I am highly thankful to various researchers to post various forms of information at different platforms (books, papers, presentation, social media, google etc) that help in developing these presentations. The aim here is to provide information in most simple way to understand. I acknowledge the efforts of all.

Optimisation Algorithms:

Machine Learning:

Engineering:

Workshops:

Presentation:

  1. Probabilistic Modeling of Human Locomotion for Biped Robot Trajectory Generation (Bharat Singh, Vishu Gupta, Rajesh Kumar):The wheel-type robot has found numerous applications in hospitals, restaurants, entertainment, the automation industry, etc., and shows its applicability in solving the tasks efficiently. However, it failed to achieve the same efficiency in an unstructured environment that is mostly found in the real world. Thus, a biped robot can replace the wheel-type robot for better performance. The biped robot has many joints which make it a complex higher degree of freedom system. Hence, the designing of the controller, reference trajectory generation, state estimation and, filter design for feedback signal is a very cumbersome task. This paper focuses on the generation of the reference trajectories. Since human locomotion is optimal naturally, therefore, the human data is used for this study, which is collected at Robotics and Machine ANalytics (RAMAN) Lab, MNIT, Jaipur, India. In the literature, various authors have implemented model-based learning methods to develop a model based on data. However, these models suffer from model bias i.e., it is assumed that learned model accurately define the real system. Therefore, in this paper, the authors have proposed probabilistic models to model the human locomotion data. The reference trajectory is generated using the Bayesian ridge regression, Automatic relevance determination regression, and Gaussian process regression. The performance evaluation of developed models are based on average error, maximum error, root mean square error, and percentage normalized root mean square error. Download the presentation
  2. Mapping Model for Genesis of Joint Trajectory using Human Gait Dataset (Bharat Singh, Ankit Vijayvargiya, Rajesh Kumar): Walking gait trajectory genesis for the biped robot is a cumbersome task because of many degrees of freedom. In this paper, the authors have proposed machine learning models for kinematic modeling of human locomotion data. The human locomotion gait dataset has been taken from the MNIT gait dataset which is collected in previous work. The Gait dataset have contains the walking data of 120 subjects from different age-group. The machine learning models can become biased due to overfitting/underfitting. Thus, the K-fold cross-validation technique is employed for the training of machine learning models for mitigating biasing. In addition, two types of mappings have been developed i.e., one-to-one and many-to-one. One-to-one mapping has been used to map the knee, hip, and ankle trajectory to knee, hip, and ankle trajectory respectively. While many-to-one mapping has been used to map the combined trajectory of the knee, hip, and ankle to individual knee, hip, and ankle trajectory. The advantage of many-to-one is that it captures the connection between the knee, hip, and ankle efficiently. The accuracy of developed machine learning mapping is evaluated in terms of average error, maximum error, and root mean square error. The result shows that the Lasso family is performing the best among the developed models and also the many-to-one mapping outperforms the one-to-one mapping. Finally, an open discussion is presented for future research direction for gait generation and applications. Download the presentation
  3. Data Driven Kinematic Modeling of Human Gait for Synthesize Joint Trajectory (Bharat Singh, Ankit Vijayvargiya, Rajesh Kumar): Synthesis of reference joint trajectories for the legged robot is a very difficult task due to higher degrees of freedom. The gait dataset can be used to develop the models which can provide the required references. This paper presents the kinematic modeling of human gait data, which is used as the reference joint trajectory for a Biped robot, 8 deep learning models are proposed. Gait data-set of 120 subjects are collected at RAMAN Lab, MNIT Jaipur, India using the vision-based methodology. All subjects belong to the 5-60 years age group. Four type of novel mappings, one-to-one (knee-to-knee, hip-to-hip, and ankleto-ankle), many-to-one (knee+hip+ankle-to-knee/hip/ankle), oneto-many (knee/ankle/hip-to-knee+hip+ankle), and many-to-many (knee+hip+ankle-to-knee+hip+ankle), are also developed. These mapping provides the reference trajectories to biped robot and relationships between the knee/hip/ankle trajectories is also obtained. Performance evaluation of developed models is measured by average error, maximum error and root mean square error. Results show that the bidirectional deep learning technique performs better for different mappings. Finally, a discussion is provided for the applicability of developed mapping robots in real biped robots. Download the presentation
  4. Biped Robot Data-Driven Gait Trajectory Genesis for Multiple Inclines and Speed (Suchit Patel, Bharat Singh, Rajesh Kumar): This paper presents a data-driven gait model for continuous parameterization of joint kinematics which yields the genesis of biped robot trajectory. This work employed data-driven approaches such as Deep Neural Network (DNN) and Long Short Term Memory (LSTM) for parameterization using the human locomotion data-set which consists of 10-able subjects walking data on varying inclines and speeds. It allows a smooth and non-switching prediction surface which provides the reference gait trajectory. Additionally, to constrain the model from following the high variance points from the mean trajectory, a loss function that incorporates the standard error of the inter-subject mean is also proposed.  Performance evaluation shows that the LSTM performs far better than the DNN in terms of mean and max error for both trained and untrained data-set. Finally, the impact of varying speeds with an incline on the predicted kinematic trajectory for both models is also presented. Download the presentation
  5. Design and Analysis of Trajectory Tracking Controllers for Noisy 2-Link Robotic Manipulator (Vaishnavi J, Bharat Singh, Rajesh Kumar): In the real world, the tracking controllers designed for ideal scenarios will be inaccurate due to presence of noise. This paper deals with trajectory tracking control of a two-link manipulator using three different tracking controllers namely, Proportional Integral Derivative (PID), Adaptive Neuro-Fuzzy Inference System (ANFIS), and Back-stepping controller in a noisy environment. Here, the PID controller is designed using the singular perturbation technique, it helps in dealing with external noise. However, its performance is not good with variable noise. Thus, the ANFIS controller is developed in which the rules are based on data collected from previously designed PID controller. Lastly, the back-stepping controller based on the virtual control signal using Lyapunov for zero error tracking is developed. The three controllers for the manipulator have been tested with the introduction of variable random noise and a fixed noise quantity. Performance analysis of these controllers is based on ISE (Integral Square Error), IAE (Integral Absolute Error), ITSE (Integral Time Squared Error), and ITAE (Integral Time Absolute Error). The simulation results illustrate the accuracy of the ANFIS controller which has better tracking in comparison to the other two control schemes with comparable torque inputs. Download the presentation
  6. Inverse Kinematics Solution for 5-DoF Robotic Manipulator using Meta-heuristic Techniques (Vaishnavi J, Bharat Singh, Ankit Vijayvargiya, Rajesh Kumar): Robotic manipulators have become the key component in the automation industry because of their accuracy. Finding the inverse kinematic (IK) solution (to compute the joint angles for desired end-effector position and orientation) for a robot manipulator is the most challenging problem. Conventional approaches like numerical, geometric, and algebraic have failed to provide the solution due to redundancy/singularity present in IK. This paper presents the IK solution for the 5-DoFs manipulator using fourteen meta-heuristic techniques. The end-effector position is determined by solving Forward kinematics using the Denavit–Hartenberg (DH) parameters. The objective function is designed to minimize the Euclidean distance between the actual and desired position/orientation of the end-effector. Comparative analysis of these techniques is based on the computation time and positional error. Result shows that the differential evolution (DE) algorithm outperforms all other techniques in terms of Cartesian (4.42598×10-8 cm) and orientation (4.4259810×10-8 rad) error. Whereas the grey wolf optimization (GWO) algorithm outperforms all in terms of computation time (0.308856 sec). Download the presentation
  7. Human activity recognition using accelerometer and gyroscope data from smartphones
    (Khimraj, Praveen Kumar Shukla, Ankit Vijayvargiya, Rajesh Kumar): Human Activity Recognition is a procedure for arranging the activity of an individual utilizing responsive sensors of the smartphone that are influenced by human activity. Its standouts among the most significant building blocks for numerous smartphone applications, for example, medical-related applications, tracking of fitness, context-aware mobile, survey system of human, and so forth. This investigation centers around acknowledgment of human activity utilizing sensors of the smartphone by some machine learning and deep learning characterization approaches. Data received from the accelerometer sensor and gyroscope sensor of the smartphone are grouped to recognize the human activity. Download the presentation
  8. Comparative analysis of machine learning techniques for the classification of knee abnormality (Ankit Vijayvargiya, Rajesh Kumar, Nilanjan Dey, João Manuel RS Tavares):
    Knee abnormality is a major problem in elderly people these days. It can be diagnosed by using Magnetic ResonanceImaging (MRI) or X-Ray imaging techniques. X-Ray is only used for primary evaluation, while MRI is an efficient way to diagnose knee abnormality, but it is very expensive. In this work, Surface EMG (sEMG) signals acquired from healthy and knee abnormal individuals during three different lower limb movements: Gait, Standing and Sitting, were used for classification. Hence, first Discrete Wavelet Transform (DWT) was used for Denoising the input signals; then, eleven different time-domain features were extracted by using a 256 msec windowing with 25% of overlapping. After that, the features were normalized between 0 (zero) to 1 (one) and then selected by using the backward elimination method based on the p-value test. Five different machine learning classifiers: k-nearest neighbor, support vector machine, decision tree, random forest and extra tree, were studied for the classification step. Our result shows that the Extra Tree Classifier with ten cross-validations gave the highest accuracy (91%) in detecting knee abnormality from the sEMG signals under analysis. Download the presentation
  9. Implementation of Machine Learning Algorithms For Human Activity Recognition (Ankit Vijayvargiya, Nidhi Kumari, Palak Gupta, Rajesh Kumar): Human Activity Recognition (HAR) is technically the problem of forecasting an individual’s actions based on evidence of their gesture using sensors functioning as accelerometer and gyroscope. It plays a major role in contrasting sectors such aspersonal biometric signature, daily life monitoring, anti-terrorists along with anti-crime securities, medical-related applications, and so on. These days, smart phones are well-resourced with leading processors and built-in sensors. This comes up with the possibility to unfold a new arena of data mining. This paper signifies the analysis of HAR focused on data composed via accelerometer sensors of smart phones. Further, it illustrates the use of time-domain features which are acquired with the help of a windowing approach termed as overlapping. It is accompanied by a window size of 250ms along with overlapping of 25%. Numerous machine learning classifiers such as k-nearest neighbors, linear discriminant analysis, bagging classifier, gradient boosting classifier, decision tree, random forest, and support vector machine usingthree different kernels were practiced. The outcomes exhibit that random forest with 5-fold cross-validation imparts the highest accuracy (92.71%) in recognition of human activities. Download the presentation
  10. A Comparative Assessment of Machine Learning Techniques for Epilepsy Detection using EEG Signal (Balan Dhanka, Ankit Vijayvargiya, Rajesh Kumar, Ghanshyam Singh): Epilepsy is a psychological issue that causes ridiculous, repetitive seizures. A seizure is an unexpected surge of
    electrical activity in the cerebrum. Since traditional time or recurrence, area examination is discovered insufficient to portray the qualities of non-fixed signals, for example, electroencephalography (EEG) signal. In this paper, we propose to change the EEG information utilizing Detrended Fluctuation Analysis, Singular Value Decomposition Entropy, Higuchi Fractal Dimension, and Petrosian Fractal Dimension works and plan vectors. These vectors are utilized as highlights for ten distinctive Machine Learning classifiers: Logistic Regression, Support Vector Classifier, Linear Support Vector Classifier, K Nearest Neighbors Classifier, Naïve Bayes Classifier, Random Forest Classifier, Decision Tree Classifier, Stochastic Gradient Descent Classifier, Multilayer Perceptron Classifier, Gaussian Process Classifier, to group epileptic seizurestarting from various parts and condition of the cerebrum. Our outcomes show that the proposed strategy gives better exactness, accuracy, and review in contrast with the current strategies for multiclass Epileptic Seizure Classification. Download the presentation
  11. Deep Learning Frameworks for COVID-19 Detection(Ankit Vijayvargiya, Akshit Panchal, Abhijeet Parashar, Ayush Gautam, Jayesh Sharma, Rajesh Kumar): The COVID-19 (previously known as “2019 novel coronavirus”) took the big form and outspread rapidly around the world becoming a pandemic. Artificial intelligence tools come out to be one of the fastest solutions to detect the disease and in another way helping to control the spread. This paper signifies how chest X-ray images use deep learning techniques which are very useful for analyzing images to detect the virus and spotting high-risk patients for controlling the spread. Further, it shows how the Convolutional Neural Network (CNN) technology of deep learning helps to detect the virus quickly. A CNN is a type of artificial neural network that is used for image pre-processing and consists of many layers that aid in detection. A sequential CNN model is proposed with different kernel sizes, filters, and having different parameters using a dataset of 2159 images. The output shows that a model with an adequate amount of filters, max-pooling layers, dropout layers and dense layers imparts the highest accuracy of 99.53% in detecting the coronavirus. Download the presentation
  12. WD-EEMD based Voting Classifier for hand gestures classification using sEMG signals (Puru Lokendra Singh, Samidha Mridul Verma, Ankit Vijayvargiya, Rajesh Kumar): In the biomedical field, there are many applications available based on surface EMG (sEMG) signal classification such as human-machine interaction, diagnosis of kinesiological studies and neuromuscular diseases. However, these signals are complicated because noise is generated during the recording of the sEMG signal. In this study, a hybridization of two signal pre-processing techniques, Wavelet Decomposition and Ensemble Empirical Mode Decomposition, called WD-EEMD with Voting classifier, is introduced to classify hand gestures based on sEMG signals. A study of different Decision Tree ensembles has been done for the classification process. Signals are preprocessed, segmented and then classified after extracting relevant features from them. The final prediction of the signal’s class is done via a voting mechanism. Different studied pre-processing techniques, similar to that of the proposed methodology with different classifiers have been compared. A new performance metric called confidence has been introduced to analyze the classification procedure. The models have been evaluated and compared on performance criteria like accuracy and overall confidence (gross and true confidence). It has been observed that Gradient Tree Boosting along with WD-EEMD gives the best classification accuracy with high confidence. Download the presentation
  13. Machine Learning based Risk Classification of Musculoskeletal Disorder among the Garment Industry Operators (Aastha Arora, Ankit Vijayvargiya, Rajesh Kumar, Manoj Tiwari): The occurrence of work-related injury risks is extremely high in the garment industry but often ignored. These disorders not only damage the physical health of the workers but also proves to be a prominent factor while talking about loss in work time; ultimately leading to low productivity and efficiency. This paper presents a systematic approach to predict the automated diagnosis of musculoskeletal disorder among the sewing machine operators of the garment industry. The working videos of 20 participants- 10 healthy (normal) and 10 unhealthy (abnormal) were recorded from both sides- left and right. For posture evaluation, OpenPose algorithm is applied to estimate 2D human pose and to extract the joint angles of neck, trunk, upper arm and lower arm of both left and right sides, using the python math library. The extracted angles were then normalized between the range 0 (zero) to 1 (one) to prepare a classification model using the KNN Classifier. Stratified k-fold cross-validation was implemented using 10 folds which gave the accuracy of 91.3% in diagnosing the musculoskeletal disorder among the sewing machine operators. Download the presentation
  14. Comparative Assessment among Different Convolutional Neural Network Architectures for Alzheimer’s disease Detection (Gargi Sharma, Ankit Vijayvargiya, Rajesh Kumar): Alzheimer’s disease (AD) is a type of Dementia affecting the brain cells. An intense amount of research has been done on the subject of Alzheimer’s disease Detection and even now huge amounts of research are going on towards this subject. Over time different deep learning models have been implemented including various transfer learning models. In this work a comprehensive analysis of eight transfer learning models has been done to classify AD in 4 classes; Non-Demented, Very Mild-Demented, Mild-Demented and Moderate Demented. The transfer learning models implemented here are: DenseNet-169, Inception ResNet-V2, MobileNet-V2, ResNet-101, Inception-V3, ResNet-50, VGG-16, VGG-19. The transfer learning model with the highest scores gave an accuracy of 98.0%and precision of 98.01%, which is a good score for medical imaging problems.Download the presentation
  15. Centralize Energy Storage Scheduling for Prosumers in Residential Microgrid (Anita Seervi, Vikash Kumar Saini, Rajesh Kumar, M A Mahmud): The residential microgrid is being governed by local distributed energy resources and energy storage. Storage offers various benefits for residential grid operations, such as peak shaving, demand shift. When electricity prices are high during peak demand hours at the same time, storage will support reducing the electricity bill of prosumers. However, scheduling energy storage devices is challenging due to uncertainty in renewable energy generation. This paper provides an analysis of distributed and centralized energy storage. The charging and discharging scheduling of energy storage for residential applications has been performed as a linear programming approach. The cost of electricity has been calculated, keeping peak demand within the specified limit. The impact of seasonal on PV profile is also considered. The Result found that centralized storage is more suitable for residential applications.Download the presentation
  16. Cloud Energy Storage Systems for Consumers and Prosumers in Residential Microgrids (Vikash Kumar Saini, Vishu Gupta, Rajesh Kumar, B. K. Panigrahi, M. A. Mahmud): Distributed energy storage systems (DESSs) have huge potential to balance distributed renewable power generation and load demands for consumers of prosumers. DESSs are capable to reduce barriers by eliminating intermittencies in distributed renewable energy sources in microgrids. Since the electricity prices are higher during the peak hours, DESSs can be used to store energy during residential off-peak (but solar peak) hours and utilized this energy during residential peak hours for reducing electricity bills of consumers or prosumers in microgrids. However, energy storage systems are still expensive components for residential microgrids and these need to be effectively utilized in order to provide a cost-effective solution. In continuous analysis economic benefit also carried out. Maximum rate of return on investment, maximum profit margin, and annual revenue are includes in economic analysis. This paper presents an alternative solution, a cloud energy storage system (CESS) for effectively utilizing DESSs in residential microgrids while reducing both electricity bills and installation costs for ESSs. This work presents an analysis on the feasibility and profit ability of CESSs.Download the presentation
  17. Renewable Energy Forecasting for Energy Storage Sizing: A Review (Anita Seervi, Vikash Kumar Saini, Rajesh Kumar, M A Mahmud): The increasing low carbon technologies in electricity generation (i.e. wind and solar), the system operator faced various challenges due to their intermittent and fluctuating nature. To address these challenges, energy storage have a potential solution. The optimal size of energy storage is most crucial issue at the planning stage. Renewable energy forecast error affects the storage sizing and scheduling. However, accurate forecasting is still a challenge in the perspective of energy storage. Therefore, this review paper explores the forecasting models with sizing of storage methods for enhancing the penetration of renewable energy generation. Further, the energy storage application in power system categorized based on specific area i.e. bulk energy services, ancillary services, power quality management, transmission and distribution infrastructure services and local energy management services also discussed. Download the presentation
  18. Data Driven Temperature Estimation of PMSM with Regression Models (Parul Sharma, Ankit Vijayvargiya, Bharat Singh, Rajesh Kumar): Monitoring temperature parameters in Permanent Magnet Surface Machine (PMSM) is crucial since the machine possesses applications in several diverse areas, for instance, traction drives, electric vehicles, besides applications that require control of rotor temperature in order to ensure safety and cost effectiveness. The thermal losses produced in the permanent magnet synchronous machine such as copper, iron, and mechanical loss and the cooling modes are responsible for temperature rise in the PMSM. The basic method to evaluate the temperature of internal components like Lumped Parameter Thermal Networks (LPTN) does not provide the degree of freedom in terms of choosing model parameters, physical comprehensibility, and real-time requirements. Hence, in this work a real-time data collected from the Germany laboratory is considered to perform the various machine learning techniques in which the effect of stator temperature, coolant temperature, and ambient temperature are contemplated to estimate the permanent magnet surface temperature. This paper has accomplished a data driven temperature estimation of four regression models that are Linear Regression, Stochastic Gradient Regression, Random Sample Consensus (RANSAC) Regression, and Random ForestRegression training models by considering Mean Absolute Error (MAE), Mean Square Error (MSE), and R2-Score as measuring parameters and for unambiguous understanding, the predicted and the actual results are elaborated. This study is evident that Random Forest Regression model is proved to have the best outcomes out of all four regression models exercised in this research. Download the presentation
  19. Optimal Number of Wind Turbine in Farm Layout with Power Maximization (Vikash Kumar Saini, Bharat Singh, Dinesh Kumar Mahto, Akhilesh Mathur, Rajesh Kumar): Designing of wind farm layout is complex and challenging task due to its space and cost associated constraints. Usually, turbines are positioned in proximity. Thus, the generated power of downstream turbine is affected by the wake effect produced by upstream turbine. This is challenge to power system planning and reliability forecasting. Therefore, it is a non-linear constraint optimization problem. Thus, in this paper, differential evolution strategy is used to find optimal positioning and number of wind turbine for maximization of power generation incorporating wake effect. Strategy implemented have two characteristics: (a) dimension is reduced to two (b) population size parameter is eliminated. Results provide the optimal number of wind turbine in farm layout with maximization of power output. Download the presentation
  20. Gated Recurrent Unit (GRU) Based Short Term Forecasting for Wind Energy Estimation (Vikash Kumar Saini, Bhawana Bhardwaj, Vishu Gupta, Rajesh Kumar, Akhilesh Mathur): Penetration of renewable energy generation is increasing rapidly due to the increasing demand of the grid. From the power system operation and economy point of view, wind power integration plays a vital role in the emerging grid. However, due to the presence of uncertainty in the wind speed, its prediction is also a significant factor in the integration of wind power. The accuracy of wind speed prediction can help power system operator to overcome the risk of unreliability power supply. Although several data prepossessing and prediction approaches have been described in the literature, these approaches facing the problem of high accuracy. To solve this problem, this paper presents the various machine learning algorithms for accurate forecasting of wind speed and the consequent estimation of generated energy. All algorithms are applied to hourly wind speed data for the region of Jodhpur, Rajasthan (India). The numerical results indicating that the GRU performance outperformed them. Download the presentation
  21. Predictive Analysis of Traditional, Deep Learning and Ensemble Learning Approach for short-term Wind Speed Forecasting (Vikash Kumar Saini, Fairy Mathur, Vishu Gupta, Rajesh Kumar): Renewable energy is increasing rapidly to reduce fossil fuel consumption, due to environmental pollution and limited availability of fossil fuels. Several renewable energy sources, such as wind, solar, tidal are available in nature. However, harvesting wind energy is significantly more carbon free. The intermittent nature of speed is a big challenge to power dispatch in the grid network. The pattern of consumption of is a new challenge which requires an intelligent grid for reliable operation. The wind speed forecast may be a solution of this problem from system operator aspect. Various forecasting models are present in the literature which offer methods for accurate forecasting. Traditional and AI based i.e. Ensemble & Deep Learning models are presented in this paper for the one year wind data set. The models are trained on 70% data and the rest of 30% data is used for testing. Ensemble learning model i.e. XGBoost results are compared to ARIMA, LSTM and Random Forest. The performance of the models has been examined using error indices such as Mean Absolute Error (MAE), Mean Square Error (MSE) and Root Mean Square Error (RMSE) on the training and testing datasets. Although all of the models produced good outcomes, the results indicate enhanced performance of XGBoost in comparison to the other techniques. Download the presentation
  22. Short term forecasting based on hourly wind speed data using deep learning algorithms (Vikash Kumar Saini, Rajesh Kumar, Akhilesh Mathur, Akash Saxena): Application of Internet of Things (IoT) in smart grid is evident in current trends. Smart grid management have greater impact on market economics, security, and distribution of energy. Smart grid is an integration of several components like, wind, solar, cyber-security etc. One of major concern in smart grid is optimal control of wind generation and accurate prediction of wind speed. This paper aims to predict the wind speed with meteorological time series data as input variable using deep learning topology for one-year wind speed data. The dynamic recurrent type network (RNN) integrates and processed with the Extreme Learning-Machine (ELM), nonlinear autoregressive network with exogenous inputs (NARX), and Long short-term memory (LSTM) model. Three models having the same Network’s architecture, intermediate layer in architecture have 19 neurons and an activation function. Feature selection method is used for feature extraction from wind data (have four features as wind speed, pressure, humidity, air temperature) and applied to models. Comparative analysis of different models are assessed by performance matrices such as MAPE, MAE, and RMSE. Download the presentation
  23. Modelling and Analysis of Home Energy Management System Using Intelligent Algorithms (Srinivas Yelisetti, Rajesh Kumar, Vishu Gupta, Akash Saxena, Ravita Lamba): In the Home Energy Management System (HEMS), mathematical optimization has been approved to be a convincing tool for reducing the energy consumption. This HEMS approach considers the mathematical modelling of the heating, ventilation, air conditioning (HVAC) and lighting system with occupant’s comfortable thermal conditions. In this paper, the proposed model is implemented for standalone residence. Further, to minimize energy consumption, intelligent algorithms and CPLEX solver in YALMIP toolbox have been applied to the formulated objective. The results have been analysed in terms of energy consumption and computation time consumed by different algorithms and toolbox. The results aided in determining the most promising optimization approach among the different algorithms and toolbox. Eventually, research was helped to assess the efficacy as well as the adaptability of the propounded framework. Download the presentation
  24. Modelling and Simulation of Home Energy Management System with Occupants Comfort (Srinivas Yelisetti, Rajesh Kumar, Ravita Lamba, Akash Saxena): Energy conservation and occupant comfort have become important aspects of residential buildings. The building’s energy use is greatly influenced by occupant behaviour through the use of heating, ventilation, and air conditioning (HVAC), lighting, etc. to improve occupant comfort inside the building. A home energy management system (HEMS) can be implemented to minimize building energy consumption while maintaining occupant comfort. In this paper, a detailed thermal and electrical model of HEMS has been developed to achieve efficient energy usage and occupant comfort simultaneously. In the proposed model, the HVAC system has been operated by considering all internal heat gains. The proposed model described the thermal, visual, and air quality comforts of the occupants. The Particle Swarm optimization (PSO) algorithm has been applied to the proposed model and the existing model. Finally, the proposed model has been compared with the existing model to validate the effectiveness of the proposed model. Download the presentation
  25. Performance Analysis of Comfort Maximization Model with Five Different Weather Conditions in India (Srinivas Yelisetti, Rajesh Kumar, Ravita Lamba, Akash Saxena): In households, maintaining the occupant’s comfort within a comfortable range is very important. In this paper, the mathematical model of three key aspects of occupant’s comfort, which are thermal, air quality and visual comfort, has been developed and analysed in MATLAB. Also, the proposed model has been analyzed for five different cities’ weather conditions in India: Jaipur, Delhi, Mumbai, Kolkata, and Chennai. The presented model is applied to a room in a single-storey standalone residential building. Further, the Particle Swarm Optimization (PSO) algorithm is applied to maximise the occupant’s comfort. To maximise the overall comfort, the proposed model has been implemented in two cases, i.e. equal priority and variable priority for three individual comforts. The results for a typical household show that optimisation can achieve the desired comfort. Download the presentation
  26. Directed Bee Colony Optimization (DBC): The paper presents a new optimization algorithm inspired by group decision-making process of honey bees. The honeybees search for the best nest site among many possible sites taking care of both speed and accuracy. The nest site selection is analogous to finding the optimality in an optimization process. Such similarities between two processes have been used to cultivate a new algorithm by learning from each other. Various experiments have been conducted for better understanding of the algorithm. A comprehensive experimental investigation on the choice of various parameters such as number of bees, starting point for exploration, choice of decision process etc. has been made, discussed and used to formulate a more accurate and robust algorithm. The proposed Directed Bee Colony algorithm (DBC) has been tested on various benchmark optimization problems. To investigate the robustness of DBC, the scalability study is also conducted. The experiments conducted clearly show that the DBC generally outperformed the other approaches. The proposed algorithm has exceptional property of generating a unique optimal solution in comparison to earlier nature inspired approaches and therefore, can be a better option for real-time online optimization problems.(Composed by Rajesh Kumar) Download paper   Download the code
  27. Termite Spatial Correlation based Particle Swarm Optimization (TSC-PSO): This  proposes a new Termite Spatial Correlation based Particle Swarm Optimization (TSC-PSO) algorithm inspired by the movement strategy shown within Termites (Cornitermes cumulans). TSC-PSO modifies the velocity equation in the original PSO algorithm by replicating the step correlation based termite motion mechanism that exhibits individually in nature and works with decentralized control to collectively perform the overall task. Further, the algorithm incorporates the mutation strategy within it to make it suitable to avoid stagnation conditions while performing optimization in complex search spaces. For deriving its utility various benchmark functions of different geometric properties have been used. Experiments clearly demonstrate the success of the proposed algorithm in different benchmark conditions against various state-of-the-art optimization algorithms. (Composed by My Team consisting of Avinash Sharma, B. K. Panigrahi and Swagtam Das) Download paper   Download the code
  28. Structured Clanning-based Ensemble Optimization (SCEO): inspired by the social organization of the Elephant clan is proposed for solving complex numerical optimization problems. The proposed algorithm is inspired by the complex and diversified behaviour present within the fission-fusion-based social structure of the elephant society. The population of elephants can consist of various groups with relationship between individuals ranging from mother-child bond, bond groups, independent males, and strangers. The algorithm tries to model this individualistic behaviour to formulate an ensemble-based optimization algorithm. The algorithm performance on these test benchmarks is compared to various state-of-the-art optimization algorithms. Experiments clearly showcase the success of the proposed algorithm in optimizing the benchmark functions to better values. (Composed by My Team consisting of Avinash Sharma,  Akash Saxena and B. K. Panigrahi) Download paper    Download the code
  29. Framework for Generation and Detection of P300 waveforms: P300 is a unique signal elicited by the brain on receiving a specific stimulus. It is most commonly known for being used as a EEG based speller, where the user can, simply with his focus, spell a word. Since the EEG data consisted of a lot of data to work with, we ended up with a large variety and quantity of features to experiment with. To enable modular organization, fast code execution and stacked feature vectors we setup a framework which utilisescaching, parallel computing and modularity. Finally after intensive testing we obtained highly accurate results, competitive with work done previously. (Composed by My Team  Mr. A. Kumar, Mr. S. Shah ) Download paper   Download the code
  30. Backtracking Search Optimization based Neural Network (BSANN): BSANN is a complex stochastic search based Evolutionary Algorithm which smartly backtracks from new to old populations during its evolution to reduce error caused from entering a non-ideal search space. It is based on a multilayer Neural Network Architecture. It has shown excellent results in the domain of EEG pattern detection beating the results of best available algorithms(including winner of the BCI competition) in the field of Motor Imagery BCI. A python library implementation of both the original BSA as well as BSA-NN supporting parallel processing has been attached below. (Composed by My Team  Mr. S. Shah, Mr. S. Agarwal) Download paper    Download the code
  31. Vectorized Particle Swarm Optimization Algorithm: PSO code presented here is based on basic swarming techniques where the global and the personal best solution of agents lead to the global best position according to the problem. The codes of PSO and IPSO (Inertial Particle Swarm Optimization) are presented here and benchmark functions such as Ackley, FoxHoles, Rosenbroch etc. are given along the codes for verification. The code has been vectorized in MATLAB so as to make it even faster. Multiple steps have been reduced resulting in higher efficiency and speed of the algorithm. The matlab files for the PSO toolbox are given below for download. The syntaxes and try run codes are presented alongside. (Composed by My Team consisting of Mr. B. P. Singh @MNIT) Download the code
  32. Download Benchmark functions: Rosenbrock,  Restrigin,  Griewank, FoxHoles,  Ackley
  33. Gummi : It is a LaTeX editor for the Linux platform, written in C/GTK+. It was designed with simplicity and develops pdf online structure. Gummi was released as free opensource software under the MIT license. (@Gummi) Download the software
  34. Scilab : Scilab is free and open source software for numerical computation providing a powerful computing environment for engineering and scientific applications. It supports Windows, Linux and MAC OSX. (@Scilab) Download the software
  35. Apache OpenOffice : Apache OpenOffice is the leading open-source office software suite for word processing, spreadsheets, presentations, graphics, databases and more. Apache OpenOffice is developed 100% by volunteers. It is compatible with other major office suites. (@Apache OpenOffice) Download the software