Dr. Hassan Salah | AlAzhar University cairo (original) (raw)
Papers by Dr. Hassan Salah
The return values of wave data, e.g. 1, 25, 50 and 100 years, are the most important factors in t... more The return values of wave data, e.g. 1, 25, 50 and 100 years, are the most important factors in the design of offshore structures. For offshore structures design the maximum wave height (Hmax) is determined for different return values based on long-term measurements or hindcast models. This paper presents the determination of Hmax with different return values for Port Said offshore area located on the northern coast of Egypt based on wave data measurements. The Long-Term analysis is carried out using the Peak-Over-Threshold (POT) method. The results of this study were compared with the results of the hindcast model conducted in the study area to design some offshore platforms located in this area. A comparison demonstrated that the values of Hmax based on the measured data of the study area were more accurate and relaxed than the Hmax values based on the hindcast model. This means that the use of the Hmax (based on the measured data) in the design of offshore platforms will raise th...
Numerous increase in the development of Offshore Projects drives the need to improve wave predict... more Numerous increase in the development of Offshore Projects drives the need to improve wave predicting techniques as waves are considered the key factor that impact the design of different offshore and near-shore structures (i.e. fixed offshore platforms, marine terminal jetty, harbor, breakwaters, ..etc.). For this purpose, several parametric methods were developed based on dimensionless analysis to predict the wave parameters described in literature such as; Pierson and Moskowitz (1964) Shore Protection Manuel (1984), Coastal Engineering Manuel (2008), ..etc. The complexity of implementing these methods, which require bathymetric surveys and considerable processing time, creates the need to develop new methods utilizing the State-of-the Art of Artificial Intelligence technology (e.g. Machine Learning). This paper focuses on the prediction of sea wave parameters (i.e. Significant Wave Heights "Hs" and Significant Wave Period "Ts") used in the design of offshore and near-shore structures, applying one of the most advanced techniques in Artificial Intelligence (A.I.) called Support Vector Machine (SVM). SVM method has shown great promise to build accurate models to process any type of raw data. SVM models are based on Kernel functions (i.e. Linear, Sigmoid, Radial Basis Function and Polynomial) that are used to compute the nonlinearly separable function in the input data and then transform these data into linearly separable function to perform regression analyses with higher confidence level. The objective of this research is to apply the SVM method in predicting sea wave parameters in the southern coast of the Mediterranean sea using the historical waves and meteorological data that have been collected since 2010 in this region. Several SVM models consisting of various input combinations for the measured data have been constructed. Then, a comparison was made between SVM results and the results of conventional parametric methods. The results showed that SVM models generally have accurate results compared that resulted from the parametric methods. Furthermore, SVM using the Kernel function called a Radial Basis Function (RBF) and Polynomial gave higher accuracy in their results than other kernels function of SVM. The analysis showed that the wind speed is considered the key parameter for wave prediction. Therefore, when using SVM in predicting wave parameters, it can be said that the availability of wind data from sea or land stations makes us ignore any other factors. Finally, it can be argued that the use of SVM in prediction of wave is an effective way for overcoming the limitations described by other methods. Keywords: Wave Prediction, Support Vector Machines, Kernel Function, Radial Basis Function.
The numerous increase in offshore operational activities demands improved wave forecasting techni... more The numerous increase in offshore operational activities demands improved wave forecasting techniques. If the accurate wave data is available, it is possible to carry out the marine activities easily and safely (i.e. offshore drilling, offshore platforms and pipelines installation, naval operations, near shore construction activities, etc.). This paper focuses on the prediction of significant wave heights () by support vector machine (SVM), using various kernel functions. This study aims to evaluate the influence of fetch and meteorological data over SVM approach and also perform a comparison between SVM kernel methods. Measured sea waves off Alexandria, west coast of Egypt, and meteorological data are used in this study. Six SVM (linear kernel) models comprising of various input combinations for wind speed, fetch, sea level pressure and air temperature have been performed to evaluate the wave height prediction performance of fetch and meteorological parameter. The results indicated that the SVM model (linear kernel function) gave the satisfactory results with all parameters (wind speed, fetch, sea level pressure and air temperature). Furthermore, the analysis showed that wind speed is the most important parameters for wave prediction. The results showed also that the fetch could also be useful for the wave height estimations, especially when used in combined with wind speed. Furthermore, the SVM kernels named sigmoid and a radial basis function (RBF) comprising all parameters were investigated. The results indicated that the SVM (linear kernel) gave the same results extracted from the SVM (sigmoid kernel). However, SVM (RBF kernel) gave the best prediction performance over other SVM kernels. Their results indicated that the error statistics of SVM models are generally within an acceptable range. Therefore, SVM can be used successfully for prediction of Hs.
The wave parameters are the important factor in the offshore and near-shore activities. The choic... more The wave parameters are the important factor in the offshore and near-shore activities. The choice of the appropriate parametric method for wave parameters calculation has a significant impact on the design of maritime structures (such as; fixed offshore platform, marine terminal jetty, harbor, ... etc.) from the cost point of view. Several parametric methods have been used to predict wave parameters that have been based on data from certain areas of the sea. These methods are still used in many Coastal areas such as the coast of Egypt. Therefore, evaluating the performance of these methods based on the measured data of the coast of Egypt is very important because of many recent discoveries of oil and gas. In this study, P-M, SPM and CEM methods were used to predict the 3 hourly significant wave height (Hs) and significant wave period (Ts). The data used in evaluation are significant wave heights (Hs), significant wave periods (Ts), wind speed (u) and fetch length (F). This data mea...
The waves are the most important factor to be considered in the near-shore and offshore activitie... more The waves are the most important factor to be considered in the near-shore and offshore activities. Recently, the offshore Nile delta of the Arab republic of Egypt has seen many discoveries in the oil and gas filed, especially in front of Alexandria and Port Said. Therefore, obtaining a formula for predicting wave characteristics based on the measured data of is very important, which will contribute significantly to marine studies and the preliminary engineering design in such mega projects. In this study, the 3 hourly significant wave heights ( ) and periods ( ) with corresponding data such as wind speed (u), fetch data (F), sea level pressure (p) and air temperature ( ) used in analyses based on the hourly observations data. This paper focuses on the prediction of significant wave parameters by using nonlinear regression method based on dimensionless analysis (pi-theorem). Measured offshore waves and meteorological data for Alexandria and Port Said regions, located on the Nile del...
Current Development in Oceanography
https://www.irjet.net/archives/V7/i1/IRJET-V7I172.pdf
https://www.irjet.net/archives/V6/i12/IRJET-V6I12218.pdf
Hassan Salah, Dec 19, 2019
The so-called Medicanes (Mediterranean hurricanes) are mesoscale and warm-core Cyclones that show... more The so-called Medicanes (Mediterranean hurricanes) are mesoscale and warm-core Cyclones that show great similarities with Tropical Cyclones. Due to the difference in sea-state conditions from one region to another during the passage of any Cyclone, knowing the characteristics of the local waves of a specific region are becoming an important subject, which may lead to some emergency measures such as evacuation of fixed offshore platforms, preventing the movement of ships, closing the ports, ...etc. This paper presents a study on Cyclone of October 2019 and its associated waves, as well as determining its classification. In addition, the most probable maximum wave height (H_max), caused by the Cyclone, was calculated and compared to the H_max used in the design of existing fixed platforms in the study area. The results showed that the H_max due to the cyclone is less than H_max for the existing fixed platforms in this region and it does not need to be evacuated.
The results of the present study will be highly useful to investigate any cyclone that occurs in the Mediterranean basin, which may lead to a change in the design criteria for this region in future.
tamer el gohary, 2018
activities such as offshore oil drilling, navigational courses, oil platform installations. This ... more activities such as offshore oil drilling, navigational courses, oil platform installations. This situation leads us to get accurate wave data for any targeted operational area as Alexandria at north coast of Egypt in this case of study in order to avoid its negative impacts and to handle any activity in very safe way. In this paper, we used nonlinear regression method to analyze wave heights data and this analysis resulted in deduction of an equation that help us to predict future deep wave heights for the same region. This study aimed to evaluate the results of this method as well as to make a comparison for its results with those obtained from support vector machine method that were used before in past paper for the same region data. The results explained that the use of nonlinear regression methods gave a good result compared to the results from support vector machine. However, this study revealed that support vector machine based on radial basis function is still more superior to nonlinear regression methods. Their results indicated that the error statistics of regression methods are generally within an acceptable range. Therefore, regression methods can be used successfully for prediction of wave heights.
hassan salah, 2017
The wave parameters are the important factor in the offshore and near-shore activities. The choic... more The wave parameters are the important factor in the offshore and near-shore activities. The choice of the appropriate parametric method for wave parameters calculation has a significant impact on the design of maritime structures (such as; fixed offshore platform, marine terminal jetty, harbor, ... etc.) from the cost point of view. Several parametric methods have been used to predict wave parameters that have been based on data from certain areas of the sea. These methods are still used in many Coastal areas such as the coast of Egypt. Therefore, evaluating the performance of these methods based on the measured data of the coast of Egypt is very important because of many recent discoveries of oil and gas. In this study, P-M, SPM and CEM methods were used to predict the 3 hourly significant wave height (Hs) and significant wave period (Ts). The data used in evaluation are significant wave heights (Hs), significant wave periods (Ts), wind speed (u) and fetch length (F). This data measured in the offshore area of Alexandria and Port Said located on the southern coast of the Mediterranean Sea. This study aims to find optimal parametric method that we can used safely in the prediction of wave parameters of this region. The results indicated that the P-M method with some modification in the significant wave period equation gave the best results in the prediction of wave parameters than other methods.
hassan salah, 2017
The numerous increase in offshore operational activities demands improved wave forecasting techni... more The numerous increase in offshore operational activities demands improved wave forecasting techniques. If the accurate wave data is available, it is possible to carry out the marine activities easily and safely (i.e. offshore drilling, offshore platforms and pipelines installation, naval operations, near shore construction activities, etc.). This paper focuses on the prediction of significant wave heights () by support vector machine (SVM), using various kernel functions. This study aims to evaluate the influence of fetch and meteorological data over SVM approach and also perform a comparison between SVM kernel methods. Measured sea waves off Alexandria, west coast of Egypt, and meteorological data are used in this study. Six SVM (linear kernel) models comprising of various input combinations for wind speed, fetch, sea level pressure and air temperature have been performed to evaluate the wave height prediction performance of fetch and meteorological parameter. The results indicated that the SVM model (linear kernel function) gave the satisfactory results with all parameters (wind speed, fetch, sea level pressure and air temperature). Furthermore, the analysis showed that wind speed is the most important parameters for wave prediction. The results showed also that the fetch could also be useful for the wave height estimations, especially when used in combined with wind speed. Furthermore, the SVM kernels named sigmoid and a radial basis function (RBF) comprising all parameters were investigated. The results indicated that the SVM (linear kernel) gave the same results extracted from the SVM (sigmoid kernel). However, SVM (RBF kernel) gave the best prediction performance over other SVM kernels. Their results indicated that the error statistics of SVM models are generally within an acceptable range. Therefore, SVM can be used successfully for prediction of Hs.
hassan salah, 2017
The waves are the most important factor to be considered in the near-shore and offshore activitie... more The waves are the most important factor to be considered in the near-shore and offshore activities. Recently, the offshore Nile delta of the Arab republic of Egypt has seen many discoveries in the oil and gas filed, especially in front of Alexandria and Port Said. Therefore, obtaining a formula for predicting wave characteristics based on the measured data of is very important, which will contribute significantly to marine studies and the preliminary engineering design in such mega projects. In this study, the 3 hourly significant wave heights () and periods () with corresponding data such as wind speed (u), fetch data (F), sea level pressure (p) and air temperature () used in analyses based on the hourly observations data. This paper focuses on the prediction of significant wave parameters by using nonlinear regression method based on dimensionless analysis (pi-theorem). Measured offshore waves and meteorological data for Alexandria and Port Said regions, located on the Nile delta coast of Egypt, used in this study. This study aims to find suitable formulas to predict the wave parameters, significant wave height and period , for these regions and evaluating its performance. The results indicated that the formulas gave the satisfactory results that can be used to predict the offshore wave parameters for the Nile delta of Egypt safely.
Conference Presentations by Dr. Hassan Salah
Participate in the Global petroleum show conference (conf. no. GPS18-326), Calgary, Canada, 2018
Numerous increase in the development of Offshore Projects drives the need to improve wave predict... more Numerous increase in the development of Offshore Projects drives the need to improve wave predicting techniques as waves are considered the key factor that impact the design of different offshore and near-shore structures (i.e. fixed offshore platforms, marine terminal jetty, harbor, breakwaters, ..etc.). For this purpose, several parametric methods were developed based on dimensionless analysis to predict the wave parameters described in literature such as; Pierson and Moskowitz (1964) Shore Protection Manuel (1984), Coastal Engineering Manuel (2008), ..etc. The complexity of implementing these methods, which require bathymetric surveys and considerable processing time, creates the need to develop new methods utilizing the State-of-the Art of Artificial Intelligence technology (e.g. Machine Learning).
This paper focuses on the prediction of sea wave parameters (i.e. Significant Wave Heights "Hs" and Significant Wave Period "Ts") used in the design of offshore and near-shore structures, applying one of the most advanced techniques in Artificial Intelligence (A.I.) called Support Vector Machine (SVM). SVM method has shown great promise to build accurate models to process any type of raw data. SVM models are based on Kernel functions (i.e. Linear, Sigmoid, Radial Basis Function and Polynomial) that are used to compute the nonlinearly separable function in the input data and then transform these data into linearly separable function to perform regression analyses with higher confidence level.
The objective of this research is to apply the SVM method in predicting sea wave parameters in the southern coast of the Mediterranean sea using the historical waves and meteorological data that have been collected since 2010 in this region. Several SVM models consisting of various input combinations for the measured data have been constructed. Then, a comparison was made between SVM results and the results of conventional parametric methods.
The results showed that SVM models generally have accurate results compared that resulted from the parametric methods. Furthermore, SVM using the Kernel function called a Radial Basis Function (RBF) and Polynomial gave higher accuracy in their results than other kernels function of SVM. The analysis showed that the wind speed is considered the key parameter for wave prediction. Therefore, when using SVM in predicting wave parameters, it can be said that the availability of wind data from sea or land stations makes us ignore any other factors. Finally, it can be argued that the use of SVM in prediction of wave is an effective way for overcoming the limitations described by other methods.
Keywords: Wave Prediction, Support Vector Machines, Kernel Function, Radial Basis Function.
The return values of wave data, e.g. 1, 25, 50 and 100 years, are the most important factors in t... more The return values of wave data, e.g. 1, 25, 50 and 100 years, are the most important factors in the design of offshore structures. For offshore structures design the maximum wave height (Hmax) is determined for different return values based on long-term measurements or hindcast models. This paper presents the determination of Hmax with different return values for Port Said offshore area located on the northern coast of Egypt based on wave data measurements. The Long-Term analysis is carried out using the Peak-Over-Threshold (POT) method. The results of this study were compared with the results of the hindcast model conducted in the study area to design some offshore platforms located in this area. A comparison demonstrated that the values of Hmax based on the measured data of the study area were more accurate and relaxed than the Hmax values based on the hindcast model. This means that the use of the Hmax (based on the measured data) in the design of offshore platforms will raise th...
Numerous increase in the development of Offshore Projects drives the need to improve wave predict... more Numerous increase in the development of Offshore Projects drives the need to improve wave predicting techniques as waves are considered the key factor that impact the design of different offshore and near-shore structures (i.e. fixed offshore platforms, marine terminal jetty, harbor, breakwaters, ..etc.). For this purpose, several parametric methods were developed based on dimensionless analysis to predict the wave parameters described in literature such as; Pierson and Moskowitz (1964) Shore Protection Manuel (1984), Coastal Engineering Manuel (2008), ..etc. The complexity of implementing these methods, which require bathymetric surveys and considerable processing time, creates the need to develop new methods utilizing the State-of-the Art of Artificial Intelligence technology (e.g. Machine Learning). This paper focuses on the prediction of sea wave parameters (i.e. Significant Wave Heights "Hs" and Significant Wave Period "Ts") used in the design of offshore and near-shore structures, applying one of the most advanced techniques in Artificial Intelligence (A.I.) called Support Vector Machine (SVM). SVM method has shown great promise to build accurate models to process any type of raw data. SVM models are based on Kernel functions (i.e. Linear, Sigmoid, Radial Basis Function and Polynomial) that are used to compute the nonlinearly separable function in the input data and then transform these data into linearly separable function to perform regression analyses with higher confidence level. The objective of this research is to apply the SVM method in predicting sea wave parameters in the southern coast of the Mediterranean sea using the historical waves and meteorological data that have been collected since 2010 in this region. Several SVM models consisting of various input combinations for the measured data have been constructed. Then, a comparison was made between SVM results and the results of conventional parametric methods. The results showed that SVM models generally have accurate results compared that resulted from the parametric methods. Furthermore, SVM using the Kernel function called a Radial Basis Function (RBF) and Polynomial gave higher accuracy in their results than other kernels function of SVM. The analysis showed that the wind speed is considered the key parameter for wave prediction. Therefore, when using SVM in predicting wave parameters, it can be said that the availability of wind data from sea or land stations makes us ignore any other factors. Finally, it can be argued that the use of SVM in prediction of wave is an effective way for overcoming the limitations described by other methods. Keywords: Wave Prediction, Support Vector Machines, Kernel Function, Radial Basis Function.
The numerous increase in offshore operational activities demands improved wave forecasting techni... more The numerous increase in offshore operational activities demands improved wave forecasting techniques. If the accurate wave data is available, it is possible to carry out the marine activities easily and safely (i.e. offshore drilling, offshore platforms and pipelines installation, naval operations, near shore construction activities, etc.). This paper focuses on the prediction of significant wave heights () by support vector machine (SVM), using various kernel functions. This study aims to evaluate the influence of fetch and meteorological data over SVM approach and also perform a comparison between SVM kernel methods. Measured sea waves off Alexandria, west coast of Egypt, and meteorological data are used in this study. Six SVM (linear kernel) models comprising of various input combinations for wind speed, fetch, sea level pressure and air temperature have been performed to evaluate the wave height prediction performance of fetch and meteorological parameter. The results indicated that the SVM model (linear kernel function) gave the satisfactory results with all parameters (wind speed, fetch, sea level pressure and air temperature). Furthermore, the analysis showed that wind speed is the most important parameters for wave prediction. The results showed also that the fetch could also be useful for the wave height estimations, especially when used in combined with wind speed. Furthermore, the SVM kernels named sigmoid and a radial basis function (RBF) comprising all parameters were investigated. The results indicated that the SVM (linear kernel) gave the same results extracted from the SVM (sigmoid kernel). However, SVM (RBF kernel) gave the best prediction performance over other SVM kernels. Their results indicated that the error statistics of SVM models are generally within an acceptable range. Therefore, SVM can be used successfully for prediction of Hs.
The wave parameters are the important factor in the offshore and near-shore activities. The choic... more The wave parameters are the important factor in the offshore and near-shore activities. The choice of the appropriate parametric method for wave parameters calculation has a significant impact on the design of maritime structures (such as; fixed offshore platform, marine terminal jetty, harbor, ... etc.) from the cost point of view. Several parametric methods have been used to predict wave parameters that have been based on data from certain areas of the sea. These methods are still used in many Coastal areas such as the coast of Egypt. Therefore, evaluating the performance of these methods based on the measured data of the coast of Egypt is very important because of many recent discoveries of oil and gas. In this study, P-M, SPM and CEM methods were used to predict the 3 hourly significant wave height (Hs) and significant wave period (Ts). The data used in evaluation are significant wave heights (Hs), significant wave periods (Ts), wind speed (u) and fetch length (F). This data mea...
The waves are the most important factor to be considered in the near-shore and offshore activitie... more The waves are the most important factor to be considered in the near-shore and offshore activities. Recently, the offshore Nile delta of the Arab republic of Egypt has seen many discoveries in the oil and gas filed, especially in front of Alexandria and Port Said. Therefore, obtaining a formula for predicting wave characteristics based on the measured data of is very important, which will contribute significantly to marine studies and the preliminary engineering design in such mega projects. In this study, the 3 hourly significant wave heights ( ) and periods ( ) with corresponding data such as wind speed (u), fetch data (F), sea level pressure (p) and air temperature ( ) used in analyses based on the hourly observations data. This paper focuses on the prediction of significant wave parameters by using nonlinear regression method based on dimensionless analysis (pi-theorem). Measured offshore waves and meteorological data for Alexandria and Port Said regions, located on the Nile del...
Current Development in Oceanography
https://www.irjet.net/archives/V7/i1/IRJET-V7I172.pdf
https://www.irjet.net/archives/V6/i12/IRJET-V6I12218.pdf
Hassan Salah, Dec 19, 2019
The so-called Medicanes (Mediterranean hurricanes) are mesoscale and warm-core Cyclones that show... more The so-called Medicanes (Mediterranean hurricanes) are mesoscale and warm-core Cyclones that show great similarities with Tropical Cyclones. Due to the difference in sea-state conditions from one region to another during the passage of any Cyclone, knowing the characteristics of the local waves of a specific region are becoming an important subject, which may lead to some emergency measures such as evacuation of fixed offshore platforms, preventing the movement of ships, closing the ports, ...etc. This paper presents a study on Cyclone of October 2019 and its associated waves, as well as determining its classification. In addition, the most probable maximum wave height (H_max), caused by the Cyclone, was calculated and compared to the H_max used in the design of existing fixed platforms in the study area. The results showed that the H_max due to the cyclone is less than H_max for the existing fixed platforms in this region and it does not need to be evacuated.
The results of the present study will be highly useful to investigate any cyclone that occurs in the Mediterranean basin, which may lead to a change in the design criteria for this region in future.
tamer el gohary, 2018
activities such as offshore oil drilling, navigational courses, oil platform installations. This ... more activities such as offshore oil drilling, navigational courses, oil platform installations. This situation leads us to get accurate wave data for any targeted operational area as Alexandria at north coast of Egypt in this case of study in order to avoid its negative impacts and to handle any activity in very safe way. In this paper, we used nonlinear regression method to analyze wave heights data and this analysis resulted in deduction of an equation that help us to predict future deep wave heights for the same region. This study aimed to evaluate the results of this method as well as to make a comparison for its results with those obtained from support vector machine method that were used before in past paper for the same region data. The results explained that the use of nonlinear regression methods gave a good result compared to the results from support vector machine. However, this study revealed that support vector machine based on radial basis function is still more superior to nonlinear regression methods. Their results indicated that the error statistics of regression methods are generally within an acceptable range. Therefore, regression methods can be used successfully for prediction of wave heights.
hassan salah, 2017
The wave parameters are the important factor in the offshore and near-shore activities. The choic... more The wave parameters are the important factor in the offshore and near-shore activities. The choice of the appropriate parametric method for wave parameters calculation has a significant impact on the design of maritime structures (such as; fixed offshore platform, marine terminal jetty, harbor, ... etc.) from the cost point of view. Several parametric methods have been used to predict wave parameters that have been based on data from certain areas of the sea. These methods are still used in many Coastal areas such as the coast of Egypt. Therefore, evaluating the performance of these methods based on the measured data of the coast of Egypt is very important because of many recent discoveries of oil and gas. In this study, P-M, SPM and CEM methods were used to predict the 3 hourly significant wave height (Hs) and significant wave period (Ts). The data used in evaluation are significant wave heights (Hs), significant wave periods (Ts), wind speed (u) and fetch length (F). This data measured in the offshore area of Alexandria and Port Said located on the southern coast of the Mediterranean Sea. This study aims to find optimal parametric method that we can used safely in the prediction of wave parameters of this region. The results indicated that the P-M method with some modification in the significant wave period equation gave the best results in the prediction of wave parameters than other methods.
hassan salah, 2017
The numerous increase in offshore operational activities demands improved wave forecasting techni... more The numerous increase in offshore operational activities demands improved wave forecasting techniques. If the accurate wave data is available, it is possible to carry out the marine activities easily and safely (i.e. offshore drilling, offshore platforms and pipelines installation, naval operations, near shore construction activities, etc.). This paper focuses on the prediction of significant wave heights () by support vector machine (SVM), using various kernel functions. This study aims to evaluate the influence of fetch and meteorological data over SVM approach and also perform a comparison between SVM kernel methods. Measured sea waves off Alexandria, west coast of Egypt, and meteorological data are used in this study. Six SVM (linear kernel) models comprising of various input combinations for wind speed, fetch, sea level pressure and air temperature have been performed to evaluate the wave height prediction performance of fetch and meteorological parameter. The results indicated that the SVM model (linear kernel function) gave the satisfactory results with all parameters (wind speed, fetch, sea level pressure and air temperature). Furthermore, the analysis showed that wind speed is the most important parameters for wave prediction. The results showed also that the fetch could also be useful for the wave height estimations, especially when used in combined with wind speed. Furthermore, the SVM kernels named sigmoid and a radial basis function (RBF) comprising all parameters were investigated. The results indicated that the SVM (linear kernel) gave the same results extracted from the SVM (sigmoid kernel). However, SVM (RBF kernel) gave the best prediction performance over other SVM kernels. Their results indicated that the error statistics of SVM models are generally within an acceptable range. Therefore, SVM can be used successfully for prediction of Hs.
hassan salah, 2017
The waves are the most important factor to be considered in the near-shore and offshore activitie... more The waves are the most important factor to be considered in the near-shore and offshore activities. Recently, the offshore Nile delta of the Arab republic of Egypt has seen many discoveries in the oil and gas filed, especially in front of Alexandria and Port Said. Therefore, obtaining a formula for predicting wave characteristics based on the measured data of is very important, which will contribute significantly to marine studies and the preliminary engineering design in such mega projects. In this study, the 3 hourly significant wave heights () and periods () with corresponding data such as wind speed (u), fetch data (F), sea level pressure (p) and air temperature () used in analyses based on the hourly observations data. This paper focuses on the prediction of significant wave parameters by using nonlinear regression method based on dimensionless analysis (pi-theorem). Measured offshore waves and meteorological data for Alexandria and Port Said regions, located on the Nile delta coast of Egypt, used in this study. This study aims to find suitable formulas to predict the wave parameters, significant wave height and period , for these regions and evaluating its performance. The results indicated that the formulas gave the satisfactory results that can be used to predict the offshore wave parameters for the Nile delta of Egypt safely.
Participate in the Global petroleum show conference (conf. no. GPS18-326), Calgary, Canada, 2018
Numerous increase in the development of Offshore Projects drives the need to improve wave predict... more Numerous increase in the development of Offshore Projects drives the need to improve wave predicting techniques as waves are considered the key factor that impact the design of different offshore and near-shore structures (i.e. fixed offshore platforms, marine terminal jetty, harbor, breakwaters, ..etc.). For this purpose, several parametric methods were developed based on dimensionless analysis to predict the wave parameters described in literature such as; Pierson and Moskowitz (1964) Shore Protection Manuel (1984), Coastal Engineering Manuel (2008), ..etc. The complexity of implementing these methods, which require bathymetric surveys and considerable processing time, creates the need to develop new methods utilizing the State-of-the Art of Artificial Intelligence technology (e.g. Machine Learning).
This paper focuses on the prediction of sea wave parameters (i.e. Significant Wave Heights "Hs" and Significant Wave Period "Ts") used in the design of offshore and near-shore structures, applying one of the most advanced techniques in Artificial Intelligence (A.I.) called Support Vector Machine (SVM). SVM method has shown great promise to build accurate models to process any type of raw data. SVM models are based on Kernel functions (i.e. Linear, Sigmoid, Radial Basis Function and Polynomial) that are used to compute the nonlinearly separable function in the input data and then transform these data into linearly separable function to perform regression analyses with higher confidence level.
The objective of this research is to apply the SVM method in predicting sea wave parameters in the southern coast of the Mediterranean sea using the historical waves and meteorological data that have been collected since 2010 in this region. Several SVM models consisting of various input combinations for the measured data have been constructed. Then, a comparison was made between SVM results and the results of conventional parametric methods.
The results showed that SVM models generally have accurate results compared that resulted from the parametric methods. Furthermore, SVM using the Kernel function called a Radial Basis Function (RBF) and Polynomial gave higher accuracy in their results than other kernels function of SVM. The analysis showed that the wind speed is considered the key parameter for wave prediction. Therefore, when using SVM in predicting wave parameters, it can be said that the availability of wind data from sea or land stations makes us ignore any other factors. Finally, it can be argued that the use of SVM in prediction of wave is an effective way for overcoming the limitations described by other methods.
Keywords: Wave Prediction, Support Vector Machines, Kernel Function, Radial Basis Function.