Green Roof Hydrological Modelling With GRU And LSTM Networks

Abstract Green Roofs (GRs) are becoming more popular as a low-impact building option. They have the potential to minimize peak stormwater runoff while also increasing the quality of runoff from buildings. Improvement of hydrological models for the simulation of GRs will aid design of individual roofs as well as city scale planning that relies on the predicted impacts of widespread GR implementation. Machine learning (ML) has exploded in popularity in recent years, owing to considerable increases in processing power and data availability. However, there are no studies focusing on the use of ML in hydrological simulation of GRs. We focus on two types of ML-based model: long short-term memory (LSTM) and gated recurrent unit (GRU) in modelling hydrological performance of GRs, with sequence input and a single output (SISO), and synced sequence input and output (SSIO) architectures. According to the results of this paper, LSTM and GRU are useful tools for the modelling of GRs. As the time window length (memory length, time step length of input data) increases, SISO appears to have a higher overall forecast accuracy. SSIO delivers the best overall performance, when the SSIO is close to, or even exceeds, the maximum window size.

Download Full-text

Genetic Algorithm-Optimized Long Short-Term Memory Network for Stock Market Prediction

Sustainability ◽

10.3390/su10103765 ◽

2018 ◽

Vol 10 (10) ◽

pp. 3765 ◽

Cited By ~ 27

Author(s):

Hyejung Chung ◽

Kyung-shik Shin

Keyword(s):

Genetic Algorithm ◽

Stock Market ◽

Short Term Memory ◽

Time Window ◽

Hybrid Approach ◽

Window Size ◽

Short Term ◽

Term Memory ◽

Long Short Term Memory ◽

Lstm Network

With recent advances in computing technology, massive amounts of data and information are being constantly accumulated. Especially in the field of finance, we have great opportunities to create useful insights by analyzing that information, because the financial market produces a tremendous amount of real-time data, including transaction records. Accordingly, this study intends to develop a novel stock market prediction model using the available financial data. We adopt deep learning technique because of its excellent learning ability from the massive dataset. In this study, we propose a hybrid approach integrating long short-term memory (LSTM) network and genetic algorithm (GA). Heretofore, trial and error based on heuristics is commonly used to estimate the time window size and architectural factors of LSTM network. This research investigates the temporal property of stock market data by suggesting a systematic method to determine the time window size and topology for the LSTM network using GA. To evaluate the proposed hybrid approach, we have chosen daily Korea Stock Price Index (KOSPI) data. The experimental result demonstrates that the hybrid model of LSTM network and GA outperforms the benchmark model.

Download Full-text

Active Moving Area Identification using Machine Learning. Case study: Ometepe Island, Nicaragua

10.5194/egusphere-egu21-8462 ◽

2021 ◽

Author(s):

Víctor Arcia ◽

Gerald Corzo ◽

Heyddy Calderón

Keyword(s):

Machine Learning ◽

Remote Sensing ◽

Short Term Memory ◽

Time Window ◽

Mass Movement ◽

Spatiotemporal Analysis ◽

Machine Learning Techniques ◽

Precipitation Forecast ◽

Mass Movements ◽

Time Step

This study aims to propose the use of spatio-temporal Remote Sensing information and Machine learning techniques (ML) for Active Moving Area Identification and Forecast. Mass Movements are frequent in Central American countries, mainly due to the combined extreme hydro-meteorological events with the seismic activity and the characteristics of the geological formations in the region. Ometepe Island is located in Lake Cocibolca, Nicaragua; it has two volcanoes (one active) and Mass Movements happen quite often in the area, where many of them represent a big risk for the population. The triggering factors for these Mass Movements are mainly volcanic activity in conjunction with high and quick precipitations. The process of identification of a Mass Movement from Remote Sensing images is used first as a way to characterise the data, and then a lagged time step was used to evaluate the forecasting capabilities in a time window of precipitation forecast. For this, Remote Sensing was used to create the Active Moving Area Inventory, using InSAR technique with Sentinel-1 SAR images. SNAP software was used to locate occurrences of displacements in the island. This inventory was used to develop ML models that had Rainfall and Soil Moisture as dynamic variables; and DEM, Land Use, Geomorphology, and others as static variables. These were trained and evaluated using Logistic Regression (LR), Random Forest (RF) and Long Short-Term Memory (LSTM) to detect occurrence of Displacement in a particular area of the island. The results were analysed performance-wise and compared to each other. The results of this methodology are a first step into a larger framework of spatiotemporal analysis for forecasting using Machine Learning.

Download Full-text

Comparison of Long Short Term Memory Networks and the Hydrological Model in Runoff Simulation

Water ◽

10.3390/w12010175 ◽

2020 ◽

Vol 12 (1) ◽

pp. 175 ◽

Cited By ~ 8

Author(s):

Hongxiang Fan ◽

Mingliang Jiang ◽

Ligang Xu ◽

Hua Zhu ◽

Junxiang Cheng ◽

...

Keyword(s):

Short Term Memory ◽

Window Size ◽

Test Period ◽

Data Driven ◽

Meteorological Variables ◽

Short Term ◽

Time Step ◽

Term Memory ◽

Runoff Modeling ◽

Long Short Term Memory

Runoff modeling is one of the key challenges in the field of hydrology. Various approaches exist, ranging from physically based over conceptual to fully data driven models. In this paper, we propose a data driven approach using the state-of-the-art Long-Short-Term-Memory (LSTM) network. The proposed model was applied in the Poyang Lake Basin (PYLB) and its performance was compared with an Artificial Neural Network (ANN) and the Soil & Water Assessment Tool (SWAT). We first tested the impacts of the number of previous time step (window size) in simulation accuracy. Results showed that a window in improper large size will dramatically deteriorate the model performance. In terms of PYLB, a window size of 15 days might be appropriate for both accuracy and computational efficiency. We then trained the model with 2 different input datasets, namely, dataset with precipitation only and dataset with all available meteorological variables. Results demonstrate that although LSTM with precipitation data as the only input can achieve desirable results (where the NSE ranged from 0.60 to 0.92 for the test period), the performance can be improved simply by feeding the model with more meteorological variables (where NSE ranged from 0.74 to 0.94 for the test period). Moreover, the comparison results with the ANN and the SWAT showed that the ANN can get comparable performance with the SWAT in most cases whereas the performance of LSTM is much better. The results of this study underline the potential of the LSTM for runoff modeling especially for areas where detailed topographical data are not available.

Download Full-text

Lightweight Driver Behavior Identification Model with Sparse Learning on In-Vehicle CAN-BUS Sensor Data

Sensors ◽

10.3390/s20185030 ◽

2020 ◽

Vol 20 (18) ◽

pp. 5030 ◽

Cited By ~ 1

Author(s):

Shan Ullah ◽

Deok-Hwan Kim

Keyword(s):

Embedded System ◽

Performance Metrics ◽

Short Term Memory ◽

Can Bus ◽

Driver Behavior ◽

Window Size ◽

Sensor Data ◽

Sparse Learning ◽

Area Network ◽

Time Step

This study focuses on driver-behavior identification and its application to finding embedded solutions in a connected car environment. We present a lightweight, end-to-end deep-learning framework for performing driver-behavior identification using in-vehicle controller area network (CAN-BUS) sensor data. The proposed method outperforms the state-of-the-art driver-behavior profiling models. Particularly, it exhibits significantly reduced computations (i.e., reduced numbers both of floating-point operations and parameters), more efficient memory usage (compact model size), and less inference time. The proposed architecture features depth-wise convolution, along with augmented recurrent neural networks (long short-term memory or gated recurrent unit), for time-series classification. The minimum time-step length (window size) required in the proposed method is significantly lower than that required by recent algorithms. We compared our results with compressed versions of existing models by applying efficient channel pruning on several layers of current models. Furthermore, our network can adapt to new classes using sparse-learning techniques, that is, by freezing relatively strong nodes at the fully connected layer for the existing classes and improving the weaker nodes by retraining them using data regarding the new classes. We successfully deploy the proposed method in a container environment using NVIDIA Docker in an embedded system (Xavier, TX2, and Nano) and comprehensively evaluate it with regard to numerous performance metrics.

Download Full-text

Time Window from Visual Images to Visual Short-Term Memory: Consolidation or Integration?

Experimental Psychology (formerly Zeitschrift für Experimentelle Psychologie) ◽

10.1027/1618-3169.51.1.45 ◽

2004 ◽

Vol 51 (1) ◽

pp. 45-51 ◽

Cited By ~ 21

Author(s):

Yuhong Jiang

Keyword(s):

Short Term Memory ◽

Time Window ◽

Short Interval ◽

Visual Images ◽

Short Term ◽

Temporal Window ◽

Term Memory ◽

Visual Short Term Memory ◽

Probe Task ◽

Long Time

Abstract. When two dot arrays are briefly presented, separated by a short interval of time, visual short-term memory of the first array is disrupted if the interval between arrays is shorter than 1300-1500 ms ( Brockmole, Wang, & Irwin, 2002 ). Here we investigated whether such a time window was triggered by the necessity to integrate arrays. Using a probe task we removed the need for integration but retained the requirement to represent the images. We found that a long time window was needed for performance to reach asymptote even when integration across images was not required. Furthermore, such window was lengthened if subjects had to remember the locations of the second array, but not if they only conducted a visual search among it. We suggest that a temporal window is required for consolidation of the first array, which is vulnerable to disruption by subsequent images that also need to be memorized.

Download Full-text

Human Activity Recognition using Fourier Transform Inspired Deep Learning Combination Model

International Journal of Sensors Wireless Communications and Control ◽

10.2174/2210327908666180727123657 ◽

2019 ◽

Vol 9 (1) ◽

pp. 16-31

Author(s):

Kyungkoo Jun

Keyword(s):

Fourier Transform ◽

Deep Learning ◽

Short Term Memory ◽

Window Size ◽

Sensor Data ◽

Data Sets ◽

Data Set ◽

Proposed Model ◽

Testing Data ◽

Labeling Scheme

Background & Objective: This paper proposes a Fourier transform inspired method to classify human activities from time series sensor data. Methods: Our method begins by decomposing 1D input signal into 2D patterns, which is motivated by the Fourier conversion. The decomposition is helped by Long Short-Term Memory (LSTM) which captures the temporal dependency from the signal and then produces encoded sequences. The sequences, once arranged into the 2D array, can represent the fingerprints of the signals. The benefit of such transformation is that we can exploit the recent advances of the deep learning models for the image classification such as Convolutional Neural Network (CNN). Results: The proposed model, as a result, is the combination of LSTM and CNN. We evaluate the model over two data sets. For the first data set, which is more standardized than the other, our model outperforms previous works or at least equal. In the case of the second data set, we devise the schemes to generate training and testing data by changing the parameters of the window size, the sliding size, and the labeling scheme. Conclusion: The evaluation results show that the accuracy is over 95% for some cases. We also analyze the effect of the parameters on the performance.

Download Full-text

Surrounding Vehicles’ Contribution to Car-Following Models: Deep-Learning-Based Analysis

Transportation Research Record Journal of the Transportation Research Board ◽

10.1177/03611981211018693 ◽

2021 ◽

pp. 036119812110186

Author(s):

Saeed Vasebi ◽

Yeganeh M. Hayeri ◽

Peter J. Jin

Keyword(s):

Deep Learning ◽

Traffic Flow ◽

Short Term Memory ◽

Data Availability ◽

Car Following ◽

Preceding Vehicle ◽

Long Short Term Memory ◽

The Right ◽

Mean Square Errors ◽

Deep Learning Model

Relatively recent increased computational power and extensive traffic data availability have provided a unique opportunity to re-investigate drivers’ car-following (CF) behavior. Classic CF models assume drivers’ behavior is only influenced by their preceding vehicle. Recent studies have indicated that considering surrounding vehicles’ information (e.g., multiple preceding vehicles) could affect CF models’ performance. An in-depth investigation of surrounding vehicles’ contribution to CF modeling performance has not been reported in the literature. This study uses a deep-learning model with long short-term memory (LSTM) to investigate to what extent considering surrounding vehicles could improve CF models’ performance. This investigation helps to select the right inputs for traffic flow modeling. Five CF models are compared in this study (i.e., classic, multi-anticipative, adjacent-lanes, following-vehicle, and all-surrounding-vehicles CF models). Performance of the CF models is compared in relation to accuracy, stability, and smoothness of traffic flow. The CF models are trained, validated, and tested by a large publicly available dataset. The average mean square errors (MSEs) for the classic, multi-anticipative, adjacent-lanes, following-vehicle, and all-surrounding-vehicles CF models are 1.58 × 10−3, 1.54 × 10−3, 1.56 × 10−3, 1.61 × 10−3, and 1.73 × 10−3, respectively. However, the results show insignificant performance differences between the classic CF model and multi-anticipative model or adjacent-lanes model in relation to accuracy, stability, or smoothness. The following-vehicle CF model shows similar performance to the multi-anticipative model. The all-surrounding-vehicles CF model has underperformed all the other models.

Download Full-text

Regional Associations of Cortical Thickness With Gait Variability: The Tasmanian Study of Cognition and Gait

Innovation in Aging ◽

10.1093/geroni/igaa057.750 ◽

2020 ◽

Vol 4 (Supplement_1) ◽

pp. 232-233

Author(s):

Oshadi Jayakody ◽

Monique Breslin ◽

Richard Beare ◽

Velandai Srikanth ◽

Helena Blumen ◽

...

Keyword(s):

Cortical Thickness ◽

Step Length ◽

Gait Variability ◽

Thickness Ratio ◽

Sensory Function ◽

Time Variability ◽

Step Width ◽

Time Step ◽

Cortical Regions ◽

Regional Associations

Abstract Gait variability is a marker of cognitive decline. However, there is limited understanding of the cortical regions associated with gait variability. We examined associations between regional cortical thickness and gait variability in a population-based sample of older people without dementia. Participants (n=350, mean age 71.9±7.1) were randomly selected from the electoral roll. Variability in step time, step length, step width and double support time (DST) were calculated as the standard deviation of each measure, obtained from the GAITRite walkway. MRI scans were processed through FreeSurfer to obtain cortical thickness of 68 regions. Bayesian regression was used to determine regional associations of mean cortical thickness and thickness ratio (regional thickness/overall mean thickness) with gait variability. Smaller overall cortical thickness was only associated with greater step width and step time variability. Smaller mean thickness in widespread regions important for sensory, cognitive and motor functions were associated with greater step width and step time variability. In contrast, smaller thickness in a few frontal and temporal regions were associated with DST variability and the right cuneus was associated with step length variability. Smaller thickness ratio in frontal and temporal regions important for motor planning, execution and sensory function and, greater thickness ratio in the anterior cingulate was associated with greater variability in all measures. Examining individual cortical regions is important in understanding the relationship between gray matter and gait variability. Cortical thickness ratio highlights that smaller regional thickness relative to global thickness may be important for the consistency of gait.

Download Full-text

Near Real-Time Global Solar Radiation Forecasting at Multiple Time-Step Horizons Using the Long Short-Term Memory Network

Energies ◽

10.3390/en13143517 ◽

2020 ◽

Vol 13 (14) ◽

pp. 3517 ◽

Cited By ~ 1

Author(s):

Anh Ngoc-Lan Huynh ◽

Ravinesh C. Deo ◽

Duc-Anh An-Vo ◽

Mumtaz Ali ◽

Nawin Raj ◽

...

Keyword(s):

Deep Learning ◽

Solar Radiation ◽

Short Term Memory ◽

Global Solar Radiation ◽

Multiple Time ◽

Short Term ◽

Time Step ◽

Term Memory ◽

Multiple Time Step ◽

Long Short Term Memory

This paper aims to develop the long short-term memory (LSTM) network modelling strategy based on deep learning principles, tailored for the very short-term, near-real-time global solar radiation (GSR) forecasting. To build the prescribed LSTM model, the partial autocorrelation function is applied to the high resolution, 1 min scaled solar radiation dataset that generates statistically significant lagged predictor variables describing the antecedent behaviour of GSR. The LSTM algorithm is adopted to capture the short- and the long-term dependencies within the GSR data series patterns to accurately predict the future GSR at 1, 5, 10, 15, and 30 min forecasting horizons. This objective model is benchmarked at a solar energy resource rich study site (Bac-Ninh, Vietnam) against the competing counterpart methods employing other deep learning, a statistical model, a single hidden layer and a machine learning-based model. The LSTM model generates satisfactory predictions at multiple-time step horizons, achieving a correlation coefficient exceeding 0.90, outperforming all of the counterparts. In accordance with robust statistical metrics and visual analysis of all tested data, the study ascertains the practicality of the proposed LSTM approach to generate reliable GSR forecasts. The Diebold–Mariano statistic test also shows LSTM outperforms the counterparts in most cases. The study confirms the practical utility of LSTM in renewable energy studies, and broadly in energy-monitoring devices tailored for other energy variables (e.g., hydro and wind energy).

Download Full-text

Impact of climate change on water availability in Tropical Andean catchments: a case study in the Vilcanota catchment, Peru

10.5194/egusphere-egu21-10292 ◽

2021 ◽

Author(s):

Selina Meier ◽

Randy Munoz ◽

Christian Huggel

Keyword(s):

Climate Change ◽

Water Management ◽

Water Demand ◽

Climate Models ◽

Local Population ◽

Time Step ◽

Glacier Surface ◽

Hydrological Simulation ◽

Sustainable Water Management ◽

Monthly Runoff

Water scarcity is increasingly becoming a problem in many regions of the world. On the one hand, this can be attributed to changes in precipitation conditions due to climate change. On the other hand, this is also due to population growth and changes in consumer behaviour. In this study, an analysis is carried out for the highly glaciated Vilcanota River catchment (9808 km2 &#8211; 1.2% glacier area) in the Cusco region (Peru). Possible climatic and socioeconomic scenarios up to 2050 were developed including the interests from different water sectors, i.e. agriculture, domestic and energy.The analysis consists of the hydrological simulation at a monthly time step from September 2043 to August 2050 using a simple glacio-hydrological model. For historic conditions (1990 to 2006) a combination of gridded data (PISCO precipitation) and weather stations was used. Future scenario simulations were based on three different climate models for both RCP 2.6 and 8.5. Different glacier outlines were used to simulate changes in glacier surface through the time for both historic (from satellite data) and future (from existing literature) scenarios. Furthermore, future water demand simulations were based on the SSP1 and SSP3 scenarios.Results from all scenarios suggest an average monthly runoff of about 130 m3/s for the Vilcanota catchment between 2043 and 2050. This represents a change of about +5% compared to the historical monthly runoff of about 123 m3/s. The reason for the increase in runoff is related to the precipitation data from the selected climate models. However, an average monthly deficit of up to 50 m3/s was estimated between April and November with a peak in September. The seasonal deficit is related to the seasonal change in precipitation, while the water demand seems to have a less important influence.Due to the great uncertainty of the modelling and changes in the socioeconomic situation, the data should be continuously updated. In order to construct a locally sustainable water management system, the modelling needs to be further downscaled to the different subcatchments in the Vilcanota catchment. To address the projected water deficit, a new dam could partially compensate for the decreasing storage capacity of the melting glaciers. However, the construction of the dam could meet resistance from the local population if they cannot be promised and communicated multiple uses of the new dam. Sustainable water management requires the cooperation of all stakeholders and all stakeholders should be able to benefit from it so that they will support future projects.

Download Full-text