Xiaochuan Luo

Real power transfer capability calculations using multi-layer feed-forward neural networks

This paper proposes a neural network solution methodology for the problem of real power transfer capability calculations. Based on the optimal power flow formulation of the problem, the inputs, for the neural network are generator status, line status and load status and the output is the transfer capability. The Quickprop algorithm is used in the paper to train the neural network. A case study of the IEEE 30-bus system is presented demonstrating the feasibility of this approach. The new method will be useful for reliability assessment in the new utility environment.

Power System Adequacy and Security Calculations Using Monte Carlo Simulation incorporating Intelligent System Methodology

Monte Carlo simulation has been extensively used in reliability evaluation of electric power systems. One of the issues with this approach has been the computational time for convergence of indices being estimated, especially when the systems are highly reliable. Perhaps the most commonly used approach to deal with this problem has been some version of variance reduction techniques. Recently some publications have proposed use of intelligent systems techniques such as self-organizing maps and linear vector quantization to tackle this problem. This paper will provide a perspective on this hybrid approach using Monte Carlo Simulation and intelligent system methods. The philosophy of this hybridization as well some results will be discussed

Power system reliability evaluation using learning vector quantization and Monte Carlo simulation

Abstract Artificial Neural Networks (ANN) based on the Learning Vector Quantization (LVQ) algorithm have received considerable attention as pattern classifiers. This paper proposes a new method for power system reliability evaluation combining Monte Carlo simulation and LVQ which greatly reduces the computing burden of the loss of load probability calculation compared to Monte Carlo simulation only. A case study of the IEEE RTS system is presented demonstrating the efficiency of this approach.

An expert system for diagnosis of digital relay operation

This paper presents an expert system which performs detailed diagnosis of digital relay operation by analyzing data contained in relay files and reports. Problem domain is discussed first. Then the analysis strategy is detailed: forward chaining reasoning, logic reasoning and backward chaining reasoning are employed to predict protection operation, identify unexpected protection operation and diagnose symptoms respectively. The implementation of knowledge representation by CLIPS rules is further described. Finally, an example is given to demonstrate the capability of the expert system

Fault analysis based on integration of digital relay and DFR data

This paper discusses integration of two existing automated analysis applications, DFR data analysis and digital relay data analysis, to achieve comprehensive fault analysis. As inputs to the integrated application, digital relay files and reports are introduced. The proposed strategy and implementation of integration are outlined. An example is used to demonstrate features of the integrated application developed so far.

Power system reliability evaluation using self organizing map

Artificial neural networks (ANN) based on the self organizing map (SOM) algorithm has received considerable attention. This paper proposes a new method for power system reliability evaluation by combining Monte Carlo simulation and self organizing map which greatly reduces the computing burden of the loss of load probability calculation compared to Monte Carlo simulation only. A case study of the IEEE RTS system is presented demonstrating the efficiency of this approach.

Loss-of-load state identification using self-organizing map

This paper presents a method for classifying power system states as loss-of-load or not using Kohonens self-organizing map (SOM). The main feature of SOM is the ability to map input data from an n-dimensional space to a lower dimensional (usually two dimensional) space while maintaining the original topological relationships. Real and reactive power at each load bus and available real power generation at each generation bus are taken as input features. OPF calculations are performed on the weights of each neuron in the map to determine whether the neuron is representative of loss-of-load or not. The loss-of-load status of a new system state can be quickly identified by the loss-of-load status of the nearest neuron. An example illustrating the approach shows that the SOM can accurately classify the loss-of-load status of power system states. This proposed method is useful for power system operation, power system reliability assessment and state screening.

Automated analysis of digital relay data based on expert system

Modern digital protective relays generate various flies and reports which contain abundant data regarding fault disturbances and protection system operation. This paper presents an expert system based application for automated analysis of digital relay data. In this application, forward chaining reasoning is used to predict expected protection operation while backward chaining reasoning is employed to validate and diagnosis of actual protection operation. An EMTP/C++ based digital relay model with capability of insertion of user-defined errors and generation of files and reports is developed. The analysis capability of this application is tested using the relay model.

GridCloud: Infrastructure for Cloud-based Wide Area Monitoring of Bulk Electric Power Grids

The continuing rollout of phasor measurement units enables wide area monitoring and control (WAMS/WACS), but the difficulty of sharing data in a secure, scalable, cost-effective, low-latency manner limits exploitation of this new capability by bulk electric power grid operators. GridCloud is an open-source platform for real-time data acquisition and sharing across the jurisdictions that control a bulk interconnected grid. It leverages commercial cloud tools to reduce costs, employing cryptographic methods to protect sensitive data, and software-mediated redundancy to overcome failures. The system has been tested by ISO New England and the results reported here demonstrate a level of responsiveness, availability, and security easily adequate for regional WAMS/WACS, with the capacity for nation-wide scalability in the future.

Loss-of-load probability calculation using learning vector quantization

This paper proposes a new method employing learning vector quantization (LVQ) and Monte Carlo simulation to calculate the loss-of-load probability (LOLP) of power systems. LVQ is a type of classification method whose goal is to use data samples to position the codebook vector in such a way that the nearest neighbor classification method will result in the maximum classification accuracy. The proposed method greatly reduces the computing burden of the loss-of-load probability calculation compared to Monte Carlo simulation only. A case study of the IEEE RTS system is presented, demonstrating the efficiency of this approach.