Forecasting Model Research Articles

A Novel Approach for Predicting the Mid–Long-Term Radiation Dose in the Case of a Hypothetical STSBO Nuclear Accident for an Operating Nuclear Power Plant

Four severe nuclear accident scenarios have been identified for operating nuclear power plants (ONPPs). However, there is a research gap in predicting the mid–long-term radiation doses for these scenarios. This study aims to address this gap by proposing a novel approach for predicting the mid–long-term radiation dose in the case of a hypothetical short-term station blackout (STSBO) scenario, one of the aforementioned scenarios. Firstly, the Weather Research and Forecasting (WRF) model was coupled with the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) (WRF-HYSPLIT) model to establish an atmospheric transport and diffusion model for airborne radionuclides, and the regularity of the atmospheric transport and diffusion for the airborne radionuclides was determined. Subsequently, the Residual Radioactive Material Guidelines (RESRAD) OFFSITE (RESRAD-OFFSITE) code was utilized to establish a radiation dose model for predicting the mid–long-term radiation dose resulting from the airborne radionuclides, and the evolution of the mid–long-term radiation dose was analyzed. Finally, the proposed approach was applied to an ONPP, and the results were used to predict the mid–long-term public radiation dose. The results indicated that the total radiation dose would be lower than the dose limit recommended by the International Commission on Radiological Protection (1 mSv/yr) from the second month to the 100th year after the hypothetical STSBO nuclear accident, and the total radiation dose would decrease slowly over time. Recommendations are made for offsite emergency response measures. These research findings can assist ONPPs in analyzing their environmental impacts in the event of an STSBO scenario.

Climate projection of tropical cyclone lifetime in the western North Pacific basin

Abstract In this study, the potential changes of tropical cyclone (TC) lifetime in the western North Pacific basin are examined for different future climates. Using homogeneous 9 km-resolution dynamical downscaling with the Weather Research and Forecasting (WRF) model, we show that TC averaged lifetime displays insignificant change under both low and high greenhouse-gas concentration scenarios. However, more noticeable changes in the tails of TC lifetime statistics are captured in our downscaling simulations, with more frequent long-lived TCs (lifetime of 8-11 days) and less short-lived TCs (lifetime of 3-5 days). Unlike present-day simulations, it is found that the correlation between TC lifetime and the Niño index is relatively weak and insignificant in all future downscaling simulations, thus offering little explanation for these changes in TC lifetime statistics based on the El Niño-Southern Oscillation. More detailed analyses of TC track distribution in the western North Pacific basin reveal, nevertheless, a noticeable shift of TC track patterns towards the end of the 21st century. Such a change in TC track climatology results in an overall longer duration of TCs over the open ocean, which is consistent across future scenarios and periods examined in this study. This shift in the TC track pattern is ultimately linked to changes in the Western North Pacific Subtropical High, which retreats to the south during July and to the east during August-September. The results obtained in this study provide new insights into how large-scale circulations can affect TC lifetime in the western North Pacific basin in warmer climates.

Forecasting the future demand on health workforce – the Austrian model

Abstract Background An aging population and upcoming retirements of health care professionals pose new challenges to the Austrian health care system. On the one hand the health care and nursing needs will increase while at the same time a shortage on health care personnel threatens adequate supply. We as the Austrian Public Health Institute estimate the risk of an undersupply with model-based health workforce forecasting and are therefore able to advise the federal states and the ministry of health on early measures. Methods We developed a forecasting model to estimate the future supply and demand. The main influencing factors are the demographic development as well as the upcoming retirements and graduates. Different environments and needs in different federal states (e.g. changing future skill mix) are considered and can be implemented as additional influencing factors. Usually a baseline scenario as a forward projection of the status quo is developed for defined professions and workplace settings. Then depending on the stakeholder needs additional scenarios are calculated and discussed. A usual forecast horizon is 10 to 15 years. Results Model output are time series of yearly workforce personnel numbers (heads and full-time equivalents) and demand in the defined settings. An analysis of possible gaps shows the need for action. Conclusions The Austrian health workforce model is a flexible tool to estimate future needs on health workforce based on different assumptions. But as assumptions and influencing factors must be represented as data, the data collection can be quite laborious. Comprehensive data is the basis for a comprehensive health workforce planning. Further work is needed to develop practical and evidence-based nursing-sensitive indicators for nursing workfoce reportings on a country level. In Austria the health professionals register forms the basis for the description of the nursing workforce structure, but that there is a need for data sources on care effectiveness. Key messages • Comprehensive data is the basis for a comprehensive health workforce planning. • An aging population and upcoming retirements of health care professionals pose new challenges to the Austrian health care system.

Multi-Step-Ahead Time Series Forecasting using Deep Learning and Fuzzy Time Series-based Error Correction Method

Recently time series forecasting has become one of the prime application areas of climatology, economics and industries. Many research works are conducted to forecast the time series more accurately. But few of them are concentrated on predicting the time series over an extended future horizon, and there is also a scope to improve their forecasting accuracy. This work proposes a multi-step-ahead foresting method to produce a stable and accurate forecasting result for the extended future horizon. Firstly, a deep learning-based forecasting model is proposed to predict the time series. Secondly, a fuzzy time series-based error correction model is implemented to enhance the prediction performance of the deep learning model. Here to optimize all the fuzzy time series (FTS) parameters in an integrated way, an integrated butterfly optimization (FTS-IBO) algorithm is proposed. In this study, two different types of real-world multivariate time series datasets are used to analyze the forecasting performance of the proposed model. The performance of the proposed FTS-IBO algorithm is compared with the traditional butterfly optimization (FTS-BO) algorithm. The experimental results show that the FTS-IBO technique is superior to the FTS-BO technique. The forecasting performance of the proposed model has also compared the other state-of-the-art models, and the simulation results exhibit that the proposed model produces a more accurate prediction performance for multi-step-ahead time series forecasting problems compared to other models.

Maintaining Agricultural Production Profitability—A Simulation Approach to Wheat Market Dynamics

Profitability is a complex notion, intertwined with theoretical, socio-economic, and modern economic considerations. While traditionally linked to monopoly advantages, contemporary views broaden this to include shareholder value creation. In agriculture, profitability faces impediments such as rising input costs, market volatility, and economic disparities, emphasizing technological innovation, market conditions, and policy as crucial determinants. This study presents an econometric simulation model to analyze and forecast wheat production profitability in Poland from 2023 to 2027, providing insights into production efficiencies, market dynamics, and policy impacts. Using a system of recursive equations, the model forecasts profitability, integrating econometric techniques with expert insights. It examines variables like resource use, production levels, and price changes, validated through FADN data. The findings reveal that price relations and production efficiencies are the key to profitability. The model highlights market volatility, particularly through “price scissors”, as the main factor influencing profitability. Favorable price conditions significantly boost profitability and incentivize production. The model is a critical tool for analyzing wheat profitability, highlighting the influence of market and policy changes. Future work could extend the model’s use to other regions or crops, incorporating advanced technologies to improve accuracy. This study offers valuable insights for agricultural economics, aiding stakeholders in strategic decision-making.

An Improved Neural Network Algorithm for Energy Consumption Forecasting

Accurate and efficient forecasting of energy consumption is a crucial prerequisite for effective energy planning and policymaking. The BP neural network has been widely used in forecasting, machine learning, and various other fields due to its nonlinear fitting ability. In order to improve the prediction accuracy of the BP neural network, this paper introduces the concept of forecast lead time and establishes a mathematical model accordingly. Prior to training the neural network, the input layer data are preprocessed based on the forecast lead time model. The training and forecasting results of the BP neural network when and when not considering forecast lead time are compared and verified. The findings demonstrate that the forecast lead time model can significantly improve the prediction speed and accuracy, proving to be highly applicable for short-term energy consumption forecasting.

Time Series Forecasting of Thermal Systems Dispatch in Legal Amazon Using Machine Learning

This paper analyzes time series forecasting methods applied to thermal systems in Brazil, specifically focusing on diesel consumption as a key determinant. Recognizing the critical role of thermal systems in ensuring energy stability, especially during low rain seasons, this study employs bagged, boosted, and stacked ensemble learning methods for time series forecasting focusing on exploring consumption patterns and trends. By leveraging historical data, the research aims to predict future diesel consumption within Brazil’s thermal energy sector. Based on the bagged ensemble learning approach a mean absolute percentage error of 0.089% and a coefficient of determination of 0.9752 were achieved (average considering 50 experiments), showing it to be a promising model for the short-time forecasting of thermal dispatch for the electric power generation system. The bagged model results were better than for boosted and stacked ensemble learning methods, long short-term memory networks, and adaptive neuro-fuzzy inference systems. Since the thermal dispatch in Brazil is closely related to energy prices, the predictions presented here are an interesting way of planning and decision-making for energy power systems.

Evaluation of WRF model performance with different microphysics schemes for extreme rainfall prediction in Lagos, Nigeria: Implications for urban flood risk management

In Nigeria, particularly in urban areas like Lagos, flooding is a frequent natural hazard. In 2011, Lagos experienced one of its worst floods resulting in significant economic losses and displacement of people. In recent years, Lagos has continued to grapple with flooding challenges, with an equally significant flood episode occurring in 2021. This study focuses on predicting floods and forecasting extremely heavy rainfall in West Africa's equatorial zone using the Weather Research and Forecasting (WRF) model, particularly in humid tropical environments like Lagos. The study discusses the need to review existing flood models and adopt alternative flood models to address the limitations of flood prediction. As potential causes of these rainfall episodes, the interconnections between synoptic systems such as tropical easterly waves, southwesterly winds related to the West African Monsoon, and local topography and oceanic conditions are investigated. Three key metrics: root mean square error (RMSE), mean bias (MB), and mean absolute error (MAE) are used to assess the effectiveness of the computational model. Results indicate that the WRF model, specifically when using the Thompson parameterisation, can estimate the amount of rainfall accumulated over a 24-h period. This suggests that the model can predict the size of daily precipitation during intense rain events. The Thompson scheme shows better performance compared to the WSM6 scheme while evaluating the stations and episodes. During the rainfall episode on July 10, 2011, Thompson's spatial rainfall predictions were better than WSM6, resulting in a decrease in root mean square error (RMSE) of 15–31% depending on the area. Simulations of the July 2021 episode also show better performance, with a decrease in RMSE of 11–25% when comparing Thompson to WSM6 scheme. The Thompson scheme’s improved ability is directly linked to a more accurate depiction of the microphysical mechanisms that control the rainfall formation. By explicitly simulating the dynamics of ice crystals and graupel, it is possible to accurately replicate the processes of orographic lifting and moist convection that are responsible for driving intense monsoon precipitation. In addition, Thompson scheme shows a reduced degree of systemic bias in comparison to WSM6, with a 75% reduction in the average bias in rainfall accumulation over the research area. The combination of the advanced Thompson microphysics method and WRF's atmospheric dynamics shows a high level of accuracy in predicting intense rainfall and the risk of floods in this area with diverse tropical topography.

Significant of Blood Glucose Time-Series Forecasting with Temporal Fusion Transformer Model for Type 1 Diabetes

Blood glucose (BG) prediction has advanced to a state of the art thanks to deep learning models, which have been demonstrated to enhance type 1 diabetes (T1D) therapy. Nevertheless, most current models are limited to single-horizon prediction and have several practical drawbacks, including poor interpretability. In this study, we develop a novel approach to optimize the forecasting model in the training processes using the Optuna function, cross-validation, and the latest dataset. We offer a novel deep learning framework for multi-horizon BG prediction: the temporal fusion transformer (TFT). TFT employs a self-attention mechanism to extract long-term temporal dependencies and enables a model with an auto-tuning adjustment approach on hyperparameters and a cross-validation function on univariate and multivariate input models. On a clinical dataset with T1D subjects, namely ShanghaiT1DM (16 subjects), D1NAMO (9 subjects), and OhioT1DM (6 subjects) datasets, it achieved an average root mean square error (RMSE) of the three datasets used. The TFT model gave a lower error score than the baseline models of neural hierarchical interpolation for time series (N-HiTS)and long short-term memory (LSTM) with the values of 10.08+0.31 mg/dL and 12.34+0.62 mg/dL on the univariate input model and 9.18+1.21 mg/dL and 14.33+0.52 mg/dL on the multivariate input model for the prediction horizons of 30 and 60 minutes, respectively. This result explained that the TFT model was adequate for carrying out multi-horizon BG value-level forecasting and potentially deploying on-edge devices to improve clinical efforts to manage BG levels for T1D patients in real-time application.

Modeling of hybrid renewable resources and comprehensive feasibility analysis of grid interactive energy system for commercial building: a case study

ABSTRACT The growing energy demand in commercial buildings and businesses poses significant environmental challenges due to the dependence on fossil fuel-based energy. To address this issue, increasing the use of renewable energy sources and decreasing reliance on fossil fuels is crucial. Commercial buildings, in particular, have many opportunities to harness energy from renewable resources, such as solar, wind, and biomass. This study analyzed historical energy consumption data from a commercial building using MATLAB to forecast future energy demands. A range of machine learning algorithms were employed to accurately predict energy consumption, including Ensemble Boosting, Ensemble Bagging, Decision Trees, Support Vector Machines (SVM), and Neural Networks (NN). A Hybrid Renewable Energy (HRE) system was designed to meet the building’s energy needs, integrating renewable sources with grid-interactive inverters. Performance metrics for the forecasting models were calculated and documented, followed by a detailed techno-economic and environmental feasibility analysis of the HRE system. Different levels of renewable energy penetration were explored to optimize HRE designs using load-following dispatch algorithms tailored to the commercial building’s specific needs. The study revealed that achieving a renewable energy proportion of 12.6% resulted in the lowest cost of electricity (COE) at $0.135 and a Net Present Cost (NPC) of $8.84 million. Additionally, a reliability study using load flow techniques confirmed the stability of the optimized Hybrid Renewable Energy system. The energy costs for varying levels of renewable penetration, ranging from 10% to 40% is also explored in this research. In summary, transitioning to renewable energy sources in commercial buildings can reduce environmental impact while ensuring a sustainable energy supply.

A novel hybrid forecasting system for crude oil futures prices: A dual perspective of deterministic forecasting and uncertainty analysis

The crude oil market occupies a crucial place in the composition of the global economy, however, the high level of uncertainty in crude oil prices makes it a challenging task to construct reliable forecasting techniques to obtain reliable expected results. Therefore, to effectively cope with the high uncertainty and the technical limitations of existing forecasting studies, this study constructs a hybrid forecasting system from the dual perspectives of deterministic forecasting and uncertainty analysis. With the concept of “denoising-integration” as the core, the system integrates several denoising techniques, econometric methods, and artificial intelligence methods, and establishes a hybrid forecasting model with high accuracy and robustness with the assistance of multi-objective optimization theory. Furthermore, by introducing fuzzy theory onto the basis of deterministic prediction results, an interval forecasting model with a narrower interval width and better comprehensive performance is constructed under the premise of ensuring that the coverage of prediction intervals meets the requirements of the confidence level. In this study, the forecasting system is applied to four global core crude oil futures markets, and the empirical study finds that the performance is improved by more than 50 % compared with the benchmark model, which has an impressive performance advantage.

Climatology of Orographic Precipitation Gradients Over High Mountain Asia Derived From Dynamical Downscaling

AbstractWithin High Mountain Asia (HMA), the annual melting of glaciers and snowpack provides vital freshwater to populations living downstream. Precipitation over HMA can directly affect the freshwater availability in this region by altering the mass balance of glaciers and snowpack. However, available reanalyses and downscaling simulations lack the resolution required to understand important glacier‐scale variations in precipitation. This study aimed to determine the current characteristics of orographic precipitation gradients (OPG) by curve‐fitting daily precipitation as a function of elevation from a 15‐year, 4‐km grid spaced Weather Research and Forecasting (WRF) model simulation focused on the Himalayan, Karakoram, and Hindu‐Kush mountain ranges. To facilitate precipitation curve‐fitting, the WRF model grid points were separated into regions of similar orientation, referred to as facets. Akaike Information Criterion‐corrected values and an F‐test p‐value identified the need for a curvature term to account for a varying OPG with elevation. Regions with similar seasonal variability were found using ‐means clustering of the monthly mean OPG coefficients. The central Himalayan slope's intra‐seasonal variability of OPG depended on synoptic scale conditions, in which cyclonically‐forced heavy‐precipitation events produced strong sublinear increases in precipitation with elevation. Initial testing of precipitation estimates using monthly coefficients showed promising results in downscaling daily WRF precipitation; the daily mean absolute error at each grid point had a lower magnitude than the daily mean precipitation total, on average. Results provide a physically‐based context for machine learning algorithms being developed to predict OPG and downscale precipitation output from global climate models over HMA.

Technological shocks and stock market volatility over a century

This paper provides a novel perspective on the innovation-stock market nexus by examining the predictive relationship between technological shocks and stock market volatility using data over a period of more than 140 years. Utilizing annual patent data for the U.S. and a large set of economies to create proxies for local and global technological shocks and a mixed-sampling data (MIDAS) framework, we present robust evidence that technological shocks capture significant predictive information regarding future realizations of stock market volatility, both in- and out-of-sample and at both the short and long forecast horizons. Further economic analysis shows that investment portfolios created by the volatility forecasts obtained from the forecasting models that incorporate technological shocks as predictors in volatility models experience significantly lower return volatility in the out-of-sample horizons, which in turn helps to improve the risk-return profile of those portfolios. Our findings present a novel take on the nexus between technological innovations and stock market dynamics and pave the way for several interesting avenues for future research regarding the role of technological innovations on asset pricing tests and portfolio models.

Recommended coupling to global meteorological fields for long-term tracer simulations with WRF-GHG

Abstract. Atmospheric transport models are often used to simulate the distribution of greenhouse gases (GHGs). This can be in the context of forward modeling of tracer transport using surface–atmosphere fluxes or flux estimation through inverse modeling, whereby atmospheric tracer measurements are used in combination with simulated transport. In both of these contexts, transport errors can bias the results and should therefore be minimized. Here, we analyze transport uncertainties in the commonly used Weather Research and Forecasting (WRF) model coupled with the greenhouse gas module (WRF-GHG), enabling passive tracer transport simulation of CO2 and CH4. As a mesoscale numerical weather prediction model, WRF's transport is constrained by global meteorological fields via initialization and at the lateral boundaries of the domain of interest. These global fields were generated by assimilating various meteorological data to increase the accuracy of modeled fields. However, in limited-domain models like WRF, the winds in the center of the domain can deviate considerably from these driving fields. As the accuracy of the wind speed and direction is critical to the prediction of tracer transport, maintaining a close link to the observations across the simulation domain is desired. On the other hand, a link that is too close to the global meteorological fields can degrade performance at smaller spatial scales that are better represented by the mesoscale model. In this work, we evaluated the performance of strategies for keeping WRF's meteorology compatible with meteorological observations. To avoid the complexity of assimilating meteorological observations directly, two main strategies of coupling WRF-GHG with ERA5 meteorological reanalysis data were tested over a 2-month-long simulation over the European domain: (a) restarting the model daily with fresh initial conditions (ICs) from ERA5 and (b) nudging the atmospheric winds, temperatures, and moisture to those of ERA5 continuously throughout the simulation period, using WRF's built-in four-dimensional data assimilation (FDDA) in grid-nudging mode. Meteorological variables and simulated mole fractions of CO2 and CH4 were compared against observations to assess the performance of the different strategies. We also compared planetary boundary layer height (PBLH) with radiosonde-derived estimates. Either nudging or daily restarts similarly improved the meteorology and GHG transport in our simulations, with a small advantage of using both methods in combination. However, notable differences in soil moisture were found that accumulated over the course of the simulation when not using frequent restarts. The soil moisture drift had an impact on the simulated PBLH, presumably via changing the Bowen ratio. This is partially mitigated through nudging without requiring daily restarts, although not entirely alleviated. Soil moisture drift did not have a noticeable impact on GHG performance in our case, likely because it was dominated by other errors. However, since the PBLH is critical for accurately simulating GHG transport, we recommend transport model setups that tie soil moisture to observations. Our method of frequently re-initializing simulations with meteorological reanalysis fields proved suitable for this purpose.

Prediction of Scour Depth for Diverse Pier Shapes Utilizing Two-Dimensional Hydraulic Engineering Center’s River Analysis System Sediment Model

Examining scouring around bridge piers is crucial for ensuring water-related infrastructure’s long-term safety and stability. Accurate forecasting models are essential for addressing scour, especially in complex water systems where traditional methods fall short. This study investigates the application of the HEC-RAS 2D sedimentation model, which has recently become available for detailed sediment analysis, to evaluate its effectiveness in predicting scoring around various pier shapes and under different water conditions. This study offers a comprehensive assessment of the model’s predictive capabilities by focusing on variables such as water velocity, shear stress, and riverbed changes. Particular attention was paid to the influence of factors like floating debris and different pier geometries on scour predictions. The results demonstrate that while the HEC-RAS 2D model generally provides accurate predictions for simpler pier shapes—achieving up to 85% precision—it shows varied performance for more complex designs and debris-influenced scenarios. Specifically, the model overpredicted scouring depths by approximately 20% for diamond-shaped piers and underpredicted by 15% for square piers in debris conditions. Elliptical piers, in contrast, experienced significantly less erosion, with scour depths up to 30% shallower compared to other shapes. This study highlights the novel application of the HEC-RAS 2D model in this context and underscores its strengths and limitations. Identified issues include difficulties in modeling water flow and debris-induced bottlenecks. This research points to the improved calibration of sediment movement parameters and the development of advanced computational techniques to enhance scour prediction accuracy in complex environments. This work contributes valuable insights for future research and practical applications in civil engineering, especially where traditional scour mitigation methods, such as apron coverings, are not feasible.

Enhanced Short-Term Load Forecasting: Error-Weighted and Hybrid Model Approach

To tackle the challenges of high variability and low accuracy in short-term electricity load forecasting, this study introduces an enhanced prediction model that addresses overfitting issues by integrating an error-optimal weighting approach with an improved ensemble forecasting framework. The model employs a hybrid algorithm combining grey relational analysis and radial kernel principal component analysis to preprocess the multi-dimensional input data. It then leverages an ensemble of an optimized deep bidirectional gated recurrent unit (BiGRU), an enhanced long short-term memory (LSTM) network, and an advanced temporal convolutional neural network (TCN) to generate predictions. These predictions are refined using an error-optimal weighting scheme to yield the final forecasts. Furthermore, a Bayesian-optimized Bagging and Extreme Gradient Boosting (XGBoost) ensemble model is applied to minimize prediction errors. Comparative analysis with existing forecasting models demonstrates superior performance, with an average absolute percentage error (MAPE) of 1.05% and a coefficient of determination (R2) of 0.9878. These results not only validate the efficacy of our proposed strategy, but also highlight its potential to enhance the precision of short-term load forecasting, thereby contributing to the stability of power systems and supporting societal production needs.

LFformer: An improved Transformer model for wind power prediction.

Wind power forecasting has complex nonlinear features and behavioral patterns across time scales, which is a severe test for traditional forecasting techniques. To address the multi-scale problem in wind power forecasting, this paper innovatively proposes an ultra-short-term forecasting model LFformer based on Legendre-Fourier, which firstly focuses on the important information in the input sequences by using the encoder-decoder architecture, and then scales the range of the original data with the Devlin normalization method, and then utilizes the Legendre polynomials to The data sequence is projected into a bounded dimensional space, the historical data is compressed using feature representation, then feature selection is performed using the low-rank approximation method of Fourier Transform, the prediction is inputted into the multilayer perceptron through the multi-scale mixing mechanism, and finally the results are outputted after back-normalization. The experimental results show that compared with the existing prediction methods, the model realizes the improvement of prediction accuracy and stability, especially in the ultra-short-term prediction scenario, with obvious advantages. The research results are not only valuable for improving the overall operational efficiency of the wind power system, but also help to enhance the stable operation of the power grid, which provides strong technical support and guarantee for wind power enterprises to improve the competitiveness of bidding for Internet access in the power market competition.

Development of PM2.5 Forecast Model Combining ConvLSTM and DNN in Seoul

Accurate prediction of PM2.5 concentrations is essential for public health management, especially in areas affected by long-range pollutant transport. This study presents a hybrid model combining convolutional long short-term memory (ConvLSTM) and deep neural networks (DNNs) to enhance PM2.5 forecasting in Seoul, South Korea. The hybrid model leverages ConvLSTM’s ability to capture spatiotemporal dependencies and DNN’s strength in feature extraction, enabling it to outperform standalone CMAQ and DNN models. For the T1 forecast (6 h averages), the ConvLSTM-DNN model exhibited superior performance, with an RMSE of 7.2 µg/m3 compared to DNN’s 8.5 µg/m3 and CMAQ’s 10.1 µg/m3. The model also maintained high categorical accuracy (ACC) and probability of detection (POD) for critical PM2.5 levels while reducing false alarms (FARs), particularly in bad and very bad events. Although its performance decreases over extended forecast periods, the ConvLSTM-DNN model demonstrates its utility as a robust forecasting tool. Future work will focus on optimizing the network structure to improve long-term forecast accuracy.

Prediction of Carbon Dioxide Concentrations in Strawberry Greenhouse by Using Time Series Models

Carbon dioxide (CO2) concentrations play an important role in plant production, as they have a direct impact on both plant growth and yield. Therefore, the objectives of this study were to predict CO2 concentrations in the greenhouse by applying time series models using five datasets. To estimate the CO2 concentrations, this study was conducted over a four-month period from 1 December 2023 to 31 March 2024, in a strawberry-cultivating greenhouse. Fifteen sensors (MCH-383SD, Lutron, Taiwan) were installed inside the greenhouse to measure CO2 concentration at 1-min intervals. Finally, the dataset was transformed into intervals of 1, 5, 10, 30, and 60 min. The time-series data were analyzed using the autoregressive integrated moving average (ARIMA) and the Prophet Forecasting Model (PFM), with performance assessed through root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R2). The evaluation indicated that the best model performance was achieved with data collected at 1-min intervals, while model performance declined with longer intervals, with the lowest performance observed at 60-min intervals. Specifically, the ARIMA model outperformed across all data collection intervals while comparing with the PFM. The ARIMA model, with data collected at 1-min intervals, achieved an R2 of 0.928, RMSE of 7.359, and MAE of 2.832. However, both ARIMA and PFM exhibited poorer performances as the interval of data collection increased, with the lowest performance at 60-min intervals where ARIMA had an R2 of 0.762, RMSE of 19.469, and MAE of 11.48. This research underscores the importance of frequent data collection for precise environmental control in greenhouse agriculture, emphasizing the critical role of short-interval data collection for accurate predictive modeling.

EMD combined with ensemble of machine learning predictors for foreign exchange rate forecasting

Forecasting foreign exchange rates is a critical financial challenge. In this paper, we build on recent trends and address the limitations of prior research by proposing a novel approach. Our method combines empirical mode decomposition (EMD) with ensemble of machine learning predictors in foreign exchange rate forecasting. To demonstrate that our proposed method (called EMD-ML) is effective, we used the new approach to forecast six foreign exchange rate time series at a specific time. The first experiment was implemented to compare the proposed forecasting model EMD–LSTM, which combines empirical mode decomposition (EMD) with ensemble of Long Short-Term Memory (LSTM) models, and the single LSTM model. The results indicate that the proposed EMD–LSTM model is more effective than the single LSTM. Besides, to aim at comparing deep-learning models against shallow machine learning models in combination with the EMD decomposition, the second experiment compared EMD-LSTM with the approach which combines EMD with an ensemble of k-nearest neighbors’ predictors (called EMD-KNN method) and the results of the second experiment show that EMD-LSTM cannot outperform EMD-KNN in foreign exchange rates forecasting.