16 skills found
mahdi-usask / Wind Speed Forecasting For Wind Power Generation Plant. Neural Network ML Based Prediction Algo. For largescale wind power penetration Wind speed prediction is a basic requirement of wind energy generation. There are many artificial neural network (ANN), ARMA, ARIMA approaches proposed in the recent literature in order to tackle this problem. This paper will use the artificial neural network (ANN) approach to get a prediction of wind speed using historical wind speed data. The historical data used here were gathered from NREL website ,as hourly basis from 80 meter hub height. The measurement location is NREL Flatirons Campus (M2). The readings displayed are derived from instruments mounted on or near a 82 meter (270 foot) meteorological tower located at the western edge of the Flatirons Campus (formerly NWTC) and about 11 km (7 miles) west of Broomfield, and approximately 8 km (5 miles) south of Boulder, Colorado. The tower is located at 39o 54' 38.34" N and 105o 14' 5.28" W (datum WGS84) with its base at an elevation of 1855 meters (6085 feet) above mean sea level. Data from year 2014 to 2018, in total 5 years of data has been used here as dataframe. Here the neural network has been implemented by Tensorflow’s Keras API. The used model is “sequential”. Four dense layer has been used in the optimized model. LSTM(Long- short-term memory) architecture has been used here as neural network architecture. Activation function being used in the dense layers are dropout function. The optimizer being used here is Adam. Here various range of Dropout function has been examined and chosen the best fit for this model. Also this paper examined various kinds of optimization method and used the best fitted one. The model performances were evaluated using the mean squared error using adam optimizer. Various kinds of data analytic techniques has been used here for better visualization and in depth understanding of the dataset and its variables. Since it is mostly a time series data so in the analytic part how the data is being changed with time has been shown. From the result of the predicted dataset it can be state that, this wind speed prediction model works best for all kinds of winds speed besides overfitted/ abnormal wind speeds which is a very rare case scenario.
uzh-rpg / Hybrid Ann SnnImplementation of "A Hybrid ANN-SNN Architecture for Low-Power and Low-Latency Visual Perception". CVPRW 2024
raymelon / TrafficLightNeuralNetwork:traffic_light: An Artificial Neural Network based Traffic Light Controller for intersections. Computational Intelligence class final project.
shahanHasan / Intrusion Detection System Adversarial Attacks Network Intrusion Detection System on CSE-CIC-IDS2018 using ML classifiers and DNN ( ANN , CNN , RNN ) | Hyper-parameter Optimization { learning rate, epochs, network architectures, regularisation } | Adversarial Attacks - Label flip , Adversarial samples , KNN (defence)
thehapyone / Predictive Maintenance ProjectIn this project, high dimensional noisy data collected from thousands of trucks during the course of 4 years was compressed using Artificial Neural Networks (ANN). This compressed meaningful information was used for performing predictive maintenance on turbochargers. Our novel deep learning ANN architecture, compressed vehicle data by over 87% while still improving fault forecasting prediction by 23% and even with extreme data size reduction of 99.7% we still see a significant performance improvement of 6.31%.
deep-div / AI ML DL DS PlaygroundA collection of simple and practical examples of ANN, CNN, and RNN models using PyTorch. Perfect for learning core deep learning architectures through code.
Xingfush / ANN To SNN For Inception ResNetThis is a project for the proposed Homeostasis-based ANN-to-SNN conversion for Inception and Residual architecture.
sminerport / SequencePredictionANNPredict next number in a sequence using a simple ANN. Modularized code with classes for data preparation, neural network architecture, and training.
CoAxLab / AzadGame-playing ANNs that use a stumbler-strategist architecture.
ajinkyaT / GA ANNOptimising ANN architecture using genetic algorithm.
GeorgiosEtsias / Optimizing ANN ArchitectureThe project utilizes genetic algorithms to optimize the architecture of Artificial Neural Networks (ANN) conducting regression analysis on monochromatic laboratory figures
gevinova / MACSANNThis is a python implementation of Memetic algorithm with crossover to search architecture of neural network (MACSANN). The objective of the algorithm is to find an architecture of an ANN that could solve a given problem.
misbahiradat / ANN RegressionOur PyTorch project simplifies multi-output regression tasks. It's equipped with a flexible ANN architecture, automated activation functions, and versatile loss functions. Easily train, evaluate, and visualize models, with room for future enhancements.
Srujanx / ANN From Scratch TitanicImplementing 3-2-2-1 ANN Architecture on Titanic Dataset. No sklearn or Keras , the primary objective being deeper dive into understanding neural networks. Manual feature selection and encoding considering top 3 features only.
hamidrahmanifard / ML Model GAGDIn this study, I predicted the tertiary oil recovery with the Gas-Assisted Gravity Drainage (GAGD) method in fractured porous media using shallow and deep neural network (NN) algorithms. I explored the tertiary oil recovery prediction versus viscosity, density, surface tension, porosity, permeability, wettability index, connate water saturation, residual water saturation after flooding, production rate, production time, capillary number, dimensionless time, and bond number. For this purpose, using 263 sets of experimental data from the literature [91,92], I first assessed the relationship between various parameters and tertiary oil recovery and determined a subset of the most influential parameters. Running DOE using ANOVA over the variables mentioned above showed that the tertiary oil recovery is a strong function of the wettability index, connate water saturation, residual water flooding, and production time. As the next step, I conducted a comparative study on one to four hidden layers ANN models to find the best architecture of the NN algorithms for predicting the tertiary oil recovery. My benchmarks for selecting the best algorithm were RMSE, MRE, and R2. Note that because of the acceptable performance of the Levenberg-Marquardt (LM) algorithm in terms of error and execution time, I used this algorithm in training the neural network models. Finally, for the transfer functions, I deployed tansig function for all layers except the output layer where purlin function is utilized.
ArchanaBathula / Intrusion Detection System For Internet Of ThingsA well-performing machine learning algorithm called Artificial Neural Network (ANN) is developed with improved network architecture. The usage of a renowned meta-heuristic algorithm called Spotted Hyena Optimization (SHO) is used for selecting the optimal hidden neurons for ANN. The main objective of the improved training is to minimize the error difference between the target and the measured output, so as to enhance the detection accuracy. Finally, the experimental results and simulations prove the stability and robustness of the proposed model in terms of a variety of performance metrics over other machine learning models.