Deep learning approach to forecasting hourly solar irradiance
- Authors: Obiora, Chibuzor Nkemdilim
- Date: 2020
- Subjects: Machine learning , Electromagnetic waves
- Language: English
- Type: Masters (Thesis)
- Identifier: http://hdl.handle.net/10210/445210 , uj:38941
- Description: Abstract: In this dissertation, six artificial intelligence (AI) based methods for forecasting solar irradiance are presented. Solar energy is a clean renewable energy source (RES) which is free and abundant in nature. But despite the environmental impacts of fossil energy, global dependence on it is yet to drop appreciably in favor of solar energy for power generation purposes. Although the latest improvements on the technologies of photovoltaic (PV) cells have led to a significant drop in the cost of solar panels, solar power is still unattractive to some consumers due to its unpredictability. Consequently, accurate prediction of solar irradiance for stable solar power production continues to be a critical need both in the field of physical simulations or artificial intelligence. The performance of various methods in use for prediction of solar irradiance depends on the diversity of dataset, time step, experimental setup, performance evaluators, and forecasting horizon. In this study, historical meteorological data for the city of Johannesburg were used as training data for the solar irradiance forecast. Data collected for this work spanned from 1984 to 2019. Only ten years (2009 to 2018) of data was used. Tools used are Jupyter notebook and Computer with Nvidia GPU... , M.Ing. (Electrical and Electronic Engineering Management)
- Full Text:
- Authors: Obiora, Chibuzor Nkemdilim
- Date: 2020
- Subjects: Machine learning , Electromagnetic waves
- Language: English
- Type: Masters (Thesis)
- Identifier: http://hdl.handle.net/10210/445210 , uj:38941
- Description: Abstract: In this dissertation, six artificial intelligence (AI) based methods for forecasting solar irradiance are presented. Solar energy is a clean renewable energy source (RES) which is free and abundant in nature. But despite the environmental impacts of fossil energy, global dependence on it is yet to drop appreciably in favor of solar energy for power generation purposes. Although the latest improvements on the technologies of photovoltaic (PV) cells have led to a significant drop in the cost of solar panels, solar power is still unattractive to some consumers due to its unpredictability. Consequently, accurate prediction of solar irradiance for stable solar power production continues to be a critical need both in the field of physical simulations or artificial intelligence. The performance of various methods in use for prediction of solar irradiance depends on the diversity of dataset, time step, experimental setup, performance evaluators, and forecasting horizon. In this study, historical meteorological data for the city of Johannesburg were used as training data for the solar irradiance forecast. Data collected for this work spanned from 1984 to 2019. Only ten years (2009 to 2018) of data was used. Tools used are Jupyter notebook and Computer with Nvidia GPU... , M.Ing. (Electrical and Electronic Engineering Management)
- Full Text:
Optimised soft-core processor architecture for noise jamming
- Authors: Mfana, Madodana
- Date: 2019
- Subjects: Noise control
- Language: English
- Type: Masters (Thesis)
- Identifier: http://hdl.handle.net/10210/417944 , uj:35416
- Description: M.Ing. (Electrical & Electronic Engineering) , Abstract: Noise jamming is a traditional electronic counter measure (ECM) that existed since the establishment of electronic warfare (EW). Traditional noise jamming techniques have been shown to be failing when interacting with intelligent Radar systems such as pulse Doppler radar. Hence there is a need to introduce new noise jamming techniques with digital architecture that will provide improved performance against smart pulse Doppler radar. The work is undertaken to investigate the feasibility of digitizing noise jamming. It focuses on analog-to-digital conversion optimization towards noise jamming architecture, as a result digitization will allow for an opportunity for adaptation of intelligent processing that previously didn’t exist. In this dissertation, certain contributions to the field of noise jamming were made by introducing state of the art odd/even order sampling architecture by proving four case studies. Case study 1 experimentally investigates sample frequency behaviour. Case study 2 uses simulation to investigate step-size and dynamic range behaviour. Case study 3 uses FPGA implementation and SNR to investigate quantization error behaviour. Case study 3 also uses SNR to investigate superiority of proposed odd/even order sampling. Lastly case study 4 uses field measurements, FPGA implementation and SNR to investigate practical implementation of digitized noise jamming. The main contribution is concerned with an architecture that digitizes, reduces sample frequency, optimizes digital resource utilization while reducing noise jamming signal-to-noise ratio. The approach evaluates and empirically compares three sampling techniques from lecture Mod-Δ, Mod-Δ (Gaussian) and Mod-Δ (Sinusoidal) with proposed novel odd/even order sampling. Sampling techniques are evaluated in terms of quantization error, mean square error and signal-to-noise ratio. It was found that the proposed novel odd/even order sampling achieved most case SNR performance of 6 dB in comparison to 18 dB for Mod-Δ. Sampling frequency findings indicated that the proposed novel odd/even order sampling had achieved sampling frequency of 2 kHz in comparison to 8 kHz from traditional 1st order sigma-delta. Dynamic range findings indicated that the proposed odd/even order sampling achieved a dynamic range of 1.088 volts/ms in comparison to 1.185 volts/ms from traditional 1st order sigma-delta. Findings have indicated that the proposed odd/even order sampling has superior SNR and sampling frequency...
- Full Text:
- Authors: Mfana, Madodana
- Date: 2019
- Subjects: Noise control
- Language: English
- Type: Masters (Thesis)
- Identifier: http://hdl.handle.net/10210/417944 , uj:35416
- Description: M.Ing. (Electrical & Electronic Engineering) , Abstract: Noise jamming is a traditional electronic counter measure (ECM) that existed since the establishment of electronic warfare (EW). Traditional noise jamming techniques have been shown to be failing when interacting with intelligent Radar systems such as pulse Doppler radar. Hence there is a need to introduce new noise jamming techniques with digital architecture that will provide improved performance against smart pulse Doppler radar. The work is undertaken to investigate the feasibility of digitizing noise jamming. It focuses on analog-to-digital conversion optimization towards noise jamming architecture, as a result digitization will allow for an opportunity for adaptation of intelligent processing that previously didn’t exist. In this dissertation, certain contributions to the field of noise jamming were made by introducing state of the art odd/even order sampling architecture by proving four case studies. Case study 1 experimentally investigates sample frequency behaviour. Case study 2 uses simulation to investigate step-size and dynamic range behaviour. Case study 3 uses FPGA implementation and SNR to investigate quantization error behaviour. Case study 3 also uses SNR to investigate superiority of proposed odd/even order sampling. Lastly case study 4 uses field measurements, FPGA implementation and SNR to investigate practical implementation of digitized noise jamming. The main contribution is concerned with an architecture that digitizes, reduces sample frequency, optimizes digital resource utilization while reducing noise jamming signal-to-noise ratio. The approach evaluates and empirically compares three sampling techniques from lecture Mod-Δ, Mod-Δ (Gaussian) and Mod-Δ (Sinusoidal) with proposed novel odd/even order sampling. Sampling techniques are evaluated in terms of quantization error, mean square error and signal-to-noise ratio. It was found that the proposed novel odd/even order sampling achieved most case SNR performance of 6 dB in comparison to 18 dB for Mod-Δ. Sampling frequency findings indicated that the proposed novel odd/even order sampling had achieved sampling frequency of 2 kHz in comparison to 8 kHz from traditional 1st order sigma-delta. Dynamic range findings indicated that the proposed odd/even order sampling achieved a dynamic range of 1.088 volts/ms in comparison to 1.185 volts/ms from traditional 1st order sigma-delta. Findings have indicated that the proposed odd/even order sampling has superior SNR and sampling frequency...
- Full Text:
Photovoltaic system maximum power point tracking under partial shaded weather conditions using machine learning algorithms
- Authors: Nkambule, Mpho Sam
- Date: 2019
- Subjects: Photovoltaic cells , Photovoltaic power systems , Artificial intelligence , Machine learning
- Language: English
- Type: Masters (Thesis)
- Identifier: http://hdl.handle.net/10210/418317 , uj:35462
- Description: Abstract: The rapid growth of demand of electrical energy and depletion of fossil opened door for renewable energy. The expeditious broadening of Photovoltaic (PV) energy has attracted the private and government precinct world-wide, due to the reduction of costs and being cleaner source of energy. However, most of the maximum power point tracking (MPPT) controllers are inefficient under rapidly changing environmental conditions. In addition, under partial shading conditions (PSC) MPPT controllers fail to track global maximum power point (GMPP). Therefore, it is essential to propose an MPPT controller that will be able to locate GMPP using historical weather data. The work is undertaken to investigate the feasibility of using machine learning (ML) based MPPT techniques to harness maximum power on a PV system under PSC. If successful, the ML based MPPT algorithms could lead to a reduction in power losses in a PV system. In this dissertation, certain contributions to the field of PV systems and ML based were made by introducing two online and eleven artificial intelligence (AI) MPPT techniques, by presenting four experiments under different weather conditions. The first contribution is concerned with an MPPT system that harvests maximum power under PSC, using Johannesburg real-time weather data. The system consists of an MPPT controller cascaded with a PID controller, to reduce errors of the MPPT algorithms, to improve the system’s performance. First, the system evaluates and compares the online [Perturb & Observe (P&O) and Incremental Conductance (INC)], to determine the most powerful MPPT algorithm. Secondly, the system validates the performance of the eleven AI MPPT methods [Fuzzy Logic Control (FLC) and Recurrent Neural Network (RNN), Support Vector Machines (SVM), the Weighted K-nearest neighbour (WK-NN), a Gaussian process regression (GPR), Decision Tree (DT), Multivariate linear regression (MLR), Linear discriminant analysis (LDA), Naïve Bayes classifier (NBC), Bagged Tree (BT) and Boosted Tree (BoT)] under PSC. The ML based techniques are evaluated using four types of error [root mean squared error (RMSE), mean absolute error (MAE), Mean squared error (MSE) and R-squared (𝑅2)]. For the first experiment, online methods are empirically compared in the form of four case studies conducted under various weather conditions. The results showed that INC performed significantly better than P&O under PSC. INC overperformed P&O in all case studies in terms of power extraction. Nevertheless, P&O has less settling time around maximum power point... , M.Phil. (Electrical and Electronic Engineering)
- Full Text:
- Authors: Nkambule, Mpho Sam
- Date: 2019
- Subjects: Photovoltaic cells , Photovoltaic power systems , Artificial intelligence , Machine learning
- Language: English
- Type: Masters (Thesis)
- Identifier: http://hdl.handle.net/10210/418317 , uj:35462
- Description: Abstract: The rapid growth of demand of electrical energy and depletion of fossil opened door for renewable energy. The expeditious broadening of Photovoltaic (PV) energy has attracted the private and government precinct world-wide, due to the reduction of costs and being cleaner source of energy. However, most of the maximum power point tracking (MPPT) controllers are inefficient under rapidly changing environmental conditions. In addition, under partial shading conditions (PSC) MPPT controllers fail to track global maximum power point (GMPP). Therefore, it is essential to propose an MPPT controller that will be able to locate GMPP using historical weather data. The work is undertaken to investigate the feasibility of using machine learning (ML) based MPPT techniques to harness maximum power on a PV system under PSC. If successful, the ML based MPPT algorithms could lead to a reduction in power losses in a PV system. In this dissertation, certain contributions to the field of PV systems and ML based were made by introducing two online and eleven artificial intelligence (AI) MPPT techniques, by presenting four experiments under different weather conditions. The first contribution is concerned with an MPPT system that harvests maximum power under PSC, using Johannesburg real-time weather data. The system consists of an MPPT controller cascaded with a PID controller, to reduce errors of the MPPT algorithms, to improve the system’s performance. First, the system evaluates and compares the online [Perturb & Observe (P&O) and Incremental Conductance (INC)], to determine the most powerful MPPT algorithm. Secondly, the system validates the performance of the eleven AI MPPT methods [Fuzzy Logic Control (FLC) and Recurrent Neural Network (RNN), Support Vector Machines (SVM), the Weighted K-nearest neighbour (WK-NN), a Gaussian process regression (GPR), Decision Tree (DT), Multivariate linear regression (MLR), Linear discriminant analysis (LDA), Naïve Bayes classifier (NBC), Bagged Tree (BT) and Boosted Tree (BoT)] under PSC. The ML based techniques are evaluated using four types of error [root mean squared error (RMSE), mean absolute error (MAE), Mean squared error (MSE) and R-squared (𝑅2)]. For the first experiment, online methods are empirically compared in the form of four case studies conducted under various weather conditions. The results showed that INC performed significantly better than P&O under PSC. INC overperformed P&O in all case studies in terms of power extraction. Nevertheless, P&O has less settling time around maximum power point... , M.Phil. (Electrical and Electronic Engineering)
- Full Text:
- «
- ‹
- 1
- ›
- »