Convolutional Long Short-term Memory Model Research Articles

Convolutional Long Short-Term Memory (ConvLSTM)-Based Prediction of Voltage Stability in a Microgrid

The maintenance of an uninterrupted electricity supply to meet demand is of paramount importance for maintaining the stable operation of an electrical power system. Machine learning and deep learning play a crucial role in maintaining that stable operation. These algorithms have the ability to acquire knowledge from past data, enabling them to efficiently identify and forecast potential scenarios of instability in the future. This work presents a hybrid convolutional long short-term memory (ConvLSTM) technique for training and predicting nodal voltage stability in an IEEE 14-bus microgrid. Analysis of the findings shows that the suggested ConvLSTM model exhibits the highest level of precision, reaching a value of 97.65%. Furthermore, the ConvLSTM model has been shown to perform better compared to alternative machine learning and deep learning models such as convolutional neural networks, k-nearest neighbors, and support vector machine models, specifically in terms of accurately forecasting voltage stability. The IEEE 14-bus system tests indicate that the suggested method can quickly and accurately determine the stability status of the system. The comparative analysis obtained the results and further justified the efficiency and voltage stability of the proposed model.

Incorporating physical constraints in a deep learning framework for short-term daily prediction of sea ice concentration

Grasping the rapidly changing sea ice patterns has become the key to Arctic navigation. Starting from meteorology and oceanography, this study combines the empirical orthogonal function (EOF) analysis with convolutional long short-term memory (ConvLSTM) model to realize short-term sea ice prediction. To further improve the effectiveness and predictability of the ConvLSTM model, we adopt the EOF analysis to incorporate the spatial and temporal patterns of sea ice changes into the ConvLSTM model in three ways, namely, model input factors, model internal network structure, and model loss function. Through the mean absolute error (MAE), anomaly correlation coefficient (ACC), and anomaly metrics, we find that embedding the EOF physical constraints into the model's internal structure fully employs the spatial pattern and time series information, affording an optimal prediction effect. Additionally, the impact of the weight proportion between the physical constraint and the base loss and the effect of combining the three ways are discussed. The experimental results reveal that tuning the weight proportion in the loss function can improve the model's training effect. However, multiple combinations of the same physical law make it difficult to introduce new effective information.

An Improved Method for Retrieving Subsurface Temperature Using the ConvLSTM Model in the Western Pacific Ocean

In the era of marine big data, making full use of multi-source satellite observations to accurately retrieve and predict the temperature structure of the ocean subsurface layer is very significant in advancing the understanding of oceanic processes and their dynamics. Considering the time dependence and spatial correlation of marine characteristics, this study employed the convolutional long short-term memory (ConvLSTM) method to retrieve the subsurface temperature in the Western Pacific Ocean from several types of satellite observations. Furthermore, considering the temperature’s vertical distribution, the retrieved results for the upper layer were iteratively used in the calculation for the deeper layer as input data to improve the algorithm. The results show that the retrieved results for the 100 to 500 m depth temperature using the 50 m layer in the calculation resulted in higher accuracy than those retrieved from the standard ConvLSTM method. The largest improvement was in the calculation for the 100 m layer, where the thermocline was located. The results indicate that our improved ConvLSTM method can increase the accuracy of subsurface temperature retrieval without additional input data.

Anticipating a risky future: long short-term memory (LSTM) models for spatiotemporal extrapolation of population data in areas prone to earthquakes and tsunamis in Lima, Peru

Abstract. In this paper, we anticipate geospatial population distributions to quantify the future number of people living in earthquake-prone and tsunami-prone areas of Lima and Callao, Peru. We capitalize upon existing gridded population time series data sets, which are provided on an open-source basis globally, and implement machine learning models tailored for time series analysis, i.e., based on long short-term memory (LSTM) networks, for prediction of future time steps. Specifically, we harvest WorldPop population data and teach LSTM and convolutional LSTM models equipped with both unidirectional and bidirectional learning mechanisms, which are derived from different feature sets, i.e., driving factors. To gain insights regarding the competitive performance of LSTM-based models in this application context, we also implement multilinear regression and random forest models for comparison. The results clearly underline the value of the LSTM-based models for forecasting gridded population data; the most accurate prediction obtained with an LSTM equipped with a bidirectional learning scheme features a root-mean-squared error of 3.63 people per 100 × 100 m grid cell while maintaining an excellent model fit (R2= 0.995). We deploy this model for anticipation of population along a 3-year interval until the year 2035. Especially in areas of high peak ground acceleration of 207–210 cm s−2, the population is anticipated to experience growth of almost 30 % over the forecasted time span, which simultaneously corresponds to 70 % of the predicted additional inhabitants of Lima. The population in the tsunami inundation area is anticipated to grow by 61 % until 2035, which is substantially more than the average growth of 35 % for the city. Uncovering those relations can help urban planners and policymakers to develop effective risk mitigation strategies.

Multi-Class Recognition of Soybean Leaf Diseases using a Conv-LSTM Model

This research presents an innovative approach for multi-class recognition of soybean leaf diseases using a Convolutional Long Short-Term Memory (Conv-LSTM) model. The model integrates the spatial learning capabilities of convolutional layers with the temporal dependencies of LSTM units, addressing the critical need for accurate disease detection in agriculture, particularly in soybean cultivation where leaf diseases significantly impact crop yield and quality. Through comparative experiments with established deep learning models such as AlexNet, VGG16, and ResNet50, the Conv-LSTM model demonstrates superior performance in terms of accuracy, precision, recall, and F1 score. By effectively capturing both spatial and temporal features in soybean leaf images, the Conv-LSTM model showcases its potential to enhance disease detection accuracy, supporting precision agriculture practices and enabling timely interventions to mitigate crop losses caused by diseases.

Convolutional LSTM model for cine image prediction of abdominal motion.

Objective.In this study, we tackle the challenge of latency in magnetic resonance linear accelerator (MR-Linac) systems, which compromises target coverage accuracy in gated real-time radiotherapy. Our focus is on enhancing motion prediction precision in abdominal organs to address this issue. We developed a convolutional long short-term memory (convLSTM) model, utilizing 2D cine magnetic resonance (cine-MR) imaging for this purpose.Approach.Our model, featuring a sequence-to-one architecture with six input frames and one output frame, employs structural similarity index measure (SSIM) as loss function. Data was gathered from 17 cine-MRI datasets using the Philips Ingenia MR-sim system and an Elekta Unity MR-Linac equivalent sequence, focusing on regions of interest (ROIs) like the stomach, liver, pancreas, and kidney. The datasets varied in duration from 1 to 10 min.Main results.The study comprised three main phases: hyperparameter optimization, individual training, and transfer learning with or without fine-tuning. Hyperparameters were initially optimized to construct the most effective model. Then, the model was individually applied to each dataset to predict images four frames ahead (1.24-3.28 s). We evaluated the model's performance using metrics such as SSIM, normalized mean square error, normalized correlation coefficient, and peak signal-to-noise ratio, specifically for ROIs with target motion. The average SSIM values achieved were 0.54, 0.64, 0.77, and 0.66 for the stomach, liver, kidney, and pancreas, respectively. In the transfer learning phase with fine-tuning, the model showed improved SSIM values of 0.69 for the liver and 0.78 for the kidney, compared to 0.64 and 0.37 without fine-tuning.Significance. The study's significant contribution is demonstrating the convLSTM model's ability to accurately predict motion for multiple abdominal organs using a Unity-equivalent MR sequence. This advancement is key in mitigating latency issues in MR-Linac radiotherapy, potentially improving the precision and effectiveness of real-time treatment for abdominal cancers.

Dissolved Oxygen Forecasting for Lake Erie’s Central Basin Using Hybrid Long Short-Term Memory and Gated Recurrent Unit Networks

Dissolved oxygen (DO) concentration is a pivotal determinant of water quality in freshwater lake ecosystems. However, rapid population growth and discharge of polluted wastewater, urban stormwater runoff, and agricultural non-point source pollution runoff have triggered a significant decline in DO levels in Lake Erie and other freshwater lakes located in populated temperate regions of the globe. Over eleven million people rely on Lake Erie, which has been adversely impacted by anthropogenic stressors resulting in deficient DO concentrations near the bottom of Lake Erie’s Central Basin for extended periods. In the past, hybrid long short-term memory (LSTM) models have been successfully used for the time-series forecasting of water quality in rivers and ponds. However, the prediction errors tend to grow significantly with the forecasting period. Therefore, this research aimed to improve the accuracy of DO forecasting models by taking advantage of Lake Erie’s real-time water quality (water temperature and DO concentration) monitoring network to establish temporal and spatial links between adjacent monitoring stations. We developed hybrid LSTM models that combine LSTM, convolutional neuron network LSTM (CNN-LSTM), hybrid CNN with gated recurrent unit (CNN-GRU) models, and convolutional LSTM (ConvLSTM) to forecast near-bottom DO concentrations in Lake Erie’s Central Basin. These hybrid LSTM models improve their capacity to handle complicated datasets with spatial and temporal variability. These models can serve as accurate and reliable tools for forecasting DO concentrations in freshwater lakes to help environmental protection agencies better access and manage the health of these vital ecosystems. Following analysis of a 21-site Lake Erie dataset for 2020 and 2021, the ConvLSTM model emerged as the most accurate and reliable, boasting an MSE of 0.51 mg/L, MAE of 0.42 mg/L, and an R-squared of 0.95 over the 12 h prediction range. The model foresees future hypoxia in Lake Erie. Notably, the temperature near site 713 holds significance for Central Basin DO forecasting in Lake Erie, as indicated by outcomes derived from the Shapley additive explanations (SHAP).

Statistical downscaling of high-resolution precipitation in India using convolutional long short-term memory networks

Abstract Statistical downscaling of the General Circulation Model (GCM) simulations are widely used for accessing climate changes in the future at different spatiotemporal scales. This study proposes a novel Statistical Downscaling (SD) model established on the Convolutional Long Short-Term Memory (ConvLSTM) Network. The methodology is applied to obtain future projection of rainfall at 0.25° spatial resolution over the Indian sub-continental region. The traditional multisite downscaling models typically perform downscaling on a single homogeneous rainfall zone, predicting rainfall at only one grid point in a single model run. The proposed model captures spatiotemporal dependencies in multisite local rainfall and predicts rainfall for the entire zone in a single model run. The study proposes a Shared ConvLSTM model providing a single end-to-end supervised model for predicting the future precipitation for entire India. The model captures the regional variability in rainfall better than a region-wise trained model. The projected future rainfall for different scenarios of climate change reveals an overall increase in the rainfall mean and spatially non-uniform changes in future rainfall extremes over India. The results highlight the importance of conducting in-depth hydrologic studies for different river basins of the country for future water availability assessment and making water resource policies.

Decoding sentiment: A sentiment analysis model for movie reviews

Sentiment analysis of movie reviews can provide valuable insights into movie reactions and preferences. To this end, this study proposes the Convolutional Long Short-Term Memory (ConvLSTM) neural network for movie review sentiment analysis. ConvLSTM can efficiently capture sequential information due to its recurrent neural network characteristics. Specifically, the movie review data are first tokenized. Next, the ConvLSTM analysis model is constructed additionally by fine-tuning its parameters to optimize the performance. The ConvLSTM model consists of multiple storage units that retain contextual information, enabling the model to identify long-distance dependencies in the text. The network is trained using a combination of positive and negative movie reviews, and the training process involves adjusting the model weights to minimize the classification error. Experimental results demonstrate the effectiveness of the proposed method in accurately predicting movie review sentiment. It outperforms traditional machine learning methods in sentiment analysis tasks. The findings demonstrate the potential of LSTM-based sentiment analysis in various applications such as movie recommendation systems and market research. This study's findings help advance the development of sentiment analysis techniques and are of great relevance in understanding and catering to audience preferences in the movie industry.

A Hybrid Data‐Driven and Data Assimilation Method for Spatiotemporal Forecasting: PM2.5 Forecasting in China

AbstractSpatiotemporal forecasting involves generating temporal forecasts for system state variables across spatial regions. Data‐driven methods such as Convolutional Long Short‐Term Memory (ConvLSTM) are effective in capturing both spatial and temporal correlations, but they suffer from error accumulation and accuracy loss as forecasting time increases due to the nonlinearity and uncertainty in physical processes. To address this issue, we propose to combine data‐driven and data assimilation (DA) methods for spatiotemporal forecasting. The accuracy of the data‐driven ConvLSTM model can be improved by periodically assimilating real‐time observations using the ensemble Kalman filter (EnKF) approach. This proposed hybrid ConvLSTM‐EnKF method is demonstrated through PM2.5 forecasting in China, which is a challenging task due to the complexity of topographical and meteorological conditions in the region, the need for high‐resolution forecasting over a large study area, and the scarcity of observations. The results show that the ConvLSTM‐EnKF method outperforms conventional methods and can provide satisfactory operational PM2.5 forecasts for up to 1 month with spatially averaged RMSE below 20 μg/m3 and correlation coefficient (R) above 0.8. In addition, the ConvLSTM‐EnKF method shows a substantial reduction in CPU time when compared to the commonly used NAQPMS‐EnKF method, up to three orders of magnitude. Overall, the use of data‐driven models provides efficient forecasts and speeds up DA. This hybrid ConvLSTM‐EnKF is a novel operational forecasting technique for spatiotemporal forecasting and is used in real spatiotemporal forecasting for the first time.

Forecasting solar irradiation using convolutional long short-term memory and feature selection of data from neighboring locations

The capacity of renewable energy systems, such as photovoltaic farms, has been steadily increasing owing to the continuous depletion of fossil-based energy sources. Information that can be used to optimize the use of renewable energy is required. This work proposes a novel method that uses data from neighboring locations to perform 24 h-ahead solar GHI forecasting. The approach is helpful in supporting the operation of photovoltaic farms, minimizing the need to collect new data on the target location. This work proposes the Convolutional Long Short-term Memory (ConvLSTM) model to maximize the performance using a feature-rich spatiotemporal dataset. The ConvLSTM model uses Hadamard convolutions in its input-to-state transition and state-to-state transition, allowing it to retain better information. The feature selection methods are used to ensure that only the critical subsets of the dataset are used to develop the model. The subsets identified by the feature selection methods are used on Vector Autoregressive, Support Vector Regression, Multilayer Perceptron, Convolutional Neural Network (CNN), CNN-LSTM and ConvLSTM networks. The winning method is compared to related studies using Stacked-LSTM with transfer-learning, attention-based LSTM and Krill-Herd Algorithm – Support vector Regression (KHA-SVR). The resulting coefficients of determination and root-mean-squared errors were statistically evaluated and compared. Ultimately, the proposed ConvLSTM model performed consistently better across all the seasons and feature subsets.

A maps-to-maps approach for simulating urban land expansion based on convolutional long short-term memory neural networks

Cellular automata (CA) have been prevalently used for the simulation of urban land change. However, how to effectively learn the spatial-temporal dynamics of urban development from time-series data remain an important challenge for CA-based models. To address this issue, we propose a new model for the simulation of urban development based on convolutional long short-term memory (ConvLSTM) neural networks. The core of the proposed model is a sequence of vanilla ConvLSTM cells integrated with the modules of channel attention and contextual embedding. Compared with conventional CA-based models, the proposed ConvLSTM model is more advanced in that it can better leverage the open access annual urban land maps to capture simultaneously the spatial structure and the temporal dependency of historical urban development, and further predict multiple maps of annual development for subsequent years (i.e., Maps-to-Maps). The performance of the ConvLSTM model is evaluated through the case studies in China’s three mega-urban regions, and ConvLSTM outperforms other state-of-the-art deep learning architectures at both the pixel level and the coarser grid level. The results also suggest the satisfactory transferability of ConvLSTM in that the model trained in one mega-urban region can be successfully re-used in others without fine tuning.

Edge-based detection and localization of adversarial oscillatory load attacks orchestrated by compromised EV charging stations

Recent reports indicate that electric vehicle charging stations (EVCSs) are susceptible to remote exploitation through their vulnerable software/cyber components. More importantly, compromised EVCS can be leveraged to perform coordinated oscillatory load attacks against the interconnected power grid, leading to power grid instability, increased operational costs, and power line tripping. In this paper, an edge-based approach for the detection and localization of coordinated oscillatory load attacks initiated by exploited EV charging stations against the power grid is investigated. The edge-based detection relies on the behavioral characteristics of the power grid in the presence of interconnected EVCS while combining cyber and physical layer features to implement deep learning algorithms for the effective detection of oscillatory load attacks at the EVCS. The proposed detection approach was evaluated by building a real-time test bed to synthesize benign and malicious data, which was generated by analyzing real-life EV charging data collected during recent years. The results demonstrate the effectiveness of the implemented approach with the Convolutional Long-Short Term Memory model producing optimal classification accuracy (99.4%). Moreover, this analysis results shed light on the impact of such detection mechanisms towards building resiliency into different levels of the EV charging ecosystem while allowing power grid operators to localize attacks and take further mitigation measures. Specifically, the detection mechanism of oscillatory load attacks is decentralized and created an effective alternative for operator-centric mechanisms to mitigate multi-operator and Man-in-the-Middle (MitM) oscillatory load attacks against the power grid. Additionally, the created test bed is leveraged to evaluate a distributed mitigation technique, which can be deployed on public/private charging stations to average out the impact of oscillatory load attacks while allowing the power system to recover smoothly within 1 second with minimal overhead. Finally, a security metric is createdto evaluate the resiliency of the ecosystem based on its design and architecture.

Deep Learning‐Based Regional Ionospheric Total Electron Content Prediction—Long Short‐Term Memory (LSTM) and Convolutional LSTM Approach

AbstractThis study evaluates the performance of deep learning approach in the prediction of the ionospheric total electron content (TEC) during magnetically quiet periods. Two deep learning techniques, long short‐term memory (LSTM) and convolutional LSTM (ConvLSTM), are employed to predict TEC values 24 hr ahead in the vicinity of the Korean Peninsula (26.5°–40°N, 121°–134.5°E). The LSTM method predicts TEC at a single point based on time series of data at that point, whereas the ConvLSTM method simultaneously predicts TEC values at multiple points using spatiotemporal distribution of TEC. Both the LSTM and ConvLSTM models are trained using the complete regional TEC maps reconstructed by applying the Deep Convolutional Generative Adversarial Network–Poisson Blending (DCGAN‐PB) method to observed TEC data. The training period spans from 2002 to 2018, and the model performance is evaluated using 2019 data. Our results show that the ConvLSTM method outperforms the LSTM method, generating more reliable TEC maps with smaller root mean square errors when compared to the ground truth (DCGAN‐PB TEC maps). This outcome indicates that deep learning models can improve the prediction accuracy of TEC at a specific point by taking into account spatial information of TEC. We conclude that ConvLSTM is a reliable and efficient approach for the prompt ionospheric prediction.

CSACL: A Channel Spatial Attention Convolutional LSTM Model for Short-Term Sea Surface Temperature Prediction

CSACL: A Channel Spatial Attention Convolutional LSTM Model for Short-Term Sea Surface Temperature Prediction

Daily-Scale Prediction of Arctic Sea Ice Concentration Based on Recurrent Neural Network Models

Arctic sea ice prediction is of great practical significance in facilitating Arctic route planning, optimizing fisheries management, and advancing the field of sea ice dynamics research. While various deep learning models have been developed for sea ice prediction, they predominantly operate at the seasonal or sub-seasonal scale, often focusing on localized areas, and few cater to full-region daily-scale prediction. This study introduces the use of spatiotemporal sequence data prediction models, namely, the convolutional LSTM (ConvLSTM) and predictive recurrent neural network (PredRNN), for the prediction of sea ice concentration (SIC). Our analysis reveals that, when solely utilizing SIC historical data as the input, the ConvLSTM model outperforms the PredRNN model in SIC prediction. To enhance the models’ capacity to capture spatiotemporal relationships between multiple variables, we expanded the range of input data types to form the ConvLSTM-multi and PredRNN-multi models. Experimental findings demonstrate that the prediction accuracy of the four models significantly surpasses the CMIP6 model in three prospective climate scenarios (SSP126, SSP245, and SSP585). Of the four models, the ConvLSTM-multi model excels in assimilating the influence of reanalysis data on sea ice within the sea ice edge region, thus exhibiting superior performance than the PredRNN-multi model in predicting daily Arctic SIC over the subsequent 10 days. Furthermore, sensitivity tests on various model parameters highlight the substantial impact of sea surface temperature and prediction date on the accuracy of daily sea ice prediction, and meteorological and oceanographic parameters primarily affect the prediction accuracy of the thin-ice region at the edge of the sea ice.

Proactive Service Caching in a MEC System by Using Spatio-Temporal Correlation among MEC Servers

Optimizingthe cache hit rate in a multi-access edge computing (MEC) system is essential in increasing the utility of a system. A pivotal challenge within this context lies in predicting the popularity of a service. However, accurately predicting popular services for each MEC server (MECS) is hindered by the dynamic nature of user preferences in both time and space, coupled with the necessity for real-time adaptability. In this paper, we address this challenge by employing the Convolutional Long Short-Term Memory (ConvLSTM) model, which can capture both temporal and spatial correlations inherent in service request patterns. Our proposed methodology leverages ConvLSTM for service popularity prediction by modeling the distribution of service popularity in a MEC system as a heatmap image. Additionally, we propose a procedure that predicts service popularity in each MECS through a sequence of heatmap images. Through simulation studies using real-world datasets, we compare the performance of our method with that of the LSTM-based method. In the LSTM-based method, each MECS predicts the service popularity independently. On the contrary, our method takes a holistic approach by considering spatio-temporal correlations among MECSs during prediction. As a result, our method increases the average cache hit rate by more than 6.97% compared to the LSTM-based method. From an implementation standpoint, our method requires only one ConvLSTM model while the LSTM-based method requires at least one LSTM model for each MECS. Thus, compared to the LSTM-based method, our method reduces the deep learning model parameters by 32.15%.

Direct integration of reservoirs' operations in a hydrological model for streamflow estimation: coupling a CLSTM model with MOHID-Land

Abstract. Knowledge about streamflow regimes and values is essential for different activities and situations in which justified decisions must be made. However, streamflow behavior is commonly assumed to be non-linear, being controlled by various mechanisms that act on different temporal and spatial scales, making its estimation challenging. An example is the construction and operation of infrastructures such as dams and reservoirs in rivers. The challenges faced by modelers to correctly describe the impact of dams on hydrological systems are considerable. In this study, an already implemented solution of the MOHID-Land (where MOHID stands for HYDrodinamic MOdel, or MOdelo HIDrodinâmico in Portuguese) model for a natural flow regime in the Ulla River basin was considered as a baseline. The watershed referred to includes three reservoirs. Outflow values were estimated considering a basic operation rule for two of them (run-of-the-river dams) and considering a data-driven model of a convolutional long short-term memory (CLSTM) type for the other (high-capacity dam). The outflow values obtained with the CLSTM model were imposed in the hydrological model, while the hydrological model fed the CLSTM model with the level and the inflow of the reservoir. This coupled system was evaluated daily using two hydrometric stations located downstream of the reservoirs, resulting in an improved performance compared with the baseline application. The analysis of the modeled values with and without reservoirs further demonstrated that considering dams' operations in the hydrological model resulted in an increase in the streamflow during the dry season and a decrease during the wet season but with no differences in the average streamflow. The coupled system is thus a promising solution for improving streamflow estimates in modified catchments.

An active learning convolutional neural network for predicting river flow in a human impacted system

The South Platte river system contains a mixture of natural streams, reservoirs, and pipeline projects that redirect water to front range communities in Colorado. At many timepoints, a simple persistence model is the best predictor for flow from pipelines and reservoirs but at other times, flows change based on snowmelt and inputs such as reservoir fill rates, local weather, and anticipated demand. Here we find that a convolutional Long Short-Term Memory (LSTM) network is well suited to modeling flow in parts of this basin that are strongly impacted by water projects as well as ones that are relatively free from direct human modifications. Furthermore, it is found that including an active learning component in which separate Convolutional Neural Networks (CNNs) are used to classify and then select the data that is then used for training a convolutional LSTM network is advantageous. Models specific for each gauge are created by transfer of parameter from a base model and these gauge-specific models are then fine-tuned based a curated subset of training data. The result is accurate predictions for both natural flow and human influenced flow using only past river flow, reservoir capacity, and historical temperature data. In 14 of the 16 gauges modeled, the error in the prediction is reduced when using the combination of on-the-fly classification by CNN followed by analysis by either a persistence or convolutional LSTM model. The methods designed here could be applied broadly to other basins and to other situations where multiple models are needed to fit data at different times and locations.

Stacked 1D Convolutional LSTM (sConvLSTM1D) Model for Effective Prediction of Sunspot Time Series

Stacked 1D Convolutional LSTM (sConvLSTM1D) Model for Effective Prediction of Sunspot Time Series