Transactions on Sustainable Energy

Huan Long; Mehrdad Eghlimi; Zijun Zhang

A multi-objective optimization model for optimizing the capacity size of the solar and wind component in a large scale PV/wind system is presented in this research. Objectives considered in the optimization include minimizing the power generation cost, maximizing the power supply reliability, and maximizing the average power fill rate. Decision variables, the number of PV arrays and wind turbines to be installed as well as the hourly load dispatches, are considered in the planning. A land budget is given to constrain the optimization. A solution algorithm is proposed to solve the proposed model and obtain exact solutions. Computational studies are performed to analyze the tradeoff among objectives in the optimization over the system's life-span. Combinations of load patterns, solar/wind resource availabilities, and penalties of power insufficiency construct different optimization scenarios. The computational results picture relationships among optimized objectives. The stability of the wind power generation contributes more than that of the solar power generation in achieving a more reliable power supply.

Data-driven wind turbine power generation performance monitoring
IEEE Transactions on Industrial Electronics

Huan Long; Long Wang; Zijun Zhang; Zhe Song; Jia Xu

This paper investigates the wind turbine power generation performance monitoring based on supervisory control and data acquisition (SCADA) data. The proposed approach identifies turbines with weakened power generation performance through assessing the wind power curve profiles. Profiles that statistically summarize the curvatures and shapes of a wind power curve over consecutive time intervals are constructed by fitting power curve models into SCADA data sets with a least square method. To monitor the variations of wind power curve profiles over time, multivariate and residual approaches are introduced and applied. Two blind industrial studies are conducted to validate the effectiveness of the proposed monitoring approach, and the results demonstrate high accuracy in detecting the abnormal power curve profiles of wind turbines and their associated time intervals.

A two-echelon wind farm layout planning model
IEEE Transactions on Sustainable Energy

Huan Long; Zijun Zhang

In this paper, a two-echelon layout planning model is proposed to determine the optimal wind farm layout to maximize its expected power output. In the first echelon, a grid composed of cells with equal size is utilized to model the wind farm, whereas the center of each cell is the potential slot for locating a wind turbine. Optimization models are developed to determine the optimal size of grid cells and the optimal cells for locating wind turbines. In the second echelon, the selected grid cells are then translated to sets of Cartesian coordinates. The model for determining the optimal coordinate rather than the center in a grid cell for locating each wind turbine is formulated. Due to the model complexity in both echelons, the random key genetic algorithm (RKGA) and particle swarm optimization (PSO) algorithm are applied to obtain the optimal solutions in the first and second echelon separately. The comparative analysis between the proposed two-echelon planning model and the traditional grid/coordinate-based planning models is conducted

Analysis of daily solar power prediction with data-driven approaches
Applied Energy

Huan Long; Zijun Zhang; Yan Su

Daily solar power prediction using data-driven approaches is studied. Four famous data-driven approaches, the Artificial Neural Network (ANN), the Support Vector Machine (SVM), the k-nearest neighbor (kNN), and the multivariate linear regression (MLR), are applied to develop the prediction models. The persistent model is considered as a baseline for evaluating the effectiveness of data-driven approaches. A procedure of selecting input parameters for solar power prediction models is addressed. Two modeling scenarios, including and excluding meteorological parameters as inputs, are assessed in the model development. A comparative analysis of the data-driven algorithms is conducted. The capability of data-driven models in multi-step ahead prediction is examined. The computational results indicate that none of the algorithms can outperform others in all considered prediction scenarios.