Program Topic: Cyber-Security and Big Data
- T1-3 – Thursday 08:20-09:40
- 08:20 – Comparison of Artificial Intelligence Techniques for Energy Consumption Estimation
- 08:40 – Parameter Selection of Multi-Class SVM with Evolutionary Optimization Methods for Static Security Evaluation in Power Systems
- 09:00 – Ensemble Regression Model-Based Anomaly Detection for Cyber-Physical Intrusion Detection in Smart Grids
- 09:20 – A Study of Resource-Constrained Cyber Security Planning for Smart Grid Networks
- T2-3 – Thursday 11:00-12:20
- 11:00 – Anomaly Detection in a Smart Grid Using Wavelet Transform, Variance Fractal Dimension and an Artificial Neural Network
- 11:20 – Incremental Mining of Frequent Power Consumption Patterns From Smart Meters Big Data
- 11:40 – Supervised Household’s Loads Pattern Recognition
- 12:00 – Dynamic Threshold Algorithm with Simplified Appliance Identification for Smart Meter Privacy
T1-3 – Thursday 08:20-09:40
08:20 Comparison of Artificial Intelligence Techniques for Energy Consumption Estimation
In this article, a comparison study of three artificial intelligence (AI) techniques for energy consumption estimation are presented. The models considered are: multilayer perceptron (MLP); radial basis function (RBF) and support vector machine (SVM). The energy consumption is modeled as a function of activity, structural and intensity changes. The models are applied to Canadian industrial manufacturing data from 1990 to 2000. Comparisons were based on Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Relative Absolute Error (RAE), Root Relative Square Error (RRSE) as well as Simulation Time. The best results were obtained for the Multilayer Perceptron.
08:40 Parameter Selection of Multi-Class SVM with Evolutionary Optimization Methods for Static Security Evaluation in Power Systems
Static Security Evaluation (SSE) is one of the most important real-time studies in power systems. Static security can be assessed by different machine learning methods. In this paper, one of the well-known classifiers, namely the Support Vector Machine (SVM), is used for solving the SSE problem. Proper operation of the SVM heavily depends on the appropriate choice of parameters. The optimization problem aims at determining these parameters. This work presents a study of several heuristic optimization methods for static security. In particular, Modified Particle Swarm Optimization (MPSO), Differential Evolution (DE), Ant Colony Optimization for continuous domain (ACOr), and Harmony Search (HS) are employed for determining the optimal SVM parameters. In addition to studying the performance of various optimization techniques, this work is concerned about viewing the SSE problem as a 2-class, a 3-class, and a 4-class classification problem, where each class is associated with a particular level of security. The performance of each method is presented in terms of classification accuracy and execution speed. It is shown that most optimization methods exhibit a similar performance. However, choosing the best optimization method seems to be dependent on the number of classes, and thus, on the number of security levels required by SSE. The New England 39-bus benchmark system is used for simulation.
09:00 Ensemble Regression Model-Based Anomaly Detection for Cyber-Physical Intrusion Detection in Smart Grids
The shift from centralised large production to distributed energy production has several consequences on the current power system operation. The increasing number of the distributed energy resources (DERs) replacing large power plants influences the dependency of the power system on the small scale, distributed production. Many of these DERs can be accessed and controlled remotely, posing a cybersecurity risk. This paper investigates an intrusion detection system that evaluates the DER operation in order to discover unauthorized control actions. The proposed anomaly detection method is based on an ensemble of non-linear artificial neural network DER models that detects and evaluates anomalies in the DER operation. The proposed method is validated against real measurement data. Based on the obtained results the proposed ensemble anomaly detection method with normal model training data selection achieves 0.947 precision and 0.976 accuracy, which improves the precision and accuracy of a classic model-based anomaly detection by 75.7% and 9.2% respectively.
09:20 A Study of Resource-Constrained Cyber Security Planning for Smart Grid Networks
This paper studies cyber security planning issues in resource-constrained smart grid networks. In particular, it proposes a centrality-based trust system placement scheme for energy SCADA systems. It aims to utilize centrality measurements to improve cyber protection in resource-constrained scenarios. The role of centrality measurements is to rank nodes based on their importance in a network. Trust systems are specialized security devices that are capable of firewalling and network intrusion detection. They monitor both types of traffic, ingress and egress. They are mainly deployed to provide cyber protection to supervisory control and data acquisition (SCADA) systems. Due to budgetary constraints, only a selected number of nodes are equipped with trust systems. Those nodes are known as the trust nodes. The proposed scheme uses linear programming problem (LPP) formulations to select the trust nodes. Numerical results are obtained through case studies for the IEEE BUS 30 and BUS 57 test system topologies. The results reveal that the proposed scheme is capable of improving quality of cyber protection in resource-constrained scenarios.
T2-3 – Thursday 11:00-12:20
11:00 Anomaly Detection in a Smart Grid Using Wavelet Transform, Variance Fractal Dimension and an Artificial Neural Network
This paper presents a scheme for detecting anomalous power consumption patterns attack using wavelet transform, variance fractal dimension and artificial neural network for smart grid. The main procedure of proposed algorithm consists of the following steps: I) Finding normal and anomalous patterns of power consumption to train the proposed method. II) Applying Wavelet Transform to power consumption patterns to extract features. III) Applying variance fractal dimension to the extracted features from step 3 as an input. IV) Training artificial neural network with extracted features from step3. IV) Launching the trained artificial neural network from step 4 to detect the anomalous power consumption attack based on a threshold. In the simulations, the proposed method can detect anomalous power consumption attack with 51% accuracy in the worst case scenario.
11:20 Incremental Mining of Frequent Power Consumption Patterns From Smart Meters Big Data
The key elements for understanding power consumption of a typical home are related to the activities that users are performing, the time at which appliances are used, and the interdependencies with other appliances that may be used concurrently. This information can be extracted from context rich smart meters big data. However, the main challenge is how to mine complex interdependencies among different appliances usage within a home where multiple concurrent data streams are occurring. Furthermore, generation of energy consumption data from a smart meter is an ongoing continuous process and over period of time inter-appliance associations can change or new ones can establish. In this paper, we propose a near real-time incremental mining of frequent power consumption patterns from smart meters big data. Our model exploits the benefits of pattern growth strategy and mine in quantum of 24 hour period, i.e. frequent patterns are extracted from data comprising of appliance usage tuples for 24 hours period, in a progressive manner. The details and the results of evaluating the proposed mechanism using real smart meters dataset are presented in this paper.
11:40 Supervised Household’s Loads Pattern Recognition
The deployment of smart meters is a promising innovation that comes to enhance the energy efficiency measures in the smart grid. The smart meter enables distributors to better understand the electrical network and reduce complexity of the management operations. It offers to households monitoring and control possibilities to their everyday energy consumption through the distribution of detailed information on household consumption and its evolution. This involves disaggregation of individual household loads in term of their individual energy consumption known as Non intrusive loads monitoring. In this paper, we present a supervised NILM approach based on dynamic fuzzy c-means events clustering and KNN label matching. First, a filtering method is involved to enhance the edge/events detection step. Then we perform a dynamic Fuzzy c-means clustering procedures to build appliances signature data based on active and reactive power measurements taking into account the time of day usage. The data base is further refined to map potential clusters centers that best identify the different appliances. A performance evaluation of the proposed approach is conducted showing a recognition rate over 90% for high consumption loads and promising results for low consumption loads.
12:00 Dynamic Threshold Algorithm with Simplified Appliance Identification for Smart Meter Privacy
Smart meter is one of the key components of smart power grids. The recent interest of smart meters adoption has created two research directions that are in conflict of each other in terms of data collection and user privacy, i.e. appliance identification from the measurements of smart meters and the privacy compensation for such measurements. This paper proposes a dynamic threshold algorithm (DTA) based on the prediction results from a simplified two-step filtering appliance identification algorithm (AIA) to compensate the smart meter data. The battery capacity required by the proposed DTA to achieve the same privacy level is less than 25% of that required by fixed threshold algorithms. Furthermore, the integration of AIA and DTA provides a trade-off method to solve the conflict through advancing the simplified two-step filtering AIA before DTA.