Browse
Recent Submissions
Item Attention mechanism based dilated causal convolution network stacked with bidirectional lstm for one day ahead electrical load forecasting(UMT.Lahore, 2022) WAQAS AMEER AWANThis research proposes two novel deep learning based neural network architecture for multivariate Short-term electrical load forecasting problem to achieve stability and reliability in power system operations. In order to achieve forecasting accuracy the existing architectures trapped into local trends and unable to minimize overfitting in learning process. To minimize overfitting and analysis global behavior of data in learning process a multilayer deep neural network introduced which consists of dilated causal convolution neural network cascaded with bidirectional long-short term memory and attention mechanism. Dilation minimizes the number of parameters while Convolution layers in DCNN extract the patterns in the local trends of the load data and Bi-LSTM forecasts the load value depending on the patterns induced by the DCNN. Attention mechanism guides the process of reasoning by giving due attention to certain parameters in data. GDP per capita and population data added into existing Malaysian data to increase features. Linear Regression, Polynomial Regression, AR, SARIMAX, LSTM, CNN-LSTM, CNN-BiLSTM, CNN-ABiLSTM are also developed and tested. MAE, RMSE, MAPE and R-Squared are used as performance metrics. The experimental results reveals that proposed architectures performs relatively better and improves the prediction accuracy up to 1%. Hence, DCNN-ABiLSTM auspicious architectures to overcome overfitting and utilize global trends while learning load data.Item Biomass energy for pakistan(UMT.Lahore, 2022) Abdullah KhalidPakistan has been experiencing a severe energy crisis in recent decades due to insufficient electricity generation capacity relative to demand. Shortages of electricity and gas have had a direct impact on domestic and commercial activities. Natural gas is the essential resource for the confined electricity, due to the continued reduction of this source and demand, companies generating power now rely on relatively expensive furnace oil. This crisis can be overcome by looking for sources of fuel that are economically rich and widely available within the country. The majority of the Pakistani population lives in rural and remote areas currently has no access to electricity. However, renewable energy resources, particularly biomass, solar and wind, can play a key role in electrifying remote areas of the country. A hybrid power system is a promising energy generation procedure that involves a combination of diverse energy systems, mostly renewable for optimal output configuration. In modern renewable energy research (RE), optimum conditions for the production and consumption of energy systems are considered an essential feature of the economic burden of shipping. Wind, solar and biomass are three renewable energy sources that are primarily focused on this thesis and on this basis a hybrid feed system is proposed.Item Design & implementation of cpvit system in combination with thermo couple module for university of management & technology(UMT.Lahore, 2022) Muhammad YasirThe past decade has proved itself to be an era of innovations in the field of Renewable energy especially in the Solar Based Power Generation. Much work is done on it but there is always a room for betterment and innovation. This thesis proposes a technique of increasing the efficiency of solar based generation of electricity with help of solar concentrator along with the thermal constraint. This proposed system is actually a hybrid system in nature where you get output energy not only in form of Electrical Energy but as well as in the form of thermal energy. The thermal constraints act as a double usable variable, which is used for cooling as well as the medium for carrying thermal energy. Water has been used as a thermal energy carrying material. We simulated the run time data of Lahore City and simulated the system for all months of the year. Different variables where analyzed in order to find out the maximum efficiency of the system.Item An optimal energy management system for residential and industrial microgrids(UMT.Lahore, 2022) Muhammad Bilal NasirThis thesis presents an optimal scheme for the integration of renewable sources with utility grid to minimize the operational cost of the residential and industrial power systems. With the changing paradigm of solar photovoltaic in low distribution networks, utilities have allowed net-metering and Feed-in-Tariff through which residential and industrial consumers can substantially contribute to energy generation. However, in conventional settings system may underperform if resources are not scheduled optimally. For instance, with the time of usage pricing, it is possible to intake energy from the grid when grid prices are lower and supply to the grid when grid supply is higher. The proposed scheme will therefore allow the optimal resource utilization considering intermittent renewable generation as well as a time-varying utility tariff. The complete comparative analysis of on-grid and off-grid models is carried out. The results indicate that the annual saving is about 32.0% by using on-grid proposed scheme where Feed-in-Tariff is availableItem EEG-based seizure prediction with machine learning(UMT.Lahore, 2022) Muhammad Mateen QureshiEpilepsy is one of the most recognized neurological illnesses, affecting millionsof individuals worldwide. The illness has long been important in the biomedicalfield because of the threats it poses to human life. The research purposeis to develop a methodology that combines signal processing and machinelearning to predict patient-specific seizure attack so that it can be medicatedbefore the actual seizure attack. A novel method for seizure predictionis proposed that combines support vector machine and wavelet packet decomposition.The essential characteristic of this proposed method is thatonly one hour of data for processing and one to two channels for training andtesting were used, resulting in a computationally efficient technique. Firstly,raw data is being segmented of patient-specific, then discrete wavelet decompositionis performed on this segmented data which is decomposed usingdiscrete wavelet transform into four bands i.e. delta, theta, alpha and beta.Secondly, four features are extracted from this decomposed data. Thirdly,the feature matrix extracted from this decomposition is fed into the classifierto categorize seizure phase i.e. (pre-ictal and inter-ictal). In the end, oncethe pre-ictal state is detected by the support vector machine(SVM) classifier,using the Kalman filtering an alarm is generated. False-positive rate, sensitivity,and accuracy were measured as performance indicators. An averagedaccuracy of 94.9%, 97.43% of sensitivity, and 0.138 false positive rate wereachieved. Keywords: EEG, Seizure, SVM, Wavelet Packet Decomposition,Pre/Inter-IctalItem Power quality control of wind energy conversion system through adaptive dynamic programming(UMT.Lahore, 2022) Syed Muhammad AshhadDeveloping an Optimal Control strategy for Doubly Fed Induction Generator (DFIG) based Wind Energy Conversion System (WECS) in a large scale wind farm is a challenging task because output parameters of WECS are dependent on the speed of wind. Catering the problem of variable wind speed for Wind Farms has become an important issue for reliable integration of wind farms with the grid. Speed of wind depends on the weather, temperature and different other conditions which make wind speed variable throughout the day. DFIG is connected with two types of converters which are called Rotor Side Converter (RSC) and Grid Side Converter (GSC). In order to get a constant and stable output power from a wind farm, both of the converters must be effectively controlled. Both the converters are complementary for each other, rotor side converter manages the excitation power provided to the rotor and the grid side converter manages power delivered to the grid and also reduces load variation problems associated with the grid. In order to control these converters and obtain fruitful results, an adaptive dynamic programming (ADP) based control technique has been presented in this proposal. Adaptive dynamic programming is an Artificial Intelligence based control technique which is based on two networks called Action and Critic Networks. These two networks are designed with the help of artificial neural networks (ANN). The proposed topology will handle grid side converter and rotor side converter in such a way that efficient, reliable and smooth output power is obtained through a windfarm. Four different scenarios with different wind speeds are simulated and results are present. Matlab/Simulink is used to carry out the Simulation study and the controller effectiveness is shown through results.Item Load forecasting of an optimized green residential system using machine learning algorithms(UMT.Lahore, 2022) Nabeel ZahoorLoad forecasting of a micro-grid system has become a challenging task due to its high volatile nature and uncertainty. Residential energy consumption is one of the most talk about and confused topic among different domains of loads in term of future information, and mostly affected by irregular human activity and changing weather conditions. That’s why there is a lot of ambiguity in load forecasting matter. Many techniques and algorithms are used in the past to solve this issue, but an improvement gap remains. Therefore, modified techniques and algorithms are needed to reduce energy consumption as well as to enhance the smartness of the system. Load forecasting of an optimized residential system using machine learning (ML) algorithms is proposed for an islanded green residential system. Load profile of residential electricity consumption is developed by real time data collected. The maximum previous studies are carried out by using the real time data collected out of country, but the data used in the proposed study is collected in Lahore, Pakistan. Photovoltaic (PV) and wind energy (WE) units are considered as renewable energy sources in the presence of battery to entertain the residential loads in the proposed prototype. An efficient energy management system (EMS) is introduced to create a balance between power generation and consumption with the help of smart appliances under controlled framework and to overcome the peak time consumption. Prediction of load and proper energy utilization are presented to ensure the stability as well as the durability of the system. For an efficient micro-grid energy management, the residential load is forecasted using different ML algorithms at different parameters in order to check best fit of the model. ML Algorithm named non-linear autoregressive exogenous (NARX) neural network (NN) has given the best results among all ML algorithms used. As a result, an efficient model is designed for a standalone DC micro-grid.