Kasun Amarasinghe

Building Energy Management Systems: The Age of Intelligent and Adaptive Buildings

Building automation systems (BAS), or building control systems (BCS), typically consist of building energy management systems (BEMSs), physical security and access control, fire/life safety, and other systems (elevators, public announcements, and closed-circuit television). BEMSs control heating, ventilation, and air conditioning (HVAC) and lighting systems in buildings; more specifically, they control HVACs primary components such as air handling units (AHUs), chillers, and heating elements. BEMSs are essential components of modern buildings, tasked with seemingly contradicting requirements?minimizing energy consumption while maintaining occupants? comfort [1]. In the United States, about 40% of total energy consumption and 70% of electricity consumption are spent on buildings every year. These numbers are comparable to global statistics that about 30% of total energy consumption and 60% of electricity consumption are spent on buildings. Buildings are an integral part of global cyber-physical systems (smart cities) and evolve and interact with their surroundings. As buildings undergo years of exploitation, their thermal characteristics deteriorate, indoor spaces (especially in commercial buildings) get rearranged, and usage patterns change. In time, their inner (and outer) microclimates adjust to changes in surrounding buildings, overshadowing patterns, and city climates, not to mention building retrofitting. Thus, even in cases of ideally designed BEMS/HVAC systems, because of ever-changing and uncertain indoor and outdoor environments, their performance frequently falls short of expectations. Unfortunately, the complexity of BEMSs, large amounts of constantly changing data, and evolving interrelations among sensor feeds make identifying these suboptimal behaviors difficult. Therefore, traditional data-mining algorithms and data-analysis tools are often inadequate.This article provides an overview of issues related to modern BEMSs with a multitude of (often conflicting) requirements. Because of massive and often incomplete data sets, control, sensing, and the evolving nature of these complex systems, computational intelligence (CI) techniques present a natural solution to optimal energy efficiency, energy security, and occupant comfort in buildings. The article further presents an overall architecture where CI can be used in BEMSs and concludes with a case study of the practical applications of using CI techniques in the BEMS domain.

Building energy load forecasting using Deep Neural Networks

Ensuring sustainability demands more efficient energy management with minimized energy wastage. Therefore, the power grid of the future should provide an unprecedented level of flexibility in energy management. To that end, intelligent decision making requires accurate predictions of future energy demand/load, both at aggregate and individual site level. Thus, energy load forecasting have received increased attention in the recent past. However, it has proven to be a difficult problem. This paper presents a novel energy load forecasting methodology based on Deep Neural Networks, specifically, Long Short Term Memory (LSTM) algorithms. The presented work investigates two LSTM based architectures: 1) standard LSTM and 2) LSTM-based Sequence to Sequence (S2S) architecture. Both methods were implemented on a benchmark data set of electricity consumption data from one residential customer. Both architectures were trained and tested on one hour and one-minute time-step resolution datasets. Experimental results showed that the standard LSTM failed at one-minute resolution data while performing well in one-hour resolution data. It was shown that S2S architecture performed well on both datasets. Further, it was shown that the presented methods produced comparable results with the other deep learning methods for energy forecasting in literature.

Intelligent Buildings of the Future: Cyberaware, Deep Learning Powered, and Human Interacting

Intelligent buildings are quickly becoming cohesive and integral inhabitants of cyberphysical ecosystems. Modern buildings adapt to internal and external elements and thrive on ever-increasing data sources, such as ubiquitous smart devices and sensors, while mimicking various approaches previously known in software, hardware, and bioinspired systems. This article provides an overview of intelligent buildings of the future from a range of perspectives. It discusses everything from the prospects of U.S. and world energy consumption to insights into the future of intelligent buildings based on the latest technological advancements in U.S. industry and government.

Next Generation Emergency Communication Systems via Software Defined Networks

The existing Emergency Communication System (ECS) infrastructure is becoming increasingly outdated with many members of the pubic moving away from landline based telecommunications and broadcast television in favor of cellular telephones and internet-based streaming entertainment services. Current systems for public services such as E911 and Emergency Alert System broadcasts are no longer a reliable means for reaching the public. In addition, both wired and wireless telecommunications systems can become overwhelmed, as was the case following Hurricane Katrina in 2005 and the World Trade Center disaster in 2001, and in fact, when communications are needed most urgently, the difficulty of maintaining effective communication increases exponentially. While the use of Internet based alternatives could resolve some of these problems, existing Internet infrastructure offers no dedicated or priority bandwidth to the user for emergency communications (e.g. E911 or Emergency Alert System). The current Internet capacity can also be overloaded due to high volume network data streams. Under these conditions, emergency communications (e.g. inbound and outbound communications reporting catastrophic or emergency events) may have their packets dropped resulting in incomplete and/or delayed communications. To alleviate these problems, this paper presents a novel framework for ECS using network virtualization via Software Defined Networks (SDN). A table top demonstration of ECS using SDN was developed at the University of Idaho, Idaho Falls. This paper details the foundational technologies and overviews the steps taken at the University of Idaho to develop ECS suing SDN.

Neural Network based downscaling of Building Energy Management System data

Building Energy Management Systems (BEMSs) are responsible for maintaining indoor environment by controlling Heating Ventilation and Air Conditioning (HVAC) and lighting systems in buildings. Buildings worldwide account for a significant portion of world energy consumption. Thus, increasing building energy efficiency through BEMSs can result in substantial financial savings. In addition, BEMSs can significantly impact the productivity of occupants by maintaining a comfortable environment. To increase efficiency and maintain comfort, modern BEMSs rely on a large array of sensors inside the building that provide detailed data about the building state. However, due to various reasons, buildings frequently lack sufficient number of sensors, resulting in a suboptimal state awareness. In such cases, a cost effective method for increasing state awareness is needed. Therefore, this paper presents a novel method for increasing state awareness through increasing spatial resolution of data by means of data downscaling. The presented method estimates the state of occupant zones using state data gathered at floor level using Artificial Neural Networks (ANN). The presented method was tested on a real-world CO2 dataset, and compared to a time based estimation of CO2 concentration. The downscaling method was shown to be capable of consistently producing accurate estimates while being more accurate than time based estimations.

EEG based brain activity monitoring using Artificial Neural Networks

Brain Computer Interfaces (BCI) have gained significant interest over the last decade as viable means of human machine interaction. Although many methods exist to measure brain activity in theory, Electroencephalography (EEG) is the most used method due to the cost efficiency and ease of use. However, thought pattern based control using EEG signals is difficult due two main reasons; 1) EEG signals are highly noisy and contain many outliers, 2) EEG signals are high dimensional. Therefore the contribution of this paper is a novel methodology for recognizing thought patterns based on Self Organizing Maps (SOM). The presented thought recognition methodology is a three step process which utilizes SOM for unsupervised clustering of pre-processed EEG data and feed-forward Artificial Neural Networks (ANN) for classification. The presented method was tested on 5 different users for identifying two thought patterns; “move forward” and “rest”. EEG Data acquisition was carried out using the Emotiv EPOC headset which is a low cost, commercial-off-the-shelf, noninvasive EEG signal measurement device. The presented method was compared with classification of EEG data using ANN alone. The experimental results for the 5 users chosen showed an improvement of 8% over ANN based classification.

Optimal stop word selection for text mining in critical infrastructure domain

Eliminating all stop words from the feature space is a standard practice of preprocessing in text mining, regardless of the domain which it is applied to. However, this may result in loss of important information, which adversely affects the accuracy of the text mining algorithm. Therefore, this paper proposes a novel methodology for selecting the optimal set of domain specific stop words for improved text mining accuracy. First, the presented methodology retains all the stop words in the text preprocessing phase. Then, an evolutionary technique is used to extract the optimal set of stop words that result in the best classification accuracy. The presented methodology was implemented on a corpus of open source news articles related to critical infrastructure hazards. The first step of mining geo-dependencies among critical infrastructures from text is text classification. In order to achieve this, article content was classified into two classes: 1) text content with geo-location information, and 2) text content without geo-location information. Classification accuracy presented methodology was compared to accuracies of four other test cases. Experimental results with 10-fold cross validation showed that the presented method yielded an increase of 1.76% or higher in True Positive (TP) rate and a 2.27% or higher increase in the True Negative (TN) rate compared to the other techniques.

Artificial neural networks based thermal energy storage control for buildings

Heating, Ventilation and Air Conditioning (HVAC) system is largest energy consumer in buildings. Worldwide, buildings consume 20% of the total energy production. Therefore, increasing efficiency of the HVAC system will result in significant financial savings. As one solution, Thermal Energy Storage (TES) tanks are being utilized with buildings to store excess energy to be reused later. An optimal control strategy is crucial for optimal usage. Therefore, this paper presents a novel control framework based on Artificial Neural Networks (ANN) for optimally controlling a TES for achieving increased savings. The presented ANN controller utilizes 3 main inputs: 1) current TES energy availability, 2) predicted building power requirement, and 3) predicted utility load/price. In addition to the design details of the control framework, this paper presents implementation details of the ANN controller. Further, experiments on several test cases were carried out and the paper presents the experimental setup and obtained results for each test case. Performance of the presented ANN control framework was compared against a classical proportional derivative (PD) controller. It was observed that the presented framework resulted in better cost savings than the classical controller consistently for all the experimental test cases.

Deep neural networks for energy load forecasting

Smartgrids of the future promise unprecedented flexibility in energy management. Therefore, accurate predictions/forecasts of energy demands (loads) at individual site and aggregate level of the grid is crucial. Despite extensive research, load forecasting remains to be a difficult problem. This paper presents a load forecasting methodology based on deep learning. Specifically, the work presented in this paper investigates the effectiveness of using Convolutional Neural Networks (CNN) for performing energy load forecasting at individual building level. The presented methodology uses convolutions on historical loads. The output from the convolutional operation is fed to fully connected layers together with other pertinent information. The presented methodology was implemented on a benchmark data set of electricity consumption for a single residential customer. Results obtained from the CNN were compared against results obtained by Long Short Term Memories LSTM sequence-to-sequence (LSTM S2S), Factored Restricted Boltzmann Machines (FCRBM), “shallow” Artificial Neural Networks (ANN) and Support Vector Machines (SVM) for the same dataset. Experimental results showed that the CNN outperformed SVR while producing comparable results to the ANN and deep learning methodologies. Further testing is required to compare the performances of different deep learning architectures in load forecasting.

Data driven decision support for reliable biomass feedstock preprocessing

Biomass feedstock preprocessing through comminution is an essential first step in biofuel production. Chemical, physical and mechanical variability in feedstock prevents the preprocessing plants from assuming constant control parameters. Constant control parameters can lead to suboptimal capability and reliability. However, adapting the control parameters to account for the variabilities is not a trivial task. This paper presents a framework for adapting control parameters through data driven methodologies. The framework named PDU- RS is a decision support system for human in the loop control. PDU-RS is implemented on the Biofuels National User Facility Preprocessing Process Demonstration Unit (PDU), operated by the Idaho National Laboratory (INL) in Idaho Falls, Idaho. PDU-RS aims at ensuring reliability in the overall operations of the PDU while maximizing throughput. Presented implementation of the PDU-RS uses Gaussian Processes (GP) for knowledge extraction from data. This paper elaborates on the PDU-RS and presents the experimental results of implementing the PDU-RS on the real Biomass PDU. The experimental results demonstrated that the PDU-RS is able to produce significantly higher throughputs while ensuring higher reliability when compared to the traditional control methodology used with the system.