Net Load Forecast Research Articles

Power system flexibility analysis using net-load forecasting based on deep learning considering distributed energy sources and electric vehicles

Today, wind and solar energy sources have opened their place in the power system due to their environmental appeal. With the presence of these renewable energy sources (RESs), forecasting net load (NL) and its ramp are of huge importance owing their fundamental role in evaluating and improving the flexibility of the power grid. However, it is challenging to analysis flexibility and NL ramps using NL forecasts due to the uncertainty of RESs. Furthermore, the surge in the number of electric vehicles (EVs) as notable electricity consumers, and how their charging strategy can affect the NL curve are matters of concern. Most NL forecasting and flexibility analyses overlook the concurrent integration of wind and solar sources. Moreover, the impact of controlled EVs charging on grid flexibility remains largely unexplored. Existing studies also fail to assess the proper combination of wind and solar energy sources for future flexibility needs. This paper presents a framework for forecasting NL and detecting ramps with the simultaneous presence of EVs and RESs to investigate the impact of controlled charging of EVs on flexibility. Investigating charging of EVs as a solution is delved deeply in this paper to reduce NL ramps and improve flexibility. The numerical results show that controlled EVs charging may lead to 3% ramp reduction and 52% increment in the system flexibility. This paper also proposes to forecast and find the proper combination of RESs for the next years, leading to a 75% increment in the system flexibility compared to the business as usual RESs penetration.

PSGformer: A novel multivariate net load forecasting model for the smart grid

Smart grid intelligently transforms modern power system through the introduction of various advanced techniques, promoting the growth of distributed renewable energy sources and improving the efficiency of load management. The construction of smart grid recognizes the importance of net load forecasting, which represents the difference between the demanded load and the renewable energy generation, and has a significant impact on power system monitoring and control. Recently, some Transformer-based models have shown excellent performances in load forecasting. However, most of them perform weakly in face of net load forecasting due to the lack of emphasis on the specific trend variations and the inter-feature dependencies in cyclical temporal patterns. To address the two issues, we present the PSGformer that consists of multiscale series decomposition, seasonal module and trend module, which can focus on unique properties that affect the accuracy of net load forecasting by processing different components obtained from the series decomposition. The seasonal module seamlessly introduces graph attention network into the Transformer to compensate for the capability of capturing local inter-feature dependencies. The trend module adopts a new segment-wise iterative approach to capture trend patterns. Compared to the six baseline models, PSGformer has higher forecasting accuracy and stability on four load datasets from different countries.

Net load forecasting method in distribution grid planning based on LSTM network

Distribution grid planning involves multiple nodes, lines, equipment, and other elements. Due to the large scale of the system, there are complex interactions in space. The net load is affected by the load changes of different nodes. If this spatial complexity is not fully considered, the net load prediction results may be inaccurate. Therefore, in order to ensure the effect of net load forecasting, a method of net load forecasting in distribution grid planning based on the Long Short-Term Memory (LSTM) network is proposed. This method fully considers the characteristics of distribution grid planning and constructs a net load forecasting model for distribution grid planning based on the LSTM network. This model selects the 3σ criterion detects and corrects the singular values in the historical load data, and obtains the reasonable maximum time series results of each day; The adaptive noise complete set empirical mode decomposition method is used to decompose the sequence results and generates Intrinsic Mode Function (IMF) components of each time series; According to the component results, a load forecasting model based on LSTM network is constructed, and the initial learning rate and cell number parameters of LSTM network are optimized by improving the Pelican optimization algorithm to improve the precision of load forecasting of LSTM network. The test results show that the method can detect singular values in the data and weaken the impact of grid planning on netload forecasting; It can effectively complete the decomposition of historical load data, and each component after decomposition will not be aliased; The prediction error of net load is less than 1.25%, which can provide a reliable basis for grid planning of distribution network.

Direct short-term net load forecasting in renewable integrated microgrids using machine learning: A comparative assessment

Modern microgrids require accurate net load forecasting (NLF) for optimal operation and management at high shares of renewable energy sources. Machine learning (ML) principles can be used to develop precise and reliable NLF models. This paper evaluates the performance of different ML models, that are optimally trained using supervised learning regimes, for direct short-term net load forecasting (STNLF) in renewable microgrids. Different categories of ML models, such as neural network, ensemble, linear regression, nearest neighbor, and support vector machine were used. The comparative assessment was conducted utilizing historical net load, meteorological, and time-related categorical data acquired from the renewable integrated microgrid of the University of Cyprus in Nicosia, Cyprus. The results showed that all STNLF ML models achieved normalized root mean square error (nRMSE) values below 10%. Amongst the investigated models, the Bayesian neural network (BNN) presented the highest forecasting accuracy, exhibiting a daily average error of 3.58%. In addition, the BNN model yielded robust forecasts regardless of the season and weather conditions. Finally, the results demonstrated that optimally constructed ML models can be applied to provide STNLF in renewable integrated microgrids, which can be used by microgrid operators to efficiently control and manage their assets.

An Overview of Short-Term Load Forecasting for Electricity Systems Operational Planning: Machine Learning Methods and the Brazilian Experience

The advent of smart grid technologies has facilitated the integration of new and intermittent renewable forms of electricity generation in power systems. Advancements are driving transformations in the context of energy planning and operations in many countries around the world, particularly impacting short-term horizons. Therefore, one of the primary challenges in this environment is to accurately provide forecasting of the short-term load demand. This is a critical task for creating supply strategies, system reliability decisions, and price formation in electricity power markets. In this context, nonlinear models, such as Neural Networks and Support Vector Machines, have gained popularity over the years due to advancements in mathematical techniques as well as improved computational capacity. The academic literature highlights various approaches to improve the accuracy of these machine learning models, including data segmentation by similar patterns, input variable selection, forecasting from hierarchical data, and net load forecasts. In Brazil, the national independent system operator improved the operation planning in the short term through the DESSEM model, which uses short-term load forecast models for planning the day-ahead operation of the system. Consequently, this study provides a comprehensive review of various methods used for short-term load forecasting, with a particular focus on those based on machine learning strategies, and discusses the Brazilian Experience.

Interpretable transformer-based model for probabilistic short-term forecasting of residential net load

Short-term residential load forecasting is of great significance for the demand-side energy management of the grid and the home energy management of resident customers. The massive penetration of distributed renewable energy, especially small-scale solar photovoltaic (PV), in the residential sector urgently requires us to move from traditional load forecasting to net load forecasting. In recent years, deep learning techniques that can improve model forecasting performance have developed rapidly in the field of residential load forecasting. However, the opaque nature of deep learning makes its practical application very difficult. This paper proposes a Transformer-based probabilistic residential net load forecasting method that utilizes quantile regression to quantify uncertainty in future load demand. Meanwhile, to improve the interpretability of deep learning model, local variable selection network is developed to automatically select relevant features and provide feature-level explanations. Additionally, interpretable sparse self-attention mechanism is proposed to extract long-term temporal dependencies. Numerical experiments are carried out with data from real household smart meters. The results show that the proposed model outperforms other state-of-the-art forecasting models. In terms of point forecasting, compared with the most common deep time series forecasting model LSTM, the proposed model decreases by 21.3%, 27.3% and 20.9% On three point forecasting performance metrics. Compared with Vanilla Transformer, the proposed InterFormer decreases by 9.9%, 10.5% and 9.0% On three point forecasting performance metrics. In terms of probabilistic forecasting, compared with LSTM and Vanilla Transformer, the average pinball loss of the proposed model decreases by 26.2% and 17.0%, respectively. Additionally, and most importantly, our model provides users with explanations in terms of feature importance and temporal patterns.

Net Load Forecasting With Disaggregated Behind-the-Meter PV Generation

As worldwide use of residential photovoltaic (PV) systems grows, system operators and utilities will need to transition from forecasting pure demand to forecasting net load with behind-the-meter (BTM) PV generation. However, PV generation can be difficult to predict and the measurements of PV generation from BTM residential systems are often invisible behind a measurement of the net load, making net load forecasting challenging. This paper proposes a novel two-stage framework for net load forecasting in areas with limited observability and high BTM PV generation. First, the profiles of observable customers are used to disaggregate the net load measurements into the pure load and PV generation. Then, separate models are used to forecast the PV generation and pure load individually, and the results are combined for a net load forecast. This paper also proposes a compensator for correcting the error of the net load forecast, using historical forecast errors of the PV generation, pure load, and net load. The proposed framework is tested through two case studies for areas with high BTM PV penetration and less than 10% observable customers. The two-stage forecasting model is compared to two benchmark methods – a time series forecasting model, and a model that forecasts the net load directly using historical net load measurements. Results show that the proposed disaggregation-forecasting framework reduces the error of the net load forecast compared to both benchmark models. In addition, when the net load forecast error is periodic, the compensator can correct the error to improve the forecast accuracy.

Geometric Deep-Learning-Based Spatiotemporal Forecasting for Inverter-Based Solar Power

Conventional power grids dominated by synchronous generators gradually shift to variable renewable-energy-integrated grids. With the growth of building-level photovoltaic (PV) panels and other inverter-based resource (IBR) deployments in recent years, market retailers and distribution operators have had to deal with the additional operational and management challenges posed by unobserved energy flow if its behind-the-meter (BTM) configuration. Also, the intermittent and stochastic nature of IBR-based solar power introduces uncertainty into the net load forecasting of electric distribution systems. Hence, management and control with high uncertainty in load prediction due to unobservable PV is a technical challenge. This article presents deep-learning-based algorithms for BTM PV power generation using a limited number of sensors in a given distribution system and extending to adjacent geographical areas. The proposed BTM PV forecasting method is based on geometric deep learning—a spatiotemporal graph neural network by processing the relationship between data in a non-Euclidean graph structure. The predictions of short-term BTM PV are aggregated with loss estimation at the data aggregation point with the net load forecasting to compute the true load forecasting with superior performance. The developed BTM PV forecasting method significantly improved true load forecasting results, which were validated and analyzed using a collection of actual BTM PV and load measurements in a test distribution feeder.

Net-load forecasting of renewable energy systems using multi-input LSTM fuzzy and discrete wavelet transform

Net-load forecasting of renewable energy systems using multi-input LSTM fuzzy and discrete wavelet transform

Energy Trading in Local Electricity Markets With Behind-the-Meter Solar and Energy Storage

Distributed energy resources, especially residential behind-the-meter photovoltaics (BTM PV), have been playing increasingly important roles in modern smart grids. Residential netload, which is closely tied with customers' gross load consumption and weather, is usually the only data available for the market operator in a local electricity market (LEM). This paper seeks to design customized prices for an LEM that consists of an agent, BTM PV, energy storage (ES), prosumers, and consumers. The LEM agent who owns a community-scale ES system is responsible for operating the market, determining the internal price, and facilitating the energy sharing within the community. A hierarchical energy trading infrastructure is considered, where the LEM agent acts as the mediator between the external utility grid and customers. A two-stage decision-making framework, including both look-ahead ES scheduling and real-time customized price design, is developed for the agent's profit maximization. Besides, the impacts of netload forecasting and BTM PV disaggregation are also investigated. The customer's consumption behavior is modeled as a utility maximization problem. Compared with the benchmark uniform price design, it is found that the customized pricing scheme could further improve the LEM agent's profit by 4% to 130%, depending on the weather conditions and seasonal load patterns.

TransformGraph: A novel short-term electricity net load forecasting model

The development of the smart grid recognizes the importance of projecting electrical net load forecasting, which represents the difference between load demand and installed renewable energy sources (RES), such as wind and solar power. For net load forecasting, many approaches like statistical models, machine learning, and deep learning have been developed. Because of the significant uncertainty of RES, these models suffer from poor forecasting accuracy. In this study, we propose the TransformGraph, a novel Transformer model for electricity net load forecasting, which combines the Transformer and graph convolutional network (GCN). Firstly, the GCN is used as an input embedding layer to encode multivariate input sequences, which can fill the gap that the correlation information is not thoroughly considered in the Transformer. Then, the self-attention mechanism in the Transformer is used to capture the temporal dependence of the sequence data. Finally, the predicted load values are output using a feed forward neural network. Data from three countries derived from Open Power System Data (OPSD) sources are employed for the case study. Comparisons between the TransformGraph and the other five forecasting models demonstrate that the proposed one has higher forecasting accuracy and stability.

Highly accurate peak and valley prediction short-term net load forecasting approach based on decomposition for power systems with high PV penetration

The increasing penetration of photovoltaic has been reshaping the electricity net load curve, which has a significant impact on power system operation and short-term dispatch scheduling. Accurate short-term net load forecasting is essential to ensure reliable and economical operations of a power system. Nonetheless, most of the existing net load forecasting approaches are mostly focused on net load forecasting at household, distribution or microgrid level, but not at grid system-wide level. They also suffer from low accuracy due to the presence of uncertainties on high-frequency fluctuations in the net load. This paper proposes a new improved two-stage net load forecasting method at grid system-wide level. Firstly, it contributes to eliminate the high-frequency components with insignificant amount of energy from the original net load by using the “Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise-ICEEMDAN” technique. Then, net load decomposition outcomes form the inputs of a computationally efficient and accurate “Long Short-Term Memory-LSTM” network algorithm to produce an accurate day-ahead forecasting, which lays out the foundation of day-ahead power dispatch scheduling. The superiority of the suggested algorithm was confirmed by comparing the obtained results against five other algorithms that use different empirical based decomposition techniques along with Back Propagation (BP) or LSTM. Statistical metrics, Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) were computed to show the model accuracy. The validity of the proposed method is verified by the net load data of “South West Interconnected System” power network in Western Australia, which refers to the total demand on the conventional generators made up of the consumers’ actual demand plus system losses, minus the solar power harnessed by the rooftop PV panels installed within the grid system. Achieving a very high day-ahead net load forecasting accuracy of 96.67% confirms our hypothesis on ICEEMDAN’s capability to decompose the net load carefully into different meaningful components.

Short-term net load forecast in distribution networks with PV penetration behind the meter

In recent years there has been a strong expansion of photovoltaic (PV) distributed generation systems. A high PV penetration level can cause uncertainty in the operation and management processes carried out by electric utilities, since most meters register the net load, i.e., the actual load minus the power generated by the PV systems behind the meter. The goal of this paper was to analyze the difference in the net load forecasting error achieved by models using or not using behind-the-meter PV generation data. The PV plant is connected to the lower voltage side of the power substation, representing a penetration level of more than 35% of the total load. The study shows that the best forecasting results are obtained with an indirect approach using two forecasting models, one for the total load and the other for the PV generation. However, the difference with respect to the results obtained with a unique net load forecasting model is almost negligible, which may be of special interest for power system distributors or other agents who do not have access to behind-the-meter generation data.

Lyapunov Optimization in Online Battery Energy Storage System Control for Commercial Buildings

The high instantaneous discharging capability of battery energy storage systems (BESSs) make them ideal candidates for reducing peak loads in commercial buildings. An efficient online BESS control algorithm can be beneficial for reducing the monthly electricity bill of individual commercial buildings. Conventional model-based BESS control algorithms rely heavily on accurate long-horizon net load forecast, which is difficult to obtain. To address this problem, we develop a Lyapunov optimization-based online BESS control algorithm and derive its theoretical performance bound. Comprehensive numerical study results using real world commercial building smart meter data in southern California show that our proposed Lyapunov optimization-based online control algorithm with time-varying weighting parameters yields higher savings in electricity cost and requires less computation time in comparison to the state-of-the-art baseline algorithms.

Flexibility study of the Greek power system using a stochastic programming approach for estimating reserve requirements

This paper presents a detailed flexibility and cost assessment study of the Greek power system for the time period 2021–2025 where for the first time the reserve requirements are determined based on the statistical analysis of the residual load variability and forecast error at hourly steps. In particular, the long-term uncertainties are modeled via forecast scenarios that incorporate different assumptions regarding temperature, load growth, renewable energy sources penetration and hydraulic conditions. In the operational time frame, the net load forecast errors are incorporated via scenarios and a two-stage stochastic unit commitment algorithm with balancing actions is implemented. Forecast error scenarios are generated with a non-parametric approach, using multivariate copula functions to capture the temporal dependencies. A Monte Carlo sampling is used to create uncertainty scenarios and a clustering algorithm to reduce their size. These scenarios are then used as input in the flexibility and cost evaluation study that is subsequently conducted. The study is carried out by solving the unit commitment problem in which all the required technical constraints are considered and then indicative reliability and flexibility metrics are calculated. The two-stage stochastic approach for estimating the reserve requirements is compared with the deterministic approach that has been used from the Greek Transmission System Operator for determining the reserve capacity. The obtained results present great interest, since in both cases the calculated reliability (LOLE) and flexibility metrics (IRRE, flexibility residual) take acceptable values, however in the stochastic approach there is a significant decrease in the system cost. Therefore, through this detailed flexibility study it is shown that the two-stage stochastic approach consists a useful alternative for assessing the reserve requirements of the Greek power system, since decreased operation costs is achieved while security of the system can be established. • We propose an upgraded modeling for net load forecast errors. • Reserve requirements are calculated using an advanced stochastic approach. • Detailed flexibility and adequacy study for the Greek power system. • Provide an all-around view of the system operation challenges. • The proposed approach leads to a significant decrease in the system cost.

Chance-constrained allocation of UFLS candidate feeders under high penetration of distributed generation

Under-Frequency Load Shedding (UFLS) schemes are the last resort to contain a frequency drop in the grid by disconnecting part of the demand. The allocation methods for selecting feeders that would contribute to the UFLS scheme have traditionally relied on the fact that electric demand followed fairly regular patterns, and could be forecast with high accuracy. However, recent integration of Distributed Generation (DG) increases the uncertainty in net consumption of feeders which, in turn, requires a reformulation of UFLS-allocation methods to account for this uncertainty. In this paper, a chance-constrained methodology for selecting feeders is proposed, with mathematical guarantees for the disconnection of the required amount of load with a certain pre-defined probability. The correlation in net-load forecasts among feeders is explicitly considered, given that uncertainty in DG power output is driven by meteorological conditions with high correlation across the network. Furthermore, this method is applicable either to systems with conventional UFLS schemes (where relays measure local frequency and trip if this magnitude falls below a certain threshold), or adaptive UFLS schemes (where relays are triggered by control signals sent in the few instants following a contingency). Relevant case studies demonstrate the applicability of the proposed method, and the need for explicit consideration of uncertainty in the UFLS-allocation process.

Aggregated Net-load Forecasting using Markov-Chain Monte-Carlo Regression and C-vine copula

Net-load is the difference between total load and renewable generation and acts as an effective system load to which dispatchable generators are scheduled and system flexibility requirements are estimated. This necessitates accurate net-load forecasts for optimum scheduling and flexibility requirement estimations. Net-Load Forecasting (NLF) got only little attention in existing literature even though it is essential for optimal generation scheduling and power system flexibility requirement estimations. Time series and machine learning models have been used for NLF in recent years, however, there is vast scope existing for improving accuracy. Probabilistic forecasting models are widely used for various forecasting problems such as load, wind, and solar generation forecasting as those models show improved forecasting accuracy. In this context, this paper proposes a novel probabilistic aggregated very short-term NLF model based on Markov Chain Monte Carlo (MCMC) Regression. MCMC uses data augmentation, where unobserved variables are simulated from their posterior distribution and this makes MCMC approach suitable for regression. Further, MCMC forecasts are improved by incorporating C-vine copula-based Joint Probability Distribution (JPD) of expected load, wind, and solar generation forecasting errors. The C-vine copula is suitable for such multi-variable JPD estimation due to its enormous flexibility in stochastic multivariate dependence modelling compared to standard elliptical and Archimedean copulas. Results show that the proposed NLF model outperforms reference models and can produce accurate net-load forecasts.

Operation-adversarial scenario generation

This paper proposes a modified conditional generative adversarial network (cGAN) model to generate net load scenarios for power systems that are statistically credible, conditioned by given labels (e.g., seasons), and, at the same time, “stressful” to the system operations and dispatch decisions. The measure of stress used in this paper is based on the operating cost increases due to net load changes. The proposed operation-adversarial cGAN (OA-cGAN) internalizes a DC optimal power flow model and seeks to maximize the operating cost and achieve a worst-case data generation. The training and testing stages employed in the proposed OA-cGAN use historical day-ahead net load forecast errors and has been implemented for the realistic NYISO 11-zone system. Our numerical experiments demonstrate that the generated operation-adversarial forecast errors lead to more cost-effective and reliable dispatch decisions.

FPSeq2Q: Fully Parameterized Sequence to Quantile Regression for Net-Load Forecasting With Uncertainty Estimates

The increased penetration of Renewable Energy Sources (RES) as part of a decentralized and distributed power system makes net-load forecasting a critical component in the planning and operation of power systems. However, compared to the transmission level, producing accurate short-term net-load forecasts at the distribution level is complex due to the small number of consumers. Moreover, owing to the stochastic nature of RES, it is necessary to quantify the uncertainty of the forecasted net-load at any given time, which is critical for the real-world decision process. This work presents parameterized deep quantile regression for short-term probabilistic net-load forecasting at the distribution level. To be precise, we use a Deep Neural Network (DNN) to learn both the quantile fractions and quantile values of the quantile function. Furthermore, we propose a scoring metric that reflects the trade-off between predictive uncertainty performance and forecast accuracy. We evaluate the proposed techniques on historical real-world data from a low-voltage distribution substation and further assess its robustness when applied in real-time. The experiment’s outcomes show that the resulting forecasts from our approach are well-calibrated and provide a desirable trade-off between forecasting accuracies and predictive uncertainty performance that are very robust even when applied in real-time.

Application of Real-Time Distribution Grid Monitoring for Grid Forecasting and Control Considering Incomplete Information of Resources Behind-the-Meter

This paper presents a rolling horizon optimal energy management mechanism using real-time grid monitoring data. The proposed mechanism includes novel approaches in terms of real-time data processing, net-load forecasting, and optimal scheduling of battery energy storage systems. The proposed real-time data processing and net-load forecasting techniques use fast training and computationally efficient methods based on grid monitoring data. The data processing and parameter forecasting methods are based on auto-regressive with exogenous variables (ARX). Two sets of features including similar values in previous hours, days, and weeks as well as calendar effects are used for training the forecast model. Furthermore, the impact of additional non-synchronized weather features adopted from meteorology databases on the forecasts’ accuracy is discussed. Finally, a real-time optimal scheduling is proposed to optimize battery energy storage systems (BESS), maximizing the self-consumption at grid and community levels. The application of real-time grid measurement in the proposed algorithm allows handling the impacts of loads and generators behind the meter without having their detailed information. The developed method is being effectively used in a real low voltage distribution grid in Switzerland.