Cyber-physical systems (CPS) are ubiquitous in several critical infrastructure applications. Forecasting the state of CPS, is essential for better planning, resource allocation and minimizing operational costs. It is imperative to forecast the state of a CPS multiple steps into the future to afford enough time for planning of CPS operation to minimize costs and component wear. Forecasting system state also serves as a precursor to detecting process anomalies and faults. Concomitantly, sensors used for data collection are commodity hardware and experience frequent failures resulting in periods with sparse or no data. In such cases, re-construction through imputation of the missing data sequences is imperative to alleviate data sparsity and enable better performance of down-stream analytic models. In this paper, we tackle the problem of CPS state forecasting and data imputation and characterize the performance of a wide array of deep learning architectures - unidirectional gated and non-gated recurrent architectures, sequence to sequence (Seq2Seq) architectures as well as bidirectional architectures - with a specific focus towards applications in CPS. We also study the impact of procedures like scheduled sampling and attention, on model training. Our results indicate that Seq2Seq models are superior to traditional step ahead forecasting models and yield an improvement of at least 28.5% for gated recurrent architectures and about 87.6% for non-gated architectures in terms of forecasting performance. We also notice that bidirectional models learn good representations for forecasting as well as for data imputation. Bidirectional Seq2Seq models show an average improvement of 17.6% in forecasting performance over their unidirectional counterparts. We also demonstrate the effect of employing an attention mechanism in the context of Seq2Seq architectures and find that it provides an average improvement of 57.12% in the case of unidirectional Seq2Seq architectures while causing a performance decline in the case of bidirectional Seq2Seq architectures. Finally, we also find that scheduled sampling helps in training better models that yield significantly lower forecasting error.

Increasing adoption of solar photovoltaic (PV) presents new challenges to modern power grid due to its variable and intermittent nature. Fluctuating outputs from PV generation can cause the grid violating voltage operation limits. PV smart inverters (SIs) provide a fast-response method to regulate voltage by modulating real and/or reactive power at the connection point. Yet existing local autonomous control scheme of SIs is based on local information without coordination, which can lead to suboptimal performance. In this paper, a deep reinforcement learning (DRL) based algorithm is developed and implemented for coordinating multiple SIs. The reward scheme of the DRL is carefully designed to ensure voltage operation limits of the grid are met with more effective utilization of SI reactive power. The proposed DRL agent for voltage control can learn its policy through interaction with massive offline simulations, and adapts to load and solar variations. The performance of the DRL agent is compared against the local autonomous control on the IEEE 37 node system with thousands of scenarios. The results show a properly trained DRL agent can intelligently coordinate different SIs for maintaining grid voltage within allowable ranges, achieving reduction of PV production curtailment, and decreasing system losses.

Modern cyber-physical systems are increasingly complex and vulnerable to attacks like false data injection aimed at destabilizing and confusing the systems. We develop and evaluate an attack-detection framework aimed at learning a dynamic invariant network, data-driven temporal causal relationships between components of cyber-physical systems. We evaluate the relative performance in attack detection of the proposed model relative to traditional anomaly detection approaches. In this paper, we introduce Granger Causality based Kalman Filter with Adaptive Robust Thresholding (G-KART) as a framework for anomaly detection based on data-driven functional relationships between components in cyber-physical systems. In particular, we select power systems as a critical infrastructure with complex cyber-physical systems whose protection is an essential facet of national security. The system presented is capable of learning with or without network topology the task of detection of false data injection attacks in power systems. Kalman filters are used to learn and update the dynamic state of each component in the power system and in-turn monitor the component for malicious activity. The ego network for each node in the invariant graph is treated as an ensemble model of Kalman filters, each of which captures a subset of the node's interactions with other parts of the network. We finally also introduce an alerting mechanism to surface alerts about compromised nodes.

Machine learning tasks typically involve minimizing a loss function that measures the distance of the model output and the ground-truth. In some applications, in addition to the usual loss function, the output must also satisfy certain requirements for further processing. We call such requirements model conditioning. We investigate cases where the conditioner is not differentiable or cannot be expressed in closed form and, hence, cannot be directly included in the loss function of the machine learning model. We propose to replace the conditioner with a learned dummy model which is applied on the output of the main model. The entire model, composed of the main and dummy models, is trained end-to-end. Throughout training, the dummy model learns to approximate the conditioner and, thus, forces the main model to generate outputs that satisfy the specified requirements. We demonstrate our approach on a use-case of demand charge-aware electricity load forecasting. We show that jointly minimizing the error in forecast load and its demand charge threshold results in significant improvement to existing load forecast methods.

Accurate modeling of battery capacity degradation is an important component for both battery manufacturers and energy management systems. In this paper, we develop a battery degradation model using deep learning algorithms. The model is trained with the real data collected from battery storage solutions installed and operated for behind-the-meter customers. In the dataset, battery operation data are recorded at a small scale (five minutes) and battery capacity is measured at every six months. In order to improve the training performance, we apply two preprocessing techniques, namely subsampling and feature extraction on operation data, and also interpolating between capacity measurements at times for which battery operation features are available. We integrate both cyclic and calendar aging processes in a unified framework by extracting the corresponding features from operation data. The proposed model uses LSTM units followed by a fully-connected network to process weekly battery operation features and predicts the capacity degradation. The experimental results show that our method can accurately predict the capacity fading and significantly outperforms baseline models including persistence and autoregressive (AR) models.

Cyber-physical systems (CPSs) are today ubiquitous in urban environments. Such systems now serve as the backbone to numerous critical infrastructure applications, from smart grids to IoT installations. Scalable and seamless operation of such CPSs requires sophisticated tools for monitoring the time series progression of the system, dynamically tracking relationships, and issuing alerts about anomalies to operators. We present an online monitoring system ( illiad ) that models the state of the CPS as a function of its relationships between constituent components, using a combination of model-based and data-driven strategies. In addition to accurate inference for state estimation and anomaly tracking, illiad also exploits the underlying network structure of the CPS (wired or wireless) for state estimation purposes. We demonstrate the application of illiad to two diverse settings: a wireless sensor motes application and an IEEE 33-bus microgrid.

This paper presents the development of an intelligent dynamic energy management system (I-DEMS) for a smart microgrid. An evolutionary adaptive dynamic programming and reinforcement learning framework is introduced for evolving the I-DEMS online. The I-DEMS is an optimal or near-optimal DEMS capable of performing grid-connected and islanded microgrid operations. The primary sources of energy are sustainable, green, and environmentally friendly renewable energy systems (RESs), e.g., wind and solar; however, these forms of energy are uncertain and nondispatchable. Backup battery energy storage and thermal generation were used to overcome these challenges. Using the I-DEMS to schedule dispatches allowed the RESs and energy storage devices to be utilized to their maximum in order to supply the critical load at all times. Based on the microgrid's system states, the I-DEMS generates energy dispatch control signals, while a forward-looking network evaluates the dispatched control signals over time. Typical results are presented for varying generation and load profiles, and the performance of I-DEMS is compared with that of a decision tree approach-based DEMS (D-DEMS). The robust performance of the I-DEMS was illustrated by examining microgrid operations under different battery energy storage conditions.

The analysis of large scale data logged from complex cyber-physical systems, such as microgrids, often entails the discovery of invariants capturing functional as well as operational relationships underlying such large systems. We describe a latent factor approach to infer invariants underlying system variables and how we can leverage these relationships to monitor a cyber-physical system. In particular we illustrate how this approach helps rapidly identify outliers during system operation.

The integration of hybrid energy storage system (ESS) with renewable energy has been considered an effective solution to counteract its variant and unpredictable power generation. This paper proposes an integrated fuzzy-logic-based power management system for the hybrid Ultracapacitor-Battery ESS. The integrated fuzzy-logic power control scheme is designed to coordinate the power sharing among these two types of energy storage elements without the aid of filter-based power sharing as baseline. The proposed method further applies the multi-rate operation mechanism and considers system dynamics, e.g. varying system status, the components' energy status, etc. Performance of the proposed control method is firstly investigated using simulations in MATLAB-Simulink. Meanwhile a small-scale hardware testbed is set-up and the proposed power management scheme is implemented on-line to validate its effectiveness.

Ushered by recent developments in various areas of science and technology, modern energy systems are going to be an inevitable part of our societies. Smart grids are one of these modern systems that have attracted many research activities in recent years. Before utilizing the next generation of smart grids, we should have a comprehensive understanding of the interdependent energy networks and processes. Next-generation energy systems networks cannot be effectively designed, analyzed, and controlled in isolation from the social, economic, sensing, and control contexts in which they operate. In this article, we present a novel framework to support charging and storage infrastructure design for electric vehicles. We develop coordinated clustering techniques to work with network models of urban environments to aid in placement of charging stations for an electrical vehicle deployment scenario. Furthermore, we evaluate the network before and after the deployment of charging stations, to recommend the installation of appropriate storage units to overcome the extra load imposed on the network by the charging stations. We demonstrate the multiple factors that can be simultaneously leveraged in our framework to achieve practical urban deployment. Our ultimate goal is to help realize sustainable energy system management in urban electrical infrastructure by modeling and analyzing networks of interactions between electric systems and urban populations.

The large scale penetration of electric vehicles (EVs) will introduce technical challenges to the distribution grid, but also carries the potential for vehicle-to-grid services. Namely, if available in large enough numbers, EVs can be used as a distributed energy resource (DER) and their presence can influence optimal DER investment and scheduling decisions in microgrids. In this work, a novel EV fleet aggregator model is introduced in a stochastic formulation of DER-CAM [1], an optimization tool used to address DER investment and scheduling problems. This is used to assess the impact of EV interconnections on optimal DER solutions considering uncertainty in EV driving schedules. Optimization results indicate that EVs can have a significant impact on DER investments, particularly if considering short payback periods. Furthermore, results suggest that uncertainty in driving schedules carries little significance to total energy costs, which is corroborated by results obtained with the stochastic formulation of the problem.

Abstract Time series prediction methods are used on a daily basis by analysts for making important decisions. Most of these methods use some variant of moving averages to reduce the number of data points before prediction. However, to reach a good prediction in certain applications (e.g., power consumption time series in data centers) it is important to preserve peaks and their patterns. In this paper, we introduce automated peak‐preserving smoothing and prediction algorithms, enabling a reliable long term prediction for seasonal data, and combine them with an advanced visual interface: (1) using high resolution cell‐based time series to explore seasonal patterns, (2) adding new visual interaction techniques (multi‐scaling, slider, and brushing & linking) to incorporate human expert knowledge, and (3) providing both new visual accuracy color indicators for validating the predicted results and certainty bands communicating the uncertainty of the prediction. We have integrated these techniques into a well‐fitted solution to support the prediction process, and applied and evaluated the approach to predict both power consumption and server utilization in data centers with 70–80% accuracy.

Practically every large IT organization hosts data centers---a mix of computing elements, storage systems, networking, power, and cooling infrastructure---operated either in-house or outsourced to major vendors. A significant element of modern data centers is their cooling infrastructure, whose efficient and sustainable operation is a key ingredient to the “always-on” capability of data centers. We describe the design and implementation of CAMAS (Chiller Advisory and MAnagement System), a temporal data mining solution to mine and manage chiller installations. CAMAS embodies a set of algorithms for processing multivariate time-series data and characterizes sustainability measures of the patterns mined. We demonstrate three key ingredients of CAMAS---motif mining, association analysis, and dynamic Bayesian network inference---that help bridge the gap between low-level, raw, sensor streams, and the high-level operating regions and features needed for an operator to efficiently manage the data center. The effectiveness of CAMAS is demonstrated by its application to a real-life production data center managed by HP.

This paper describes an approach for designing a power management plan that matches the supply of power with the demand for power in data centers. Power may come from the grid, from local renewable sources, and possibly from energy storage subsystems. The supply of renewable power is often time-varying in a manner that depends on the source that provides the power, the location of power generators, and the weather conditions. The demand for power is mainly determined by the time-varying workloads hosted in the data center and the power management policies implemented by the data center. A case study demonstrates how our approach can be used to design a plan for realistic and complex data center workloads. The study considers a data center's deployment in two geographic locations with different supplies of power. Our approach offers greater precision than other planning methods that do not take into account time-varying power supply and demand and data center power management policies.

The growth of engineering disciplines, such as maintainability and reliability, have been motivated by several factors, which include the ever growing of system complexity as well as the high cost associated to failures repairing activities and productivity reduction. This work adopts a methodology, which includes a hybrid modeling technique that considers the advantages of both stochastic Petri nets (SPN) and reliability block diagrams (RBD) to evaluate maintenance policies with different service level agreement (SLA). Dependability, cost and sustainability are also prominent features considered in the proposed methodology.

The advent of cloud computing has demanded more computational resources from data centers in order to provide high-availability services required in this new paradigm. To support data center dependability evaluation, this paper presents a tool, namely, ASTRO, which adopts a hybrid modeling approach, that includes Reliability Block Diagrams (RBD), Stochastic Petri Nets (SPN) and Data Center High-Level models. In addition, this tool also provides a prominent functionality in which an evaluation result can be reused in other models. ASTRO Tool also permits a hierarchical modeling, supporting the piecewise approach proposed by the methodology in which the tool is inserted.

Abstract Today's large data centres are the computational hubs of the next generation of IT services. With the advent of dynamic smart cooling and rack level sensing, the need for visual data exploration is growing. If administrators know the rack level thermal state changes and catch problems in real time, energy consumption can be greatly reduced. In this paper, we apply a cell‐based spatio‐temporal overall view with high‐resolution time series to simultaneously analyze complex thermal state changes over time across hundreds of racks. We employ cell‐based visualization techniques for trouble shooting and abnormal state detection. These techniques are based on the detection of sensor temperature relations and events to help identify the root causes of problems. In order to optimize the data centre cooling system performance, we derive new non‐overlapped scatter plots to visualize the correlations between the temperatures and chiller utilization. All these techniques have been used successfully to monitor various time‐critical thermal states in real‐world large‐scale production data centres and to derive cooling policies. We are starting to embed these visualization techniques into a handheld device to add mobile monitoring capability.

Motivation: Data centers are a critical component of modern IT infrastructure but are also among the worst environmental offenders through their increasing energy usage and the resulting large carbon footprints. Efficient management of data centers, including power management, networking, and cooling infrastructure, is hence crucial to sustainability. In the absence of a 'first-principles' approach to manage these complex components and their interactions, data-driven approaches have become attractive and tenable.