Short-term Demand Forecasting Research Articles

A novel forward operator-based Bayesian recurrent neural network-based short-term net load demand forecasting considering demand-side renewable energy

Currently, traditional electricity consumers are now shifting to a new role of prosumers since more integration of renewable energy to demand side. Accurate short-term load demand forecasting is significant to safe, stable, and reliable operation of a renewable energy-dominated power system. In this paper, a short-term load forecasting model based on a bidirectional long short-term memory network (Bi-LSTM) using kernel transfer operator is proposed to achieve short-term load demand forecasting. To consider the influence of seasonality, holiday effects and weather on load demand forecasting, and simultaneously to improve the accuracy and performance of the forecasting model, this paper implements the dimensionality reduction of the input data by introducing an improved kernel transfer operator based on the Perron-Frobenius method. On this basis, a Bayesian Bi-LSTM for short-term load demand forecasting is formulated to obtain the probability prediction interval of load demand. To verify the validity of the proposed method, the actual historical load demand in a certain region in China was used for model training and verification. The forecasting results are compared with several conventional load demand methods using probability prediction technics. Simulation analysis shows that the proposed method can effectively reflect the short-term uncertainty of load demand, and is superior to the conventional methods in terms of forecasting accuracy and computational performance.

Daily hotel demand forecasting with spatiotemporal features

PurposeGiven the importance of spatial effects in improving the accuracy of hotel demand forecasting, this study aims to introduce price and online rating, two critical factors influencing hotel demand, as external variables into the model, and capture the spatial and temporal correlation of hotel demand within the region.Design/methodology/approachFor high practical implications, the authors conduct the case study in Xiamen, China, where the hotel industry is prosperous. Based on the daily demand data of 118 hotels before and during the COVID-19 period (from January to June 2019 and from January to June 2021), the authors evaluate the prediction performance of the proposed innovative model, that is, a deep learning-based model, incorporating graph convolutional networks (GCN) and gated recurrent units.FindingsThe proposed model simultaneously predicts the daily demand of multiple hotels. It effectively captures the spatial-temporal characteristics of hotel demand. In addition, the features, price and online rating of competing hotels can further improve predictive performance. Meanwhile, the robustness of the model is verified by comparing the forecasting results for different periods (during and before the COVID-19 period).Practical implicationsFrom a long-term management perspective, long-term observation of market competitors’ rankings and price changes can facilitate timely adjustment of corresponding management measures, especially attention to extremely critical factors affecting forecast demand, such as price. While from a short-term operational perspective, short-term demand forecasting can greatly improve hotel operational efficiency, such as optimizing resource allocation and dynamically adjusting prices. The proposed model not only achieves short-term demand forecasting, but also greatly improves the forecasting accuracy by considering factors related to competitors in the same region.Originality/valueThe originalities of the study are as follows. First, this study represents a pioneering attempt to incorporate demand, price and online rating of other hotels into the forecasting model. Second, integrated deep learning models based on GCN and gated recurrent unit complement existing predictive models using historical data in a methodological sense.

BM-DDPG: An Integrated Dispatching Framework for Ride-Hailing Systems

This paper proposes an integrated dispatching framework for matching drivers with riders in ride-hailing systems. The goal is to compute matching solutions that maximize social welfare and benefit both sides of the market, such that the sustainable growth of the ride-hailing system is ensured. The proposed framework integrates data-driven proactive guidance strategies with batched matching optimization to increase social welfare, improve matching rate and reduce rider wait time. Proactive guidance strategies are computed by leveraging short-term demand forecasts based on historical data. Taken the resulting guidance strategies as inputs, the batched matching algorithm computes optimal bipartite matching between drivers and riders in a batch. Using New York City taxi data from 2016 March 1st to March 31st as input, we conduct a numerical study to evaluate the performance of the proposed framework and compare it with existing approaches in the literature. Our results show that the proposed framework improves social welfare for up to 50%. It also increases the matching rate by an average of 20% and reduces the average rider wait time by over 15%. This implies a strong potential for the proposed dispatching framework to improve service quality in ride-hailing systems.

Short-Term Electrical Load Demand Forecasting Based on LSTM and RNN Deep Neural Networks

As the development of smart grids is increasing, accurate electric load demand forecasting is becoming more important for power systems, because it plays a vital role to improve the performance of power companies in terms of less operating cost and reliable operation. Short-term load forecasting (STLF), which focuses on the prediction of few hours to one week ahead predictions and is also beneficial for unit commitment and cost-effective operation of smart power grids, is receiving increasing attention these days. Development and selection of an accurate forecast model from different artificial intelligence (AI)-based techniques and meta-heuristic algorithms for better accuracy is a challenging task. Deep Neural Network (DNN) is a group of intelligent computational algorithms which have a viable approach for modeling across multiple hidden layers and complex nonlinear relationships between variables. In this paper, a model for STLF using deep learning neural network (DNN) with feature selection is proposed. A wide range of intelligent forecast models was designed and tested based on multiple activation functions, such as hyperbolic tangent (tanh), different variants of rectifier linear unit (ReLU), and sigmoid. Among the others, DNN with leaky ReLu produced the best forecast accuracy. Regarding the precision of the methods used in this research work, certain output measures, such as absolute percentage error (MAPE), mean square error (MSE), and root mean square error (RMSE) are used. There was also a reliance on multiple parametric and variable details to determine the capability of the smart load forecasting techniques.

Short-Term Demand Forecast of E-Commerce Platform Based on ConvLSTM Network.

Based on real sales data, this article constructed LGBM and LSTM sales prediction models to compare and verify the performance of the proposed models. In this article, we forecast the product sales of stores in the future T + 3 days and use MAPE as the evaluation index. The experiment shows that the product sales prediction model based on the convolutional LSTM (ConvLSTM) network has better prediction accuracy. From a store point of view, ConvLSTM prediction model MAPE was 0.42 lower than the long short-term memory (LSTM) network and 0.68 lower than LGBM. From the perspective of commodity categories, different commodity categories are suitable for different forecasting methods. Some categories are suitable for regression forecasting, while others are suitable for time-series forecasting. Among the categories suitable for time-series forecasting, the ConvLSTM model performs the best.

A new taxonomy for vector exponential smoothing and its application to seasonal time series

In short-term demand forecasting, it is often difficult to estimate seasonality accurately, owing to short data histories. However, companies usually have multiple products with similar seasonal demand patterns. A possible solution in this case is to use the components of several time series from a homogeneous family, thus estimating seasonal coefficients based on cross-sectional information. Motivated by this practical problem, we propose a new taxonomy of Parameters, Initial States and Components (PIC), which exploits homogeneous features of time series. We then apply this framework to vector exponential smoothing. We develop a model selection mechanism based on information criteria to select the appropriate PIC restrictions. We then conduct a simulation experiment and empirical analysis on retail data in order to assess the performance of point forecasts and prediction intervals of the models within this framework.

Modeling the effects of the governmental responses to COVID-19 on transit demand: The case of Athens, Greece

Short-term demand forecasting is essential for the public transit system, allowing for effective operations planning. This is especially relevant in the highly uncertain environment created by the SARS‑CoV‑2 pandemic. In this paper, we attempt to develop accurate prediction models of transit ridership in Athens, Greece, using Autoregressive Fractional Integrated time series models enhanced with SARS‑CoV‑2-related exogenous variables. The selected exogenous variables are, from the one hand, the ratio of weekly SARS‑CoV‑2 infections over the infections 3 weeks before (capturing the dynamics of the pandemic, as a proxy for fear of transmitting the disease while commuting), and from the other hand, an index of the stringency of the government’s SARS‑CoV‑2-related measures and regulations. The developed ARFIMAX models have been fitted separately on bus and metro ridership data and wield comparable and statistically significant results. In both models, the exogenous variables prove to be statistically significant and their values are intuitive, suggesting a linear interrelation between them and transit ridership.

Short-term forecasting of origin-destination matrix in transit system via a deep learning approach

ABSTRACT Short-term travel demand forecasting is the critical first step to support transportation system management. Complex relevance among Origin-Destination (OD) pairs, temporal dependencies, and external factors bring challenges to it. An innovative deep learning approach, Multi-Fused Residual Network (MF-ResNet) is proposed to forecast travel demand. The complex relevance among OD pairs is converted into graphical-based spatial dependencies by treating OD matrix as the input of the model. The residual network units enable MF-ResNet to model not only near but also distant spatial correlations. Three conv-based residual network units model the temporal closeness, mid-term periodicity, as well as long-term periodicity features, and Fully-Connected (F-C) layers capture external factors. The fusion techniques coordinate all of the features. The proposed method is applied to the short-term forecasts of metro OD matrix in Shenzhen, China. The experimental results show that MF-ResNet can capture multiple complex dependencies robustly and outperforms traditional methods in terms of forecasting accuracy.

Moving Average Smoothing for Gregory-Newton Interpolation: A Novel Approach for Short-Term Demand Forecasting

Demand forecasting is a fundamental task when it comes to industrial stability. A high deficiency in demand forecasting results in high variability, and hence the presence of undesirable bullwhip effect in the supply chain. This causes unnecessary inventory or low service levels and consequently less profit and lower efficiency of the entire supply chain. In this paper, a new forecasting method is developed for relatively short-term demand forecasting. The new approach combines the simple moving average method with Gregory-Newton interpolation curve fitting. The developed method has two phases; a) Data Preparation (Smoothing) phase, where the data get cleaned of possible existing noise and, b) Forecasting phase where the forecast for a new period is obtained. The simple moving average is utilized twice for the 1st phase, then Gregory-Newton forward interpolation is used in the 2nd phase. The proposed method considers datasets with trend and seasonality separately. Numerical examples are conducted for testing the approach efficiency. Comparisons with other acknowledged methods of the literature are present. Results showed high potential of utilizing the new approach for real-life situations.

Electricity Load and Internet Traffic Forecasting Using Vector Autoregressive Models

This study was conducted to investigate the applicability of measuring internet traffic as an input of short-term electricity demand forecasts. We believe our study makes a significant contribution to the literature, especially in short-term load prediction techniques, as we found that Internet traffic can be a useful variable in certain models and can increase prediction accuracy when compared to models in which it is not a variable. In addition, we found that the prediction error could be further reduced by applying a new multivariate model called VARX, which added exogenous variables to the univariate model called VAR. The VAR model showed excellent forecasting performance in the univariate model, rather than using the artificial neural network model, which had high prediction accuracy in the previous study.

Station-level short-term demand forecast of carsharing system via station-embedding-based hybrid neural network

Station-based one-way carsharing system brings transformation to public mobility and spurs the growth of sharing economy. The accurate estimation of rental and return demand of carsharing stations to support vehicle relocation is essential, hence a station-level short-term demand forecasting method named as station-embedding-based hybrid neural network (SEHNN) integrated by variational graph auto-encoder (VGAE) and long short-term memory network (LSTM) is proposed. The VGAE module undertakes the functions of station feature extraction and embedding, while the LSTM module captures the time series regularity and forecast the station-level demand. The results from the real data of Lanzhou, China demonstrate that, compared with ElasticNet, ARIMA, LSTM, and ConvLSTM, the mean absolute error of the proposed model targeted at hourly demand forecasting is reduced by 56.5%, 47.2%, 38.7%, and 38.5%, respectively, and it also outperforms some widely used models among different intervals and scales including main stations and subset carsharing system.

Short-term Demand Forecasting of Shared Bicycles Based on Seasonal Grey Markov Model

According to the characteristics of periodicity, nonlinearity and randomness of shared bicycle riding data, a seasonal grey Markov model was proposed. In the new prediction model, the seasonal GM (1,1) model was used to get the prediction results firstly. Then, Markov model was used to modify the prediction residual. In the process of residual correction, the residual sequence was selected according to the new information priority principle, and the residual was modified by the expectation of the median value in the state interval, which improves the prediction accuracy of the model. Finally, the model was applied to the demand forecast of the Citi Bike shared bicycles in New York, the first three weeks of June 2019 during the Saturday peak hours. The numerical results show that the mean absolute percentage error (MAPE) of predicted value of seasonal grey Markov model is 2.72%, which is superior to the traditional GM (1,1) model, seasonal GM (1,1) model and grey Markov model.

Deep sequence to sequence Bi-LSTM neural networks for day-ahead peak load forecasting

The power industry is currently facing the problem of an electricity supply–demand imbalance. The most inexpensive and efficient solution to alleviate this imbalance is to decrease electricity demand. Local electrical utilities should deploy demand response programs (DRP), and short-term peak demand forecasting (STPDF) plays a crucial role in their successful deployment. In residential sectors, peak demand forecasting is also critical because the energy policies, technological growth, and changing climate are further increasing the peak demand. Therefore, an accurate peak demand forecasting will help utility companies in avoiding blackouts and secure a continuous power supply by implementing subsidy schemes such as DRP. However, daily peak load is volatile, nonstationary, and nonlinear in nature, and hence it is hard to predict it accurately. This research work for the first time has attempted to design, implement, and test deep bidirectional long short-term memory based sequence to sequence (Bi-LSTM S2S) regression approach for “day-ahead” peak demand forecasting and has accomplished preliminary success. The day-ahead peak electricity demand forecasting model is designed and tested using the MATLAB software. For performance comparison, shallow Bi-LSTM S2S, shallow LSTM S2S, deep LSTM S2S, Levenberg-Marquardt backpropagation artificial neural networks (LMBP-ANN), and medium Gaussian support vector regression (MG-SVR) forecasting models are also developed and tested. Mean absolute percentage error (MAPE) and Root Mean Squared Error (RMSE) are used as performance metrics. It has been found out that in terms of both performance metrics, the proposed deep Bi-LSTM S2S day-ahead “peak” forecasting model has outperformed all the other models on both public holidays and normal days. The load pattern on public holidays is always different than on normal days, and there is always less data available in contrast to the normal days. Therefore, it is hard to accurately forecast their load.

A decision integration strategy for short-term demand forecasting and ordering for red blood cell components

Blood transfusion is one of the most crucial and commonly administered therapeutics worldwide. The need for more accurate and efficient ways to manage blood demand and supply is an increasing concern. Building a technology-based, robust blood demand and supply chain that can achieve the goals of reducing ordering frequency, inventory level, wastage and shortage, while maintaining the safety of blood usage, is essential in modern healthcare systems. In this study, we summarize the key challenges in current demand and supply management for red blood cells (RBCs). We combine ideas from statistical time series modeling, machine learning, and operations research in developing an ordering decision strategy for RBCs, through integrating a hybrid demand forecasting model using clinical predictors and a data-driven multi-period inventory problem considering inventory and reorder constraints. We have applied the integrated ordering strategy to the blood inventory management system in Hamilton, Ontario using a large clinical database from 2008 to 2018. The proposed hybrid demand forecasting model provides robust and accurate predictions, and identifies important clinical predictors for short-term RBC demand forecasting. Compared with the actual historical data, our integrated ordering strategy reduces the inventory level by 40% and decreases the ordering frequency by 60%, with low incidence of shortages and wastage due to expiration. If implemented successfully, our proposed strategy can achieve significant cost savings for healthcare systems and blood suppliers. The proposed ordering strategy is generalizable to other blood products or even other perishable products.

Short-term demand forecasting for online car-hailing using ConvLSTM networks

This paper used the previous data of online car-hailing orders in Haikou provided by Didi Chuxing GAIA Initiative to predict the short-term demand for online car-hailing service. This paper contains two main steps. The first step is about converting online car-hailing demand for images that contains spatiotemporal feature of online car-hailing orders. This paper draws a picture every 72 min from 2017∕5∕8 to 2017∕8∕8, with a total of 1,840 binary vector images. The second step is to employ the deep learning method of a Conv-LSTM neutral network to the image for online car-hailing demand prediction. Conv-LSTM has excellent image prediction properties, so it is ideal for predicting such binary vector figures with spatiotemporal information. After learning the first 1460 images, the last 380 images were simulated, predicted and tested. Finally, the simulation results of the last 20 images were taken as the effect of the model. Our result shows that the Conv-LSTM neutral network can train the model with a reasonable output and is suitable for short-term forecasting network ride-hailing demand forecast with spatiotemporal feature information. By comparing five different training session times, it can be seen that when the number of training session reaches 30, the model reaches an optimum. Reasonable prediction results can provide data support for vehicle dispatching and distribution, solve problems such as energy waste and traffic congestion caused by asymmetric supply and demand, and maximize the interests of passengers, drivers and ride-hailing platforms.

Forecasting the short-term urban gas daily demand in winter based on the XGBoost algorithm

The share of natural gas in the primary energy consumption is continuously increasing. The urban gas plays a major part in the natural gas consumption in China, and thus the accurate prediction of the short-term urban gas daily demand in winter facilitates the healthy development of the natural gas consumption and helps to maintain the secure and appropriate urban gas supply. This paper thoroughly analyzed the factors affecting the urban gas demand in winter, introduced the XGBoost algorithm for the forecast, and made accurate predictions regarding the violently-fluctuating short-term urban gas daily demand in winter. In the simulation tests, the Root-Mean-Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) of the forecast made in the proposed model are 8.269 and 1.515%, respectively. Compared with other statistical algorithms and machine learning/artificial intelligence algorithms such as the multivariate regression, random forest and support vector machine, the proposed model has shown better fitting performance and higher prediction accuracy. Therefore, the XGBoost-based prediction model can provide references for the short-term urban gas daily demand forecasting.

A Novel Short-Term Load Forecasting Method by Combining the Deep Learning With Singular Spectrum Analysis

One of the major issues about the operation of power systems is the prediction of load demand. Moreover, load forecasting is of prime concern to system operators. Recently, the integration of power system elements, such as renewable energy sources, energy storages and electricity vehicle, brings more challenges, particularly when there are large fluctuations in forecasting cycle. This study concentrates on short-term load demand forecasting and proposes a hybrid method that combines Singular Spectrum Analysis (SSA) with deep-learning Neural Network (NN) techniques. In the beginning, the SSA technique is applied as an initial filter to remove noises. Next, a hybrid neural network, including Backpropagation Neural Network (BPNN) and Long Short-Term Memory (LSTM), is developed and trained. Then, the trained network is used as the core forecasting algorithm. Each SSA has different forms to combine with neural networks. The performance of the proposed forecasting algorithm is demonstrated using the power demand data recorded in Taiwan. Furthermore, this study compares the forecasting results by five models, including SSA, SSA-BPNN, ANN, SSA-LSTM, Wavelet Neural Network (WNN) and LSTM. The forecasting results reveal that the proposed forecasting model using Singular Spectrum Analysis provides the best performance on load forecasts.

Short-term electric power demand forecastingusing a hybrid model of SARIMA and SVR

Short-term electric power demand forecasting is the most basic and important application of smart grid. With the rapid development of renewable energy and clean energy in recent years, power demand forecasting gains special attention again. It has a great impact on the planning of power generation units and the purchase and sale of power market. Furthermore, itis also conducive to the realization of demand response and resource allocation efficiency and reliability, thus contributing to the Photovoltaic power generation system as well. Although generous methods have been proposed, it remains to be an open challenge owing to its limited precision. In this paper, a hybrid method of Seasonal Auto-Regressive Integrated Moving Average (SARIMA) and Support Vector Regression (SVR) is proposed for hourly forecasting, which overcomes the difficulty of generalization of a single model. The core thought of this model is using SARIMA to fit the linear part of the series and correcting the deviation by SVR. We tested the accuracy of this method on a public data set, and the result shows that it performs much better than the single SARIMA model on the fitting effect. Also, we compared our model with the recent three works, and it demonstrates a decrease of 18.102%,10.534% and 4.757% in forecasting error respectively. To our surprise, when we usedthe previous four hours of data to predict the current data by the single SVR, we got the best performance of all models. In the end, we analyzed the possible causes for this result.

Hourly and Daily Urban Water Demand Predictions Using a Long Short-Term Memory Based Model

This case study uses a long short-term memory (LSTM)-based model to predict short-term urban water demands for the Hefei City of China. The performance of the LSTM-based model is compared with the autoregressive integrated moving average (ARIMA) model, the support vector regression (SVR) model, and the random forests (RF) model based on data with time resolutions ranging from 15 min to 24 h. Additionally, this paper investigates the performance of the LSTM-based model in predicting multiple successive data points. Results show that the LSTM-based model can offer predictions with improved accuracy than the other models when dealing with data with high time resolutions, data points with abrupt changes, and data with a relatively high uncertainty level. It is also observed that the LSTM-based model exhibits the best performance in predicting multiple successive water demands with high time resolutions. In addition, the inclusion of external parameters (e.g., temperature) cannot enhance the performance of the LSTM-based model, but it can improve ARIMAX's prediction ability (ARIMAX is the ARIMA with variables). These observations provide additional and improved evaluations regarding the LSTM-based models used for short-term urban water demand forecasting, thereby enabling their wider adoption in practical applications.

Short-term water demand forecasting using hybrid supervised and unsupervised machine learning model

Regression Tree (RT) forecasting models are widely used in short-term demand forecasting. Likewise, Self-Organizing Maps (SOM) models are known for their ability to cluster and organize unlabeled big data. Herein, a combination of these two Machine Learning (ML) techniques is proposed and compared to a standalone RT and a Seasonal Autoregressive Integrated Moving Average (SARIMA) models, in forecasting the short-term water demand of a municipality. The inclusion of the Unsupervised Machine Learning clustering model has resulted in a significant improvement in the performance of the Supervised Machine Learning forecasting model. The results show that using the output of the SOM clustering model as an input for the RT forecasting model can, on average, double the accuracy of water demand forecasting. The Mean Absolute Percentage Error (MAPE) and the Normalized Root Mean Squared Error (NRMSE) were calculated for the proposed models forecasting 1 h, 8 h, 24 h, and 7 days ahead. The results show that the hybrid models outperformed the standalone RT model, and the broadly used SARIMA model. On average, hybrid models achieved double accuracy in all 4 forecast periodicities. The increase in forecasting accuracy afforded by this hybridized modeling approach is encouraging. In our application, it shows promises for more efficient energy and water management at the water utilities.