Katarina Grolinger | University of Western Ontario (original) (raw)

Papers by Katarina Grolinger

IECON 2023- 49th Annual Conference of the IEEE Industrial Electronics Society, 2023

In recent years, electric vehicles (EVs) have been widely adopted because of their environmental ... more In recent years, electric vehicles (EVs) have been widely adopted because of their environmental benefits. However, the increasing volume of EVs poses capacity issues for grid operators as simultaneously charging many EVs may result in grid instabilities. Scheduling EV charging for grid load balancing has a potential to prevent load peaks caused by simultaneous EV charging and contribute to balance of supply and demand. This paper proposes a user-preference-based scheduling approach to minimize costs for the user while balancing grid loads. The EV owners benefit by charging when the electricity cost is lower, but still within the user-defined preferred charging periods. On the other hand, the approach reduces the pressure on the grid by balancing the grid load. Two methods, the greedy algorithm and nonlinear programming, are considered along with users’ charging preferences and durations. For scheduling small numbers of charging activities, the nonlinear programming model achieves better load balancing than the greedy algorithm; however, for scheduling medium to large numbers of charging activities, the greedy algorithm has a clear advantage in terms of time complexity.

Neural Computing and Applications, Apr 21, 2022

International Journal of Electrical Power & Energy Systems

Energies

The transition to Electric Vehicles (EV) in place of traditional internal combustion engines is i... more The transition to Electric Vehicles (EV) in place of traditional internal combustion engines is increasing societal demand for electricity. The ability to integrate the additional demand from EV charging into forecasting electricity demand is critical for maintaining the reliability of electricity generation and distribution. Load forecasting studies typically exclude households with home EV charging, focusing on offices, schools, and public charging stations. Moreover, they provide point forecasts which do not offer information about prediction uncertainty. Consequently, this paper proposes the Long Short-Term Memory Bayesian Neural Networks (LSTM-BNNs) for household load forecasting in presence of EV charging. The approach takes advantage of the LSTM model to capture the time dependencies and uses the dropout layer with Bayesian inference to generate prediction intervals. Results show that the proposed LSTM-BNNs achieve accuracy similar to point forecasts with the advantage of pre...

2022 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

This paper introduces the Virtual Sensor Middleware (VSM), which facilitates distributed sensor d... more This paper introduces the Virtual Sensor Middleware (VSM), which facilitates distributed sensor data processing on multiple fog nodes. VSM uses a Virtual Sensor as the core component of the middleware. The virtual sensor concept is redesigned to support functionality beyond sensor/device virtualization, such as deploying a set of virtual sensors to represent an IoT application and distributed sensor data processing across multiple fog nodes. Furthermore, the virtual sensor deals with the heterogeneous nature of IoT devices and the various communication protocols using different adapters to communicate with the IoT devices and the underlying protocol. VSM uses the publishsubscribe design pattern to allow virtual sensors to receive data from other virtual sensors for seamless sensor data consumption without tight integration among virtual sensors, which reduces application development efforts. Furthermore, VSM enhances the design of virtual sensors with additional components that support sharing of data in dynamic environments where data receivers may change over time, data aggregation is required, and dealing with missing data is essential for the applications.

2022 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

Energy companies often implement various demand response (DR) programs to better match electricit... more Energy companies often implement various demand response (DR) programs to better match electricity demand and supply by offering the consumers incentives to reduce their demand during critical periods. Classifying clients according to their consumption patterns enables targeting specific groups of consumers for DR. Traditional clustering algorithms use standard distance measurement to find the distance between two points. The results produced by clustering algorithms such as Kmeans, K-medoids, and Gaussian Mixture Models depend on the clustering parameters or initial clusters. In contrast, our methodology uses a shape-based approach that combines Agglomerative Hierarchical Clustering (AHC) with Dynamic Time Warping (DTW) to classify residential households' daily load curves based on their consumption patterns. While DTW seeks the optimal alignment between two load curves, AHC provides a realistic initial clusters center. In this paper, we compare the results with other clustering algorithms such as K-means, K-medoids, and GMM using different distance measures, and we show that AHC using DTW outperformed other clustering algorithms and needed fewer clusters.

2022 IEEE Conference on Dependable and Secure Computing (DSC)

Engineering Applications of Artificial Intelligence

There is an increasing demand for using Unmanned Aerial Vehicle (UAV), known as drones, in differ... more There is an increasing demand for using Unmanned Aerial Vehicle (UAV), known as drones, in different applications such as packages delivery, traffic monitoring, search and rescue operations, and military combat engagements. In all of these applications, the UAV is used to navigate the environment autonomously-without human interaction, perform specific tasks and avoid obstacles. Autonomous UAV navigation is commonly accomplished using Reinforcement Learning (RL), where agents act as experts in a domain to navigate the environment while avoiding obstacles. Understanding the navigation environment and algorithmic limitations plays an essential role in choosing the appropriate RL algorithm to solve the navigation problem effectively. Consequently, this study first identifies the main UAV navigation tasks and discusses navigation frameworks and simulation software. Next, RL algorithms are classified and discussed based on the environment, algorithm characteristics, abilities, and applications in different UAV navigation problems, which will help the practitioners and researchers select the appropriate RL algorithms for their UAV navigation use cases. Moreover, identified gaps and opportunities will drive UAV navigation research. Symbol Description ∈ State belongs to all possible states ∈ Action belongs to the set of all possible Actions ∈ Reward belongs to the set of all generated Rewards Discounted factor decreases the contribution of the future rewards, where < < The Expected Summation of the Discounted Rewards; = ∑ ∞ = + + (′ , | ,) The probability of the transition to state ′ with reward from taking action in state at time A trajectory consists of a sequence of actions and states pairs, where the actions influence the states, also called an episode. Each trajectory has a start state and ends in a final state that terminates the trajectory (,) Action-value function expresses the expected return of the state-action pairs (,); (,) is (,) parameterized by

Sensors

The deterioration of infrastructure’s health has become more predominant on a global scale during... more The deterioration of infrastructure’s health has become more predominant on a global scale during the 21st century. Aging infrastructure as well as those structures damaged by natural disasters have prompted the research community to improve state-of-the-art methodologies for conducting Structural Health Monitoring (SHM). The necessity for efficient SHM arises from the hazards damaged infrastructure imposes, often resulting in structural collapse, leading to economic loss and human fatalities. Furthermore, day-to-day operations in these affected areas are limited until an inspection is performed to assess the level of damage experienced by the structure and the required rehabilitation determined. However, human-based inspections are often labor-intensive, inefficient, subjective, and restricted to accessible site locations, which ultimately negatively impact our ability to collect large amounts of data from inspection sites. Though Deep-Learning (DL) methods have been heavily explor...

Industry hires engineers primarily for solving workplace problems; consequently problem solving s... more Industry hires engineers primarily for solving workplace problems; consequently problem solving skills are an essential part of an engineering education. However, industry problems, as well as the environment engineers work in, are often quite different than what students experience at universities. This workshop explores problem based learning, the differences between problems students typically solve in the classroom and the workplace, as well as the strategies for making classroom problems emulate real-world workplace problems. As the main goal of engineering education is to prepare students for work in industry, closing the gap between classroom and workplace problems will result in better prepared graduates

2021 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE), 2021

The desire to make applications and machines more intelligent and the aspiration to enable their ... more The desire to make applications and machines more intelligent and the aspiration to enable their operation without human interaction have been driving innovations in neural networks, deep learning, and other machine learning techniques. Although reinforcement learning has been primarily used in video games, recent advancements and the development of diverse and powerful reinforcement algorithms have enabled the reinforcement learning community to move from playing video games to solving complex real-life problems in autonomous systems such as self-driving cars, delivery drones, and automated robotics. Understanding the environment of an application and the algorithms' limitations plays a vital role in selecting the appropriate reinforcement learning algorithm that successfully solves the problem on hand in an efficient manner. Consequently, in this study, we identify three main environment types and classify reinforcement learning algorithms according to those environment types. Moreover, within each category, we identify relationships between algorithms. The overview of each algorithm provides insight into the algorithms' foundations and reviews similarities and differences among algorithms. This study provides a perspective on the field and helps practitioners and researchers to select the appropriate algorithm for their use case.

Industry hires engineers primarily for solving workplace problems; consequently problem solving s... more Industry hires engineers primarily for solving workplace problems; consequently problem solving skills are an essential part of an engineering education. However, industry problems, as well as the environment engineers work in, are often quite different than what students experience at universities. This workshop explores problem based learning, the differences between problems students typically solve in the classroom and the workplace, as well as the strategies for making classroom problems emulate real-world workplace problems. As the main goal of engineering education is to prepare students for work in industry, closing the gap between classroom and workplace problems will result in better prepared graduates.

Montaža je podrucje koje ima znatan udio u troskovima i vremenu proizvodnje. Stoga su danasnji na... more Montaža je podrucje koje ima znatan udio u troskovima i vremenu proizvodnje. Stoga su danasnji napori usmjereni k unapređenju montaže, posebno fleksibilne montaže uz primjenu robota. Ovaj rad se bavi razvojem inteligentnog robotskog sustava za primjenu u montaži, koji je sposoban rjesavati razlicite zadace prilagođavajuci svoje ponasanje promjenljivom okolisu, ukljucujuci snalaženje u neizvjesnim i nepredvidivim situacijama. Da bi to postigao robot mora moci naci rjesenje u nepoznatoj situaciji, uciti ponasanje, sto znaci procedure ponasanja zajedno s odgovarajucim znanjem o strukturi radnog prostora, i prepoznati radni prostor. Problem inteligentnog robotskog ponasanja se suocava s nepredvidivo velikim brojem varijanti problema sto ne može biti obuhvaceno egzaktnom domenom znanja, stoga se razvoj zasniva na aktivnim metodama ucenja. Sustav primjenjuje metodu prisilnog ucenja baziranu na strategijskim i slucajnim pokusajima za pronalaženje rjesenja problema i neuronsku mrežu za pamc...

International Journal of Electrical Power & Energy Systems, 2021

Load forecasting is essential for energy management, infrastructure planning, grid operation, and... more Load forecasting is essential for energy management, infrastructure planning, grid operation, and budgeting. Large scale smart meter deployments have resulted in ability to collect massive energy data and have created opportunities in sensor-based forecasting. Machine learning (ML) has demonstrated great successes in sensor-based load forecasting; however, when prediction is needed on a smart meter level, typically a single model is trained for each smart meter. With a large number of meters, this becomes computationally expensive or even infeasible. On the other hand, with conventional ML, training a single model for several smart meters requires participants to share their data with the central server. Consequently, this paper proposes federated learning for load forecasting with smart meter data: this strategy enables training a single model with all participating smart meters without the need to share local data. Two alternative federated learning strategies are examined: FedSGD, which performs one step of gradient descent on client before merging updates on the server, and Fe-dAVG, which carries out several steps before the merging. Specifically, residential consumers are diverse what makes training a single model challenging as load profiles vary across consumers. The results show that the FedAVG achieves better accuracy than FedSGD while also requiring fewer communication rounds. Comparing to individual models for each meter and a single central models for all meters, FedAVG achieves comparable or better accuracy.

2019 IEEE Canadian Conference of Electrical and Computer Engineering (CCECE), 2019

Sensors, wearables, mobile and other Internet of Thing (IoT) devices are becoming increasingly in... more Sensors, wearables, mobile and other Internet of Thing (IoT) devices are becoming increasingly integrated in all aspects of our lives. They are capable of collecting massive quantities of data that are typically transmitted to the cloud for processing. However, this results in increased network traffic and latencies. Edge computing has a potential to remedy these challenges by moving computation physically closer to the network edge where data are generated. However, edge computing does not have sufficient resources for complex data analytics tasks. Consequently, this paper investigates merging cloud and edge computing for IoT data analytics and presents a deep learning-based approach for data reduction on the edge with the machine learning on the cloud. The encoder part of the autoencoder is located on the edge to reduce data dimensions. Reduced data are sent to the cloud where there are used directly for machine learning or expanded to original features using the decoder part of the autoencoder. The proposed approach has been evaluated on the human activity recognition tasks. Results show that 50% data reduction did not have a significant impact on the classification accuracy and 77% reduction only caused 1% change.

2020 IEEE International Conference on Web Services (ICWS), 2020

Automated service classification plays a crucial role in service discovery, selection, and compos... more Automated service classification plays a crucial role in service discovery, selection, and composition. Machine learning has been widely used for service classification in recent years. However, the performance of conventional machine learning methods highly depends on the quality of manual feature engineering. In this paper, we present a novel deep neural network to automatically abstract low-level representation of both service name and service description to high-level merged features without feature engineering and the length limitation, and then predict service classification on 50 service categories. To demonstrate the effectiveness of our approach, we conduct a comprehensive experimental study by comparing 10 machine learning methods on 10,000 real-world web services. The result shows that the proposed deep neural network can achieve higher accuracy in classification and more robust than other machine learning methods.

IEEE Access, 2021

Rapid expansion of smart metering technologies has enabled large-scale collection of electricity ... more Rapid expansion of smart metering technologies has enabled large-scale collection of electricity consumption data and created the foundation for sensor-based load forecasting on individual buildings or even the household level. With continuously growing energy consumption, the importance of energy management including load forecasting is increasing in order to remedy the energy effect on the environment. Numerous machine learning techniques have been proposed for sensor-based load forecasting but most are offline approaches: the model is trained once and then used to infer future consumption. However, these approaches are not able to adapt to concept drift: for example, their accuracy will degrade when the building use changes or new equipment is installed. Thus, an approach capable of learning from new data as they arrive is needed. This paper proposes adaptive online ensemble learning with Recurrent Neural Network (RNN) and ARIMA for load forecasting under concept drift. The RNN part of the ensembles consists of Online Adaptive RNN as its underlying RNN learner has the ability to model temporal dependencies present in load data while its online nature enables continuous learning from arriving data. The adaptation to the concept drift is improved by adding Rolling ARIMA to the ensemble. The performance of the proposed approach has been examined on the four individual homes with different degrees of concept drift. The results show that the proposed ensemble achieves better accuracy than its constituent algorithms alone and, moreover, the analysis demonstrates the need to examine load forecasting approaches in respect to how they handle concept drift. INDEX TERMS Load forecasting, concept drift, energy forecasting, ensemble learning, recurrent neural network, online learning.

Applied Energy, 2021

Electricity load forecasting has been attracting research and industry attention because of its i... more Electricity load forecasting has been attracting research and industry attention because of its importance for energy management, infrastructure planning, and budgeting. In recent years, the proliferation of smart meters and other sensors has created new opportunities for sensor-based load forecasting on the building and even individual household level. Machine learning approaches such as Recurrent Neural Networks (RNNs) have shown great successes in load forecasting, but these approaches employ offline learning: they are trained once and miss on the opportunity to learn from newly arriving data. Moreover, they are not well suited for handling the concept drift; for example, their predictive performance will degrade if the load changes due to the installation of new equipment. Consequently, this paper proposes Online Adaptive RNN, an approach for load forecasting capable of continuously learning from newly arriving data and adapting to new patterns. RNN is employed to capture time dependencies while the online aspect is achieved by updating the RNN weights according to new data. The performance is monitored; if it degrades, online tuning is activated to adapt the RNN hyperparameters to changes in data. The proposed approach was evaluated with data from five individual homes: the results show that the proposed approach achieves higher accuracy than the standalone offline long short term memory network and five other online algorithms. Moreover, the time to learn from new samples is only a fraction of the time needed to retrain the offline model.

IECON 2023- 49th Annual Conference of the IEEE Industrial Electronics Society, 2023

In recent years, electric vehicles (EVs) have been widely adopted because of their environmental ... more In recent years, electric vehicles (EVs) have been widely adopted because of their environmental benefits. However, the increasing volume of EVs poses capacity issues for grid operators as simultaneously charging many EVs may result in grid instabilities. Scheduling EV charging for grid load balancing has a potential to prevent load peaks caused by simultaneous EV charging and contribute to balance of supply and demand. This paper proposes a user-preference-based scheduling approach to minimize costs for the user while balancing grid loads. The EV owners benefit by charging when the electricity cost is lower, but still within the user-defined preferred charging periods. On the other hand, the approach reduces the pressure on the grid by balancing the grid load. Two methods, the greedy algorithm and nonlinear programming, are considered along with users’ charging preferences and durations. For scheduling small numbers of charging activities, the nonlinear programming model achieves better load balancing than the greedy algorithm; however, for scheduling medium to large numbers of charging activities, the greedy algorithm has a clear advantage in terms of time complexity.

Neural Computing and Applications, Apr 21, 2022

International Journal of Electrical Power & Energy Systems

Energies

The transition to Electric Vehicles (EV) in place of traditional internal combustion engines is i... more The transition to Electric Vehicles (EV) in place of traditional internal combustion engines is increasing societal demand for electricity. The ability to integrate the additional demand from EV charging into forecasting electricity demand is critical for maintaining the reliability of electricity generation and distribution. Load forecasting studies typically exclude households with home EV charging, focusing on offices, schools, and public charging stations. Moreover, they provide point forecasts which do not offer information about prediction uncertainty. Consequently, this paper proposes the Long Short-Term Memory Bayesian Neural Networks (LSTM-BNNs) for household load forecasting in presence of EV charging. The approach takes advantage of the LSTM model to capture the time dependencies and uses the dropout layer with Bayesian inference to generate prediction intervals. Results show that the proposed LSTM-BNNs achieve accuracy similar to point forecasts with the advantage of pre...

2022 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

This paper introduces the Virtual Sensor Middleware (VSM), which facilitates distributed sensor d... more This paper introduces the Virtual Sensor Middleware (VSM), which facilitates distributed sensor data processing on multiple fog nodes. VSM uses a Virtual Sensor as the core component of the middleware. The virtual sensor concept is redesigned to support functionality beyond sensor/device virtualization, such as deploying a set of virtual sensors to represent an IoT application and distributed sensor data processing across multiple fog nodes. Furthermore, the virtual sensor deals with the heterogeneous nature of IoT devices and the various communication protocols using different adapters to communicate with the IoT devices and the underlying protocol. VSM uses the publishsubscribe design pattern to allow virtual sensors to receive data from other virtual sensors for seamless sensor data consumption without tight integration among virtual sensors, which reduces application development efforts. Furthermore, VSM enhances the design of virtual sensors with additional components that support sharing of data in dynamic environments where data receivers may change over time, data aggregation is required, and dealing with missing data is essential for the applications.

2022 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

Energy companies often implement various demand response (DR) programs to better match electricit... more Energy companies often implement various demand response (DR) programs to better match electricity demand and supply by offering the consumers incentives to reduce their demand during critical periods. Classifying clients according to their consumption patterns enables targeting specific groups of consumers for DR. Traditional clustering algorithms use standard distance measurement to find the distance between two points. The results produced by clustering algorithms such as Kmeans, K-medoids, and Gaussian Mixture Models depend on the clustering parameters or initial clusters. In contrast, our methodology uses a shape-based approach that combines Agglomerative Hierarchical Clustering (AHC) with Dynamic Time Warping (DTW) to classify residential households' daily load curves based on their consumption patterns. While DTW seeks the optimal alignment between two load curves, AHC provides a realistic initial clusters center. In this paper, we compare the results with other clustering algorithms such as K-means, K-medoids, and GMM using different distance measures, and we show that AHC using DTW outperformed other clustering algorithms and needed fewer clusters.

2022 IEEE Conference on Dependable and Secure Computing (DSC)

Engineering Applications of Artificial Intelligence

There is an increasing demand for using Unmanned Aerial Vehicle (UAV), known as drones, in differ... more There is an increasing demand for using Unmanned Aerial Vehicle (UAV), known as drones, in different applications such as packages delivery, traffic monitoring, search and rescue operations, and military combat engagements. In all of these applications, the UAV is used to navigate the environment autonomously-without human interaction, perform specific tasks and avoid obstacles. Autonomous UAV navigation is commonly accomplished using Reinforcement Learning (RL), where agents act as experts in a domain to navigate the environment while avoiding obstacles. Understanding the navigation environment and algorithmic limitations plays an essential role in choosing the appropriate RL algorithm to solve the navigation problem effectively. Consequently, this study first identifies the main UAV navigation tasks and discusses navigation frameworks and simulation software. Next, RL algorithms are classified and discussed based on the environment, algorithm characteristics, abilities, and applications in different UAV navigation problems, which will help the practitioners and researchers select the appropriate RL algorithms for their UAV navigation use cases. Moreover, identified gaps and opportunities will drive UAV navigation research. Symbol Description ∈ State belongs to all possible states ∈ Action belongs to the set of all possible Actions ∈ Reward belongs to the set of all generated Rewards Discounted factor decreases the contribution of the future rewards, where < < The Expected Summation of the Discounted Rewards; = ∑ ∞ = + + (′ , | ,) The probability of the transition to state ′ with reward from taking action in state at time A trajectory consists of a sequence of actions and states pairs, where the actions influence the states, also called an episode. Each trajectory has a start state and ends in a final state that terminates the trajectory (,) Action-value function expresses the expected return of the state-action pairs (,); (,) is (,) parameterized by

Sensors

The deterioration of infrastructure’s health has become more predominant on a global scale during... more The deterioration of infrastructure’s health has become more predominant on a global scale during the 21st century. Aging infrastructure as well as those structures damaged by natural disasters have prompted the research community to improve state-of-the-art methodologies for conducting Structural Health Monitoring (SHM). The necessity for efficient SHM arises from the hazards damaged infrastructure imposes, often resulting in structural collapse, leading to economic loss and human fatalities. Furthermore, day-to-day operations in these affected areas are limited until an inspection is performed to assess the level of damage experienced by the structure and the required rehabilitation determined. However, human-based inspections are often labor-intensive, inefficient, subjective, and restricted to accessible site locations, which ultimately negatively impact our ability to collect large amounts of data from inspection sites. Though Deep-Learning (DL) methods have been heavily explor...

Industry hires engineers primarily for solving workplace problems; consequently problem solving s... more Industry hires engineers primarily for solving workplace problems; consequently problem solving skills are an essential part of an engineering education. However, industry problems, as well as the environment engineers work in, are often quite different than what students experience at universities. This workshop explores problem based learning, the differences between problems students typically solve in the classroom and the workplace, as well as the strategies for making classroom problems emulate real-world workplace problems. As the main goal of engineering education is to prepare students for work in industry, closing the gap between classroom and workplace problems will result in better prepared graduates

2021 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE), 2021

The desire to make applications and machines more intelligent and the aspiration to enable their ... more The desire to make applications and machines more intelligent and the aspiration to enable their operation without human interaction have been driving innovations in neural networks, deep learning, and other machine learning techniques. Although reinforcement learning has been primarily used in video games, recent advancements and the development of diverse and powerful reinforcement algorithms have enabled the reinforcement learning community to move from playing video games to solving complex real-life problems in autonomous systems such as self-driving cars, delivery drones, and automated robotics. Understanding the environment of an application and the algorithms' limitations plays a vital role in selecting the appropriate reinforcement learning algorithm that successfully solves the problem on hand in an efficient manner. Consequently, in this study, we identify three main environment types and classify reinforcement learning algorithms according to those environment types. Moreover, within each category, we identify relationships between algorithms. The overview of each algorithm provides insight into the algorithms' foundations and reviews similarities and differences among algorithms. This study provides a perspective on the field and helps practitioners and researchers to select the appropriate algorithm for their use case.

Industry hires engineers primarily for solving workplace problems; consequently problem solving s... more Industry hires engineers primarily for solving workplace problems; consequently problem solving skills are an essential part of an engineering education. However, industry problems, as well as the environment engineers work in, are often quite different than what students experience at universities. This workshop explores problem based learning, the differences between problems students typically solve in the classroom and the workplace, as well as the strategies for making classroom problems emulate real-world workplace problems. As the main goal of engineering education is to prepare students for work in industry, closing the gap between classroom and workplace problems will result in better prepared graduates.

Montaža je podrucje koje ima znatan udio u troskovima i vremenu proizvodnje. Stoga su danasnji na... more Montaža je podrucje koje ima znatan udio u troskovima i vremenu proizvodnje. Stoga su danasnji napori usmjereni k unapređenju montaže, posebno fleksibilne montaže uz primjenu robota. Ovaj rad se bavi razvojem inteligentnog robotskog sustava za primjenu u montaži, koji je sposoban rjesavati razlicite zadace prilagođavajuci svoje ponasanje promjenljivom okolisu, ukljucujuci snalaženje u neizvjesnim i nepredvidivim situacijama. Da bi to postigao robot mora moci naci rjesenje u nepoznatoj situaciji, uciti ponasanje, sto znaci procedure ponasanja zajedno s odgovarajucim znanjem o strukturi radnog prostora, i prepoznati radni prostor. Problem inteligentnog robotskog ponasanja se suocava s nepredvidivo velikim brojem varijanti problema sto ne može biti obuhvaceno egzaktnom domenom znanja, stoga se razvoj zasniva na aktivnim metodama ucenja. Sustav primjenjuje metodu prisilnog ucenja baziranu na strategijskim i slucajnim pokusajima za pronalaženje rjesenja problema i neuronsku mrežu za pamc...

International Journal of Electrical Power & Energy Systems, 2021

Load forecasting is essential for energy management, infrastructure planning, grid operation, and... more Load forecasting is essential for energy management, infrastructure planning, grid operation, and budgeting. Large scale smart meter deployments have resulted in ability to collect massive energy data and have created opportunities in sensor-based forecasting. Machine learning (ML) has demonstrated great successes in sensor-based load forecasting; however, when prediction is needed on a smart meter level, typically a single model is trained for each smart meter. With a large number of meters, this becomes computationally expensive or even infeasible. On the other hand, with conventional ML, training a single model for several smart meters requires participants to share their data with the central server. Consequently, this paper proposes federated learning for load forecasting with smart meter data: this strategy enables training a single model with all participating smart meters without the need to share local data. Two alternative federated learning strategies are examined: FedSGD, which performs one step of gradient descent on client before merging updates on the server, and Fe-dAVG, which carries out several steps before the merging. Specifically, residential consumers are diverse what makes training a single model challenging as load profiles vary across consumers. The results show that the FedAVG achieves better accuracy than FedSGD while also requiring fewer communication rounds. Comparing to individual models for each meter and a single central models for all meters, FedAVG achieves comparable or better accuracy.

2019 IEEE Canadian Conference of Electrical and Computer Engineering (CCECE), 2019

Sensors, wearables, mobile and other Internet of Thing (IoT) devices are becoming increasingly in... more Sensors, wearables, mobile and other Internet of Thing (IoT) devices are becoming increasingly integrated in all aspects of our lives. They are capable of collecting massive quantities of data that are typically transmitted to the cloud for processing. However, this results in increased network traffic and latencies. Edge computing has a potential to remedy these challenges by moving computation physically closer to the network edge where data are generated. However, edge computing does not have sufficient resources for complex data analytics tasks. Consequently, this paper investigates merging cloud and edge computing for IoT data analytics and presents a deep learning-based approach for data reduction on the edge with the machine learning on the cloud. The encoder part of the autoencoder is located on the edge to reduce data dimensions. Reduced data are sent to the cloud where there are used directly for machine learning or expanded to original features using the decoder part of the autoencoder. The proposed approach has been evaluated on the human activity recognition tasks. Results show that 50% data reduction did not have a significant impact on the classification accuracy and 77% reduction only caused 1% change.

2020 IEEE International Conference on Web Services (ICWS), 2020

Automated service classification plays a crucial role in service discovery, selection, and compos... more Automated service classification plays a crucial role in service discovery, selection, and composition. Machine learning has been widely used for service classification in recent years. However, the performance of conventional machine learning methods highly depends on the quality of manual feature engineering. In this paper, we present a novel deep neural network to automatically abstract low-level representation of both service name and service description to high-level merged features without feature engineering and the length limitation, and then predict service classification on 50 service categories. To demonstrate the effectiveness of our approach, we conduct a comprehensive experimental study by comparing 10 machine learning methods on 10,000 real-world web services. The result shows that the proposed deep neural network can achieve higher accuracy in classification and more robust than other machine learning methods.

IEEE Access, 2021

Rapid expansion of smart metering technologies has enabled large-scale collection of electricity ... more Rapid expansion of smart metering technologies has enabled large-scale collection of electricity consumption data and created the foundation for sensor-based load forecasting on individual buildings or even the household level. With continuously growing energy consumption, the importance of energy management including load forecasting is increasing in order to remedy the energy effect on the environment. Numerous machine learning techniques have been proposed for sensor-based load forecasting but most are offline approaches: the model is trained once and then used to infer future consumption. However, these approaches are not able to adapt to concept drift: for example, their accuracy will degrade when the building use changes or new equipment is installed. Thus, an approach capable of learning from new data as they arrive is needed. This paper proposes adaptive online ensemble learning with Recurrent Neural Network (RNN) and ARIMA for load forecasting under concept drift. The RNN part of the ensembles consists of Online Adaptive RNN as its underlying RNN learner has the ability to model temporal dependencies present in load data while its online nature enables continuous learning from arriving data. The adaptation to the concept drift is improved by adding Rolling ARIMA to the ensemble. The performance of the proposed approach has been examined on the four individual homes with different degrees of concept drift. The results show that the proposed ensemble achieves better accuracy than its constituent algorithms alone and, moreover, the analysis demonstrates the need to examine load forecasting approaches in respect to how they handle concept drift. INDEX TERMS Load forecasting, concept drift, energy forecasting, ensemble learning, recurrent neural network, online learning.

Applied Energy, 2021

Electricity load forecasting has been attracting research and industry attention because of its i... more Electricity load forecasting has been attracting research and industry attention because of its importance for energy management, infrastructure planning, and budgeting. In recent years, the proliferation of smart meters and other sensors has created new opportunities for sensor-based load forecasting on the building and even individual household level. Machine learning approaches such as Recurrent Neural Networks (RNNs) have shown great successes in load forecasting, but these approaches employ offline learning: they are trained once and miss on the opportunity to learn from newly arriving data. Moreover, they are not well suited for handling the concept drift; for example, their predictive performance will degrade if the load changes due to the installation of new equipment. Consequently, this paper proposes Online Adaptive RNN, an approach for load forecasting capable of continuously learning from newly arriving data and adapting to new patterns. RNN is employed to capture time dependencies while the online aspect is achieved by updating the RNN weights according to new data. The performance is monitored; if it degrades, online tuning is activated to adapt the RNN hyperparameters to changes in data. The proposed approach was evaluated with data from five individual homes: the results show that the proposed approach achieves higher accuracy than the standalone offline long short term memory network and five other online algorithms. Moreover, the time to learn from new samples is only a fraction of the time needed to retrain the offline model.

Load forecasting is essential for the operation and planning of a utility company. Recent large-s... more Load forecasting is essential for the operation and planning of a utility company. Recent large-scale smart meter deployments enabled the collection of fine-grained load data and created opportunities for sensor-based load forecasting. Machine learning (ML) has achieved great successes in load forecasting; however, conventional ML techniques require data transfer to the cloud or another centralized location for model training. This not only exposes data to privacy and security risks but also, with a large number of smart meters, increases network traffic. Federated Learning (FL) has a potential to alleviate mentioned concerns by training a single ML model in a distributed manner without requiring participants to share their data. Consequently, this paper proposes FedNorm, a novel asynchronous FL approach for load forecasting with smart meter data. While most FL strategies are synchronous and require all clients to complete local training in each round of aggregation, FedNorm is asynchronous and aggregates updates without waiting for lagging clients. To achieve this, FedNorm measures the clients contributions considering similarities of local and global models as well as the loss function magnitudes. The experiments demonstrate that FedNorm achieves higher accuracy than seven state-of- the-art FL techniques. Moreover, experiments show that FedNorm converges in fewer communication rounds compared to other FL models.