pattern recognition applications for power system disturbance classification
TRANSCRIPT
![Page 1: Pattern Recognition Applications for Power System Disturbance Classification](https://reader030.vdocuments.us/reader030/viewer/2022020607/575076aa1a28abdd2e9fa5f0/html5/thumbnails/1.jpg)
Keywords: Duality, interconnected power systems, optimizationmethods, power distribution, power generation dispatch, power industry.
Preprint Order Number: PE-737PRS (10-2001)Discussion Deadline: March 2002
Incremental Transmission LossAllocation Under Pool Dispatch
Galiana, F.D.; Conejo, A.; Kockar, I.
Author Affiliation: McGill University, Montreal, Quebec, Can-ada; University of Castila-La Mancha, Spain
Abstract: Incremental transmission loss analysis has been used fordecades, but recent interest in its application to loss allocation calls fornew in-depth results. This paper demonstrates that, for incrementalmethods to be applied correctly in loss allocation, it is first necessary tospecify the load distribution and loss supply strategies. Incrementalloss allocation among bus power injections is shown to be arbitraryand, therefore, open to challenge as discriminatory. Loss allocation ispossible among incremental loads and/or generators, but the proportionof the total losses assigned to either one is arbitrary. Unique,nonarbitrary incremental loss allocations are, however, possible amongthe "equivalent" incremental bilateral exchanges between generatorsand loads. From these basic components it is possible then to calculatethe allocation among generators or loads in any specified proportion.The main results, although developed initially for small increments, areextended to large variations. Finally, a general incremental loss alloca-tion algorithm is developed and tested.
Keywords: Incremental transmission loss analysis, loss allocation,loss supply, pool dispatch, equivalent bilateral contracts, numerical al-gorithm.
Preprint Order Number: PE-106PRS (10-2001)Discussion Deadline: March 2002
Including Uncertainty in LOLECalculation Using Fuzzy Set TheoryKim, J.O.; Singh, C.
Author Affiliation: Hanyang University, Korea; Texas A&M Uni-versity, College Station, TX
Abstract: This paper presents a conceptual possibilistic approachusing fuzzy set theory to manage the uncertainties in the reliability in-put data of real power systems. In this paper, an algorithm is introducedto calculate the possibilistic reliability indices according to the degreeof uncertainty in the given data. The probability distribution functioncan be transformed into an appropriate possibilistic representation us-ing the probability-possibility consistency principle (PPCP) algorithm.In this algorithm, the transformation is performed by making a compro-mise between the transformation consistency and the human experi-ence. Fuzzy classification theory is applied to reduce the number ofload data points. The fuzzy classification method determines the close-ness of load data points by assigning them to various clusters and thendetermining the distance between the clusters. The IEEE-RTS with32-generating units is used to demonstrate the capability of the pro-posed algorithm.
Keywords: Loss of load expectation, forced outage rate, probabil-ity-possibility consistency principle, fuzzy clustering
Preprint Order Number: PE-030PRS (10-2001)Discussion Deadline: March 2002
An Empirical Study of Applied Game Theory:Transmission Constrained Cournot Behavior
Cunningham, L.B.; Baldick, R.; Baughman, M.L.
Author Affiliation: University of TexasAbstract: Restructured energy markets present opportunities for the
exercise of market power, meaning market players can potentially main-tain prices in excess of competitive prices. In this paper, we investigateCoumot equilibrium in a simple example network. We analyze three mar-
ket players in a transmission-constrained system and consider nonconstantmarginal cost. Several scenarios are evaluated that show a pure strategyequilibrium can break down even when a transmission constraint exceedsthe value of the unconstrained Coumot equilibrium line flow.
Keywords: Game theory, Coumot, transmission constraints, pric-ing of power, market models
Preprint Order Number: PE-761PRS (10-2001)Discussion Deadline: March 2002
Genetic Algorithms Based Economic Dispatchfor Cogeneration Units Considering MultiplantMultibuyer Wheeling
Hong, Y.Y.; Li, C.
Author Affiliation: Chung Yuan UniversityAbstract: A new method based on genetic algorithms (GA) is pro-
posed for optimal dispatch among multi-plant (cogeneration systems)with multi-cogenerators, which transmit MW to designated buyers(load buses) via wheeling. The operation constraints in thecogeneration systems and security constraints in the third party (trans-mission system owner) were considered. Varying weighting coeffi-cients for penalty functions and determination of gene variables for GAwere discussed. The IEEE 30- and 118-bus systems were used as testsystems to illustrate the applicability of the proposed method.
Keywords: Cogeneration, genetic algorithms, wheelingPreprint Order Number: PE-612PRS (10-2001)Discussion Deadline: March 2002
Production Cost Minimization Versus ConsumerPayment Minimization in Electricity Pools
Vasquez, C.; Rivier, M.; Perez-Arriaga, IJ.P.
Author Affiliation: Universidad Pontificia Comillas, Madrid, SpainAbstract: Algorithms that involve some kind of optimization have
been adopted by several electricity pools as a tool to clear the market.Traditionally, this kind of model was used on a cost-minimizing basis,but recent papers have pointed out that alternative dispatches may beobtained that, even with higher production costs, result in lower elec-tricity prices for consumers. This paper studies this new pay-ment-minimization approach, including the long-term economicsignals that it provides and their impact on future investments. Our re-sults show that minimizing consumer payment results in discrimina-tory scheduling for certain generation resources and may cause, in thelong run, higher prices for consumers.
Keywords: Electricity auctions, market design, unit commitment,marginal pricing, power generation dispatch, deregulation.
Preprint Order Number: PE-482PRS (10-2001)Discussion Deadline: March 2002
Power System Instrumentation Measurement
Pattern Recognition Applications forPower System Disturbance Classification
Gaouda, A.M.; Kanoun, S.H.; Salama, M.M.A.; Chikhani, A.Y.
Author Affiliation: University of Waterloo, Waterloo, Ontario,Canada; Royal Military College, Kingston, Ontario, Canada
Abstract: This paper presents an automated on-line disturbanceclassification technique. This technique is based on waveletmultiresolution analysis and pattem recognition techniques. The wave-let-multiresolution transform is introduced as a powerful tool for fea-ture extraction in order to classify different disturbances. MinimumEuclidean distance, k-nearest neighbor, and neural network classifiersare used to evaluate the efficiency of the extracted features.
IEEE Power Engineering Review, January 2002 69
![Page 2: Pattern Recognition Applications for Power System Disturbance Classification](https://reader030.vdocuments.us/reader030/viewer/2022020607/575076aa1a28abdd2e9fa5f0/html5/thumbnails/2.jpg)
Keywords: Power quality, wavelet analysis, multiresolution signaldecomposition, minimum Euclidean distance, k-nearest neighbor, andneural network recognition techniques.
Preprint Order Number: PE-002PRD (10-2001)Discussion Deadline: March 2002
Power System Operation
One-Hour-Ahead Load ForecastingUsing Neural Networks
Senjyu, T.; Takara, H.; Uezato, K.; Funabashi, T.
Author Affiliation: University of the Ryukyus, Okinawa, Japan;Meidensha Corporation
Abstract: Load forecasting has always been an essential part of anefficient power system's planning and operation. Several electric powercompanies are now forecasting load power based on conventional meth-ods. However, since the relationship between load power and factors in-fluencing load power is nonlinear, it is difficult to identify itsnonlinearity by using conventional methods. Most papers deal with24-hour-ahead load forecasting or next-day peak load forecasting. Thesemethods forecast the demand power by using forecasted temperature asforecast information. But, when the temperature curves change rapidlyon the forecast day, load power changes greatly and forecast error in-creases. In conventional methods, neural networks use similar-day datato learn the trend of similarity. However, learning of all similar day'sdata is very complex, and it does not suit learning of neural networks.Therefore, it is necessary to reduce the neural network structure andlearning time. To overcome these problems, we propose aone-hour-ahead load forecasting method using the correction of simi-lar-day data. In the proposed prediction method, the forecasted loadpower is obtained by adding a correction to the selected similar day data.
Keywords: Load forecasting, recurrent neural network, on-linelearning.
Preprint Order Number: PE-396PRS (10-2001)Discussion Deadline: March 2002
Locational Market Power Screening and CongestionManagement: Experience and Suggestions
Gan, D.; Bourcier, D.V.
Author Affiliation: ISO New England, Inc., Holyoke, MAAbstract: The existing New England electricity market follows a
uniform pricing system: infra-marginal generators are paid a uniformclearing price while constrained-on generators are paid bid prices subjectto a locational market power screen. In this paper, we first report the op-eration experience of this approach during the market's first two years.We then briefly review the NEPOOL locational pricing proposal beingimplemented. Two new approaches for locational market power screen-ing are presented. The first one is based on a zonal network model andthe second is based on a nodal transmission model. Both approaches arebeing evaluated by an independent consultant and are be to recom-mended to NEPOOL. Test results of both approaches are included.
Keywords: Market power, power system, electricity market,microeconomics, linear programming.
Preprint Order Number: PE-792PRS (10-2001)Discussion Deadline: February 2002
Application of a Robust Algorithm for DynamicState Estimation of a Power SystemShih, K.R.; Huang, S.J.
Author Afflliation: National Cheng Kung University, TaiwanAbstract: In this paper, a computation algorithm using the expo-
nential function to increase the robustness of the dynamic state estima-tion is proposed. The merit of this approach lies in its immunity to the
polluted measurements, while the implementation of the method is notcomplicated when compared with other methods. To validate the effec-tiveness of the proposed method, it was tested through several examplepower systems under different scenarios that include normal operation,bad measurements, sudden load change, and topology error conditions.From test results, they help support the feasibility of the method foi-state estimation applications.
Keywords: Dynamic state estimation, robust algorithm.Preprint Order Number: PE-650PRS (10-2001)Discussion Deadline: March 2002
Cooperative Coevolutionary Algorithmfor Unit Commitment
Chen, H.; Wang, X.
Author Affiliation: Xi'an Jiaotong University, Xi'an, ChinaAbstract: This paper presents a new cooperative coevolutionary al-
gorithm (CCA) for power system unit commitment. CCA is an exten-sion of the traditional genetic algorithm (GA), which appears to haveconsiderable potential for formulating and solving more complex prob-lems by explicitly modeling the coevolution of cooperating species.This method combines the basic ideas of Lagrangian relaxation tech-nique (LR) and GA to form a two-level approach. The first level uses asubgradient-based stochastic optimization method to optimizeLagrangian multipliers. The second level uses GA to solve the individ-ual unit commitment subproblems. CCA can manage more compli-cated time-dependent constraints than conventional LR. Simulationresults show that CCA has a good convergent property and a significantspeedup over traditional GAs and can obtain high-quality solutions.The "curse of dimensionality" is surmounted, and the computationalburden is almost linear with the problem scale.
Index Terms: Evolutionary optimization, GA, Lagrangian relax-ation, unit commitment.
Preprint Order Number: PE-564PRS (10-2001)Discussion Deadline: March 2002
An Improved Tabu Search forEconomic Dispatch with Multiple MinimaLin, W.M.; Cheng, F.S.; Tsay, M.T.
Author Affiliation: National Sun Yat-Sen University, Taiwan;Cheng-Shiu Institute of Technology
Abstract: This paper develops an improved tabu search algorithm(ITS) for economic dispatch (ED) with noncontinuous and nonsmoothcost functions. ITS employs a flexible memory system to avoid the en-trapment in a local minimum, and developed the ideal of "distance" tothe fitness to accelerate optimization. The new approach extends sim-ple tabu search algorithm (STS) to real valued optimization problem,and applies parallelism to weaken the dependence of the convergencerate of modified tabu search algorithm (MTS) on the initial condition.Effectiveness of the method was compared with many conventionalmethods. Results show that the proposed algorithm can provide accu-rate solutions with reasonable performance, and has a great potentialfor other applications in the power system.
Keywords: Economic dispatch, evolutionary programming, simpletabu search, modified tabu search, improved tabu search, a move, taburestrictions, aspiration criteria, adaptive progressing scheme,recombination.
Preprint Order Number: PE-339PRS (10-2001)Discussion Deadline: March 2002
Unit Connuitment Solution MethodologyUsing Genetic AlgorithmSwarp, K.S.; Yamashiro, S.
Author Affiliation: Indian Institute of Technology Madras, India;Kitami Institute of Technology, Hokkaido, Japan
IEEE Power Engineering Review, January 200270