A Novel Hybrid Extreme Learning Machine Approach Improved by K Nearest Neighbor Method and Fireworks Algorithm for Flood Forecasting in Medium and Small Watershed of Loess Region

Ren, Juanhui; Ren, Bo; Zhang, Qiuwen; Zheng, Xiuqing

doi:10.3390/w11091848

Cited by 20 publications

(8 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Compared to other learning algorithms, such as back propagation (BP), ELM achieves swift learning and performs well in generation function processing [52][53][54][55]. Using ELM in various engineering science fields, such as feature selection [56], classification [57], and regression [51,58], has provided acceptable results.…”

Section: Introductionmentioning

confidence: 99%

A Non-Tuned Machine Learning Technique for Abutment Scour Depth in Clear Water Condition

Bonakdari

Moradi

Ebtehaj

et al. 2020

Water

View full text Add to dashboard Cite

Abutment scour is a complex three-dimensional phenomenon, which is one of the leading causes of marine structure damage. Structural integrity is potentially attainable through the precise estimation of local scour depth. Due to the high complexity of scouring hydrodynamics, existing regression-based relations cannot make accurate predictions. Therefore, this study presented a novel expansion of extreme learning machines (ELM) to predict abutment scour depth (ds) in clear water conditions. The model was built using the relative flow depth (h/L), excess abutment Froude number (Fe), abutment shape factor (Ks), and relative sediment size (d50/L). A wide range of experimental samples was collected from the literature, and data was utilized to develop the ELM model. The ELM model reliability was evaluated based on the estimation results and several statistical indices. According to the results, the sigmoid activation function (correlation coefficient, R = 0.97; root mean square error, RMSE = 0.162; mean absolute percentage error, MAPE = 7.69; and scatter index, SI = 0.088) performed the best compared with the hard limit, triangular bias, radial basis, and sine activation functions. Eleven input combinations were considered to investigate the impact of each dimensionless variable on the abutment scour depth. It was found that ds/L = f (Fe, h/L, d50/L, Ks) was the best ELM model, indicating that the dimensional analysis of the original data properly reflected the underlying physics of the problem. Also, the absence of one variable from this input combination resulted in a significant accuracy reduction. The results also demonstrated that the proposed ELM model significantly outperformed the regression-based equations derived from the literature. The ELM model presented a fundamental equation for abutment scours depth prediction. Based on the simulation results, it appeared the ELM model could be used effectively in practical engineering applications of predicting abutment scour depth. The estimated uncertainty of the developed ELM model was calculated and compared with the conventional and artificial intelligence-based models. The lowest uncertainty with a value of ±0.026 was found in the proposed model in comparison with ±0.50 as the best uncertainty of the other models.

show abstract

Section: Introductionmentioning

confidence: 99%

A Non-Tuned Machine Learning Technique for Abutment Scour Depth in Clear Water Condition

Bonakdari

Moradi

Ebtehaj

et al. 2020

Water

View full text Add to dashboard Cite

show abstract

“…Extreme Learning Machine (ELM) is an improved version of conventional AN models that can solve regression problems with reduced execution time. The ELM models demonstrated higher performance when compared with AN and Support Vector Machine (SVM) models (He et al 2014;Ren et al 2019). SVM is often used for regression and clustering purposes, respectively, when limited data are provided from the area of study.…”

Section: Ai-based Methods For River-flow Forecastingmentioning

confidence: 99%

Modelling river flow in cold and ungauged regions: a review of the purposes, methods, and challenges

et al. 2022

View full text Add to dashboard Cite

River flow forecasting models assist in the understanding, predicting, monitoring, and managing of issues related to surface-water resources, such as water quality deterioration and flooding, or developing adaptation strategies to cope with climate change and increasing water demand. This review presents an overview of the current research status and progress in river-flow forecasting, focusing on cold climates and ungauged locations. River-flow forecasting in cold regions represents a challenge because the natural processes that occur within catchments vary greatly both seasonally and annually. This variability, which highly depends on climatic and topo-geomorphological characteristics within a basin, translates into increased model uncertainty and a substantial limitation when attempting to forecast river flow in cold regions, which are often poorly gauged or ungauged. To address this limitation, the “Predictions in Ungauged Basins” initiative offers a variety of studies to improve forecasting performance by adopting regionalization, spatial calibration, interpolation, and regression approaches. Process-based models demonstrate significant improvement by including remote-sensing data to replicate and derive complex hydrological processes. Empirical models, which utilize observed data to formulate a graphical solution, unlike mathematical models that require formulating the relationships between the processes, are also implemented with the most recent developments in machine learning, showing exceptional forecasting accuracy. Although process-based models provide a wide understanding of a watershed hydrology, data are often unavailable, expensive, and time-consuming to collect. They also generate numerous calibration parameters, resulting in complex and computationally demanding methods to operate. River-flow forecasting using empirical models reduces the number of calibration parameters but could produce biased results when insufficient variables are available to explain the physical mechanisms of a watershed’s hydrology. Moreover, empirical models could be potentially sensitive to calibration and validation dataset selection. In this review, Canadian studies are primarily selected to highlight some of the efforts that may be necessary in other similar cold and ungauged regions, including: (i) coping with limited data availability through regionalization methods; (ii) providing user-friendly interfaces; (iii) advancing model structure; (iv) developing a universal method for transferring regionalization parameters; (v) standardizing calibration and validation dataset selection; (vi) integrating process-based and empirical models.

show abstract

“…By examining real-time data from sensing and social media feeds to pinpoint regions of need and organize rescue activities, hybrid deep learning models may help with disaster response and making predictions. To locate damaged regions and gauge the success of relief activities, they may also assist with following disaster reconstruction efforts by examining satellite images (Ren et al 2019).…”

Section: Introductionmentioning

confidence: 99%

Hybrid deep learning model for flood frequency assessment and flood forecasting

Pandey,

Desai,

Panwar

2023

Multidiscip. Sci. J.

View full text Add to dashboard Cite

The most common and persistent natural hazard to people across the globe is flooding. The frequency of floods in a given place is defined as the likelihood and intensity of floods occurring there within a certain period. Examining historical flood data and using techniques are often used to determine the likelihood that a flood of a certain size would occur in a specific location. The method of flood prediction involves making forecasts on the frequency and severity of flooding. It may be influenced by a number of factors, including the topography, river flow, soil moisture content, and the period of rainfall. In this research, we provide a novel Cat Swarm Optimized Spatial Adversarial Network (CSO-SAN) technique for predicting and assessing flood frequency. This technique simulates the yearly greatest flow at the river Mahanadi measurement sites at Andhiyarkore, Bamanidhi, Baronda, and Kurubhatta over 60 years. The CSO-SAN model is adapted for the flood forecasting component to predict the frequency and size of future floods. The model incorporates real-time data from various sources, such as meteorological predictions and information on river flow, to anticipate the probability and severity of upcoming floods. Compared to other conventional statistical techniques and forecasting models, the CSO-SAN model outperformed them in tests conducted on the Mahanadi river basins. The model offers a viable method for improving the precision of flood frequency evaluation and flood forecasting, with significant advantages for managing and reducing flood risk.

show abstract

A Novel Hybrid Extreme Learning Machine Approach Improved by K Nearest Neighbor Method and Fireworks Algorithm for Flood Forecasting in Medium and Small Watershed of Loess Region

Cited by 20 publications

References 41 publications

A Non-Tuned Machine Learning Technique for Abutment Scour Depth in Clear Water Condition

A Non-Tuned Machine Learning Technique for Abutment Scour Depth in Clear Water Condition

Modelling river flow in cold and ungauged regions: a review of the purposes, methods, and challenges

Hybrid deep learning model for flood frequency assessment and flood forecasting

Contact Info

Product

Resources

About