Karthikeyan Balasubramaniam

Cellular neural network based situational awareness system for power grids

Situational awareness (SA) in simple terms is to understand the current state of the system and based on that understanding predict how system states are to evolve over time. Predictive modeling of power systems using conventional methods is time consuming and hence not well suited for real-time operation. In this study, neural network (NN) based non-linear predictor is used to predict states of power system for future time instance. Required control signals are computed based on predicted state variables and control set points. In order to reduce computation the problem is decoupled and solved in a cellular array of NNs. The cellular neural network (CNN) framework allows for accurate prediction with only minimal information exchange between neighboring predictors. The predicted states are then used in computing stability metrics that give proximity to point of instability. The situational awareness platform developed using CNN framework extracts information from data for the next time instance i.e. a step ahead of time and maps this data with geographical coordinates of power system components. The geographic information system (GIS) provides a visual indication of operating status of individual components as well as that of the entire system.

Optimal operation of microgrids under conditions of uncertainty

This paper proposes a method to optimally operate microgrids under conditions of uncertainty introduced by renewable energy sources, under both grid connected and islanded mode. Uncertainty is quantified with probability distribution and confidence levels are used to establish likelihood of forecast error. The optimization problem is formulated as a quadratic programming problem and the optimality of the solution is shown mathematically by proving the convexity of the problem. The optimization is carried out with the combined objective of minimizing total operating cost and carbon emission. The proposed optimization method is then tested against a priority controller for an extended time-horizon of 24 hours. Furthermore, under islanded mode of operation, for extended time-horizon, a key decision making task of whether to provide energy to non-critical loads or to store excess energy is addressed.

CNN based power system transient stability margin and voltage stability index prediction

Operators at electric grid control centers are faced with the task of making important decisions in real-time. With the plethora of data available it becomes important to extract information from the available data, based on which knowledge of system condition can be formed. This knowledge can then be used in decision making. Metrics such as transient stability margin (TSM) and voltage stability load index (VSLI) help in assessing the stability of the system. In this study, cellular neural network (CNN) based stability margin prediction system is developed in a distributed computing framework. The developed system not only extracts information from available data but also predicts the same, one step ahead of time. Moreover, the framework employed uses distributed computing and hence could be used on a large scale power system with a linear increase in computation time instead of an exponential increase. A reduced version of New Zealands South Island power system is used as the test system to demonstrate the feasibility of CNNs for TSM and VSLI prediction.

A MILP formulation for utility scale optimal demand side response

A detailed mathematical model for optimal utilization of demand side response is formulated. Formulations for constrained controllable load (dependent load) is modeled, whose operation is dependent on certain other loads completing their schedule prior to the constrained controllable loads start time. In addition, several other types of load models, including electric vehicles (EV), adjustable and constant power loads are modeled. The ensuing problem is that of mixed integer linear programming (MILP) type. Using the developed MILP formulation optimal solution to utility scale demand response (DR) scheme is obtained in real-time (duration between dispatches). In particular, the work is aimed at reliability based DR schemes used by independent system operators (ISOs) and regional transmission organizations (RTOs). Applicability to utility scale DR schemes is shown by formulation and solution of large problems with more than one million variables - solved in real-time.

Battery Energy Storage Systems

Battery Energy Storage Systems (BESS) are becoming more popular due to their positive attributes such as flexibility, lower cost, modularity, etc. This paper provides a mathematical approach to quantify its financial and operational impacts. Specifically, two test cases are investigated: intermittency mitigation of a solar farm using a BESS and optimal scheduling of microgrid with BESS. The former problem is formulated as a mixed integer linear programming problem while the latter is formulated as a linear programming problem. Both algorithms provided accurate solutions while achieving optimality. Finally, a financial model for the optimal operation of an advanced lead-acid BESS is investigated.

Partial right eigenstructure assignment based design of robust damping controller

The paper presents the design of an output feedback based centralized, damping controller using partial right eigenstructure assignment technique. The algorithm exploits the extra degrees of freedom provided by multi-input-multi-output (MIMO) systems in partially assigning a few, selected eigenvectors in addition to corresponding eigenvalues. Singular value decomposition (SVD) is used for determining the basis vectors for the achievable closed-loop eigenvectors. A derivativefree, direct search optimization technique has been used to solve the controller design problem. Robustness of the damping of inter-area modes to operating point changes is achieved by incorporating multiple models of the system in the optimization problem. The designed controller provides a supplemtary signal to voltage reference of a static VAr compensator (SVC). Remote feedback signals have been utilized and feedback signal selection is based on the input/output controllability metric. An algorithm explaining the framework and the results of its implementation on the IEEE 68 bus system has been provided. The results show the effectiveness of the controller in exploiting the eigenstructure of the system for damping inter-area oscillations.