IT Expo

Published on Slideshow
Static slideshow
Download PDF version
Download PDF version
Embed video
Share video
Ask about this video

Scene 1 (0s)

[Audio] Greetings Everyone, We are group 7, and we are excited to present our capstone project for the IT Expo. Our team consists of Bavithra Ganesan, Jasmeet Kaur, Pritesh Dalal, and Rutvik Shah. Our project is focused on forecasting energy use in Industry settings using data mining methods. Over the past 3 months, we have dedicated ourselves to researching and developing a solution that we believe can have a significant impact on Steel Industry. We have leveraged our collective expertise in Machine Learning & Deep Learning to create a solution that we are proud to showcase today. During this presentation, we will be discussing model architecture, techniques employed in designing the model and summary of our proposed model. We encourage you to ask questions and provide feedback throughout the presentation. Thank you for joining us, and we hope you enjoy our presentation..

Scene 2 (1m 4s)

[Audio] Industrial operations need a lot of energy, which has a big impact on the environmental impact and overall operational costs of any manufacturing facility. Consequently, it is crucial to maximise energy use without sacrificing the product's quality or quantity. The purpose of this study is to determine whether data mining techniques may be used to forecast daily energy usage patterns in a South Korean steel industry. The project focuses on using artificial neural networks and machine learning algorithms to interpret data gathered from various sources inside the facility. The main goal of this study is to assess the reliability and efficiency of these data mining algorithms in predicting energy consumption trends in the steel sector. The study's findings will point out places where energy can be saved, operational efficiency can be raised, and they will provide information on how these models may be applied in real-world situations. The steel industry can lessen its influence on the environment, maximise energy use, and cut costs by precisely anticipating energy consumption trends. The suggested remedy could be applied to different industrial settings and aid in the development of sustainable energy consumption patterns..

Scene 3 (2m 26s)

[Audio] A step-by-step procedure is used to create the energy consumption prediction model for the steel sector, commencing with data retrieval, EDA, data preprocessing, data preparation, sampling, modelling, model evaluation and tuning, and model comparison & selection. Below is a quick breakdown of each action: Data Retrieval: In this process, information is gathered from a variety of databases, sensors, and records located inside the steel manufacturing facility. EDA (Exploratory Data Analysis): At this step, the data that was retrieved from multiple sources is examined. To comprehend the properties of the data, data analysis techniques are used to spot patterns, correlations, and outliers. Preprocessing the data involves cleaning, transforming, and normalising the data to get rid of errors and inconsistencies. The data must be prepared for modelling, which is why this phase is so important. Data Preparation: For the purpose of developing and assessing models, the preprocessed data is divided into training, validation, and testing datasets. Sampling: To train the model, data samples are at random chosen from the training dataset. For the model to be accurate and general, this stage is crucial. Modeling: Several machine learning techniques, including Artificial Neural Networks, Linear Regression, and Random Forests Regressor , are employed in this stage to create models for estimating energy usage in the steel sector. Model Evaluation and Tuning: Performance measurements like RMSE (Root Mean Square Error) and R-squared are used to measure how well the model performs by comparing the predicted values with the actual values. The hyperparameters are then tweaked to increase the performance of the model. The last step involves evaluating the effectiveness of several models and choosing the best one based on assessment measures. Overall, there are several stages to the energy consumption prediction model for the steel sector, from data retrieval to model selection. This method enables the model to anticipate energy usage with accuracy, pinpoint opportunities for energy savings, and boost operational effectiveness..

Scene 4 (5m 7s)

[Audio] Linear regression, random forest regressor, and artificial neural networks (ANN) are techniques used for predictive modeling. Each approach has its strengths and weaknesses, and the choice of technique depends on the nature of the data and the specific problem at hand. Linear regression is a statistical method used to model the relationship between a dependent variable and one or more independent variables. In the context of predicting energy consumption by industry, the dependent variable would be the energy consumption. Linear regression assumes that there is a linear relationship between the dependent variable and the independent variables, and it estimates the parameters of the linear equation to best fit the data. The resulting model can be used to make predictions of energy consumption for new data points. Random forest regression is a type of ensemble learning method that uses decision trees to build a model. In a random forest, multiple decision trees are trained on different subsets of the data, and the predictions of each tree are combined to produce a final prediction. Random forest regression can capture non-linear relationships between the dependent and independent variables and can handle high-dimensional data with many variables. It also has the advantage of being less prone to overfitting than single decision trees. Artificial neural networks (ANN) are a type of machine learning model that is inspired by the structure and function of the human brain. ANN consists of layers of interconnected nodes or neurons that process information and make predictions. In the context of predicting energy consumption by industry, ANN can be used to model complex non-linear relationships between the dependent and independent variables. ANN can handle large amounts of data and can learn from patterns in the data without being explicitly programmed. However, training an ANN can be computationally expensive and requires careful tuning of hyperparameters. In summary, linear regression, random forest regression, and ANN are all viable approaches for predicting energy consumption by industry..

Scene 5 (7m 29s)

[Audio] From the above graphs, it can be seen that ANN has highest R2 score and lowest MSE among all models. Although, R2 score is significantly similar for all models; the difference in other metric MSE points out that ANN is better than other two. The high R2 score implies that these models predict the energy consumption precisely and accurately. Also, there is no overfitting of data as explained above in modelling section. It can imply that right set of features are selected in feature extraction and steps of data preparation and data pre-processing have prepared the data well for modelling. It can be concluded that ANN performs better than other models and predict the energy consumption precisely and accurately..

Scene 6 (8m 21s)

[Audio] There could be next steps for future direction of research in this direction. The first is expanding the analysis to more locations and business sectors to evaluate how well the data mining models can be applied broadly. This can entail gathering information from many sources and assessing the models' performance in various scenarios. The second step could be analysing the underlying causes of energy use more thoroughly to find areas where energy can be saved, and efficiency can be increased. This can entail combining more data sources or running extra statistical analyses on the data. The last step could be analysing the effectiveness of the data mining models across extended time spans to determine their applicability in actual energy management scenarios. This could entail evaluating the viability of applying the models in practise and comparing the performance of the models to other prediction approaches or industry standards..

Scene 7 (9m 23s)

[Audio] Thank You for being here.. THANK YOU.