In the UK, the traditional settlement process aims to allocate customers' total electricity consumption into each half-hour slot based on the typical load profiles. However, the large difference between load profiles and the real energy consumption lead to the unfair and inaccurate electricity charges. In order to tackle this problem, the smart meter will be introduced and this paper, for the first time, accurately quantifies the improvement of using smart metering data in electricity settlement. The result shows that all of the customers are being mischarged with an average of 9.52 GBP overcharged and 10.34 GBP undercharged per household, which will be discussed in terms of potential causes and errors.
In this study, four representative machine learning methods (support vector machine (SVM), maximum entropy (MaxEnt), random forest (RF), and artificial neural network (ANN)) were employed to construct a landslide susceptibility map (LSM) in Xulong Gully (XLG), southwest China. The models were subsequently compared in order to select the best-performing model. This model was further improved to optimize the machine learning method. A total of 16 layers were extracted from the collected data and employed as conditional factors for the correlation analysis and subsequent modelling. The LSMs were then divided into four levels (very high susceptibility (VH), high susceptibility (H), moderate susceptibility (M) and low susceptibility (L)). The results were verified by receiver operating characteristic (ROC) curves, Root Mean Squared Error (RMSE) and Frequency Ratio (FR). The higher of the area under ROC curve (AUC) and the lower the RMSE, the more accurate and stable the performance. Following the factor performance analysis, the optimal SVM model was linearity improved to the Trace Ratio Criterion (TRC)-SVM, with a better performance and the ability to overcome the factor defect. The comprehensive comparisons and proposed LSM can support future research, as well as local authorities in the development of landslide remission strategies.
Uncertainty has become a key challenge in energy profiles at the domestic level, which is influenced by a variety of factors, such as behaviors, technologies, weather conditions, energy prices, and so on. These factors influence the household demand at various time horizons, hence result in diverse uncertainty natures across time-scales. Therefore, it is crucial to understand the temporal natures of the demand uncertainty at different time-scales, particularly intra-day and inter-days. This paper first attempts to quantify and characterize uncertainties of household electricity demand across multiple time scales. An advanced data-driven temporal-dependency Haar expansions uncertainty quantification approach is thus proposed by this paper. The proposed approach consists of two stages. At first stage, uncertainty will be decomposed into multiple components of different time scales and then quantified by Monte Carlo methods. The second stage will further characterize natures of each uncertainty component in terms of threefold: 1) temporal uncertainty distribution at a certain time scale; 2) coupling degree between two time scales; and 3) uncertainty propagation natures to the system level of uncertainty component at a certain time scale. The proposed analysis is demonstrated on Irish smart meter database. To prove the efficiency and validity of the proposed approach, the error bound and computational cost is analyzed.
With the development of deep learning techniques, supervised learning has achieved performances surpassing those of humans. Researchers have designed numerous corresponding models for different data modalities, achieving excellent results in supervised tasks. However, with the exponential increase of data in multiple fields, the recognition and classification of unlabeled data have gradually become a hot topic. In this paper, we employed a Reinforcement Learning framework to simulate the cognitive processes of humans for effectively addressing novel class discovery in the Open-set domain. We deployed a Member-to-Leader Multi-Agent framework to extract and fuse features from multi-modal information, aiming to acquire a more comprehensive understanding of the feature space. Furthermore, this approach facilitated the incorporation of self-supervised learning to enhance model training. We employed a clustering method with varying constraint conditions, ranging from strict to loose, allowing for the generation of dependable labels for a subset of unlabeled data during the training phase. This iterative process is similar to human exploratory learning of unknown data. These mechanisms collectively update the network parameters based on rewards received from environmental feedback. This process enables effective control over the extent of exploration learning, ensuring the accuracy of learning in unknown data categories. We demonstrate the performance of our approach in both the 3D and 2D domains by employing the OS-MN40, OS-MN40-Miss, and Cifar10 datasets. Our approach achieves competitive competitive results.
Due to the growth of intermittent generation and flexible demand, the difference between real metered load profiles and predicted profiles has increased significantly. This has caused a higher cost to suppliers as they have to mitigate the errors by using costly fast response generators or buying expensive energy from other suppliers. Improving load forecast accuracy is an alternative to reduce the difference and consequently the costs, but it relies on large quantities of historical load data which is not necessarily available. This paper utilises a novel control strategy for energy storage systems to mitigate forecast errors for suppliers. This results in energy cost savings and the Use-of-System (UoS) savings. In order to test the charging/discharging strategies and quantify the economic benefits, a case study has been conducted by utilizing smart metering data. The case study has produced a 33.2% cost reduction in the energy cost savings.
As an effective tool in Demand-Side Response (DSR) programmes, the "Time-Of-Use" (TOU) tariff stimulates customers to modify their usage behaviour. However, there are considerable deviations in responsiveness among different customers, even to the same TOU tariff. There are two drawbacks in the existing analyses: 1) customers' psychological features are omitted, i.e. the willingness to respond; 2) the interaction effect among features, which describe the load and socio-economic characteristics, are ignored. Therefore, this paper proposes an interaction-aware framework to identify the significant psychological, socio-economic and load criteria to evaluate customers' responsiveness to different TOU tariffs. The results find that the significant characteristics of the more responsive customers include: smooth load profiles, high consumption levels overnight, large family size. A significant psychological feature interacted with load features demonstrates that customers with lower confidence in load-reduction activities also can successfully respond, as long as their load consumption is small.
Most existing network pricing methodologies are designed for retailers and large customers. With the development of responsive technologies, domestic customers may have very different impacts on networks cost, thus calling for a cost-reflective network pricing method for mass customers in the retail market. The naive volumetric and marginal pricing methods may cause issues of inequality and mis-signaling. Inspired by the Passenger Car Equivalent (PCE) in transportation economics, this paper proposes a Unit Home Equivalent (UHE) pricing to reflect the compatibility between electricity networks and a certain type of users. The method is validated against the Distribution Use of System (DUoS) charging methodology in the U.K. by using real network and household data. The results show the proposed pricing can encourage existing customers to adjust energy usage behaviours and guide new customers and EVs to the right locations.
Deep learning has been proven of great potential in various time-series forecasting applications. To exploit the potential and extendibility of deep learning in electricity load forecasting, this paper for the first time presents a comprehensive deep learning assessment on performing load forecasting at different levels through the power systems. The assessment is demonstrated via two extreme cases: 1) regional aggregated demand with an example of New England electricity load data, and 2) disaggregated household demand with examples of 100 individual households from Ireland. The state-of-the-art deep recurrent neural network is implemented for this assessment. Compared with the shallow neural network, the proposed deep model has improved the forecasting accuracy in terms of MAPE by 23% at aggregated level and RMSE by 5% at disaggregated level.
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)