Yuping Lu

Swarm Intelligence for Optimal Reactive Power Dispatch

This paper presents swarm intelligence to solve optimal reactive power dispatch (ORPD) problem. Results derived with swarm intelligence, improved genetic algorithm and a conventional gradient-based optimization method were compared. The comparison shows that swarm intelligence would produce similar results on the IEEE 30-bus system as those from improved genetic algorithm while the conventional method was not as good as the other two in solving ORPD problem

New Artificial Neural Network Based Magnetizing Inrush Detection in Digital Differential Protection for Large Transformer

The impact of inrush on large transformer differential protection has not been solved satisfactorily. A new intelligent ANN based scheme for digital differential protection is proposed in this paper to distinguish inrush from internal fault in a transformer. The new scheme is based on multi-condition restraint which introduces voltage features as a criterion. Test results show the advantages in comparing with traditional second harmonic restraint method. The new intelligent scheme can meet the requirements for large transformer protection.

A digital sampling rate synchronization scheme for fully digital relay protection

The application of IEC61850 and non-conventional instrument transformers (NCIT) brings a new era in the development of fully digital protection system in which the protection IED receives sampled data from different digital data sources (DDS). The mismatch in sampling rate between the protection IED and the DDS requires sampling rate synchronization (SRS) to make sure the protection and other algorithms can run correctly. In this paper, a digital SRS scheme based on oversampling interpolator and quasi-continuous FIR interpolator is proposed, in which the SRS requirement in a protection IED can be realized. A poly-phase implementation structure for the oversampling interpolator is designed and the Lagrange interpolation algorithm is chosen to approximate the impulse response of the quasi-continuous FIR interpolator considering the reduction in error and computation. Design examples are given to demonstrate the effectiveness of the proposed approach.

A NEW FUZZY NEURAL NETWORK STRATEGY OF INSULATORS CONTAMINATION DETECTION FOR POWER SYSTEM TRANSMISSION LINE

The detection of insulators contamination is difficult in power systems because many factors can influence the pollution. The contamination condition of insulators is usually estimated by detecting the root mean square (r.m.s) of surface leakage current via online-monitoring system. It ignores the influence of environmental factors, such as temperature, humidity, etc. As these factors are fuzzy-characterized, a new method based on Fuzzy Neural Network (FNN) is proposed to improve traditional insulation contamination detection. The renewed structure of FNN is put forward. The evaluation of contamination severity of insulators is achieved through FNN, which are trained by the field samples. The results prove the validity of the method proposed in the paper and can be used to eliminate the insulator from flashover fault and improve the condition-based maintenance (CBM).

A New Fuzzy Neural Network Based Insulator Contamination Detection

The contamination condition of insulators is usually estimated by detecting the root mean square (r.m.s) of surface leakage current via online monitoring system, ignoring the influence of environment factors, such as temperature, humidity, etc. For the detection factors have fuzzy characters, a new method based on fuzzy neural network is proposed in order to overcome the disadvantages of traditional insulation condition detection. It is through the build of the structure of fuzzy neural network and the establishment of net weights by training samples as well to estimate the contamination severity of insulators. The test samples simulation experiment result proves the validity of the method presented in this paper, which shows an instructive significance for the prevention of the insulator from flashover fault and the condition-based maintenance (CBM)

Neural network based generator-transformer protection

This work proposes a new technique and method for application of artificial neural network (ANN) to protect generator transformer unit. This technique combines conventional generator protection theory developed through several decades with intelligent ANN technique. This combination promotes an effective approach to developing protective relays of superior performance and high reliability.

Application of intelligent algorithm in island detection of distributed generation

In the grid-connection, the distributed generation system in the status of the islanding is liable for damage to equipment, affecting the performance of the utility. Seriously, island may create a hazard for utility line-worker. Islanding detection methods for distributed generation system are reviewed. Characterized by high efficiency and performance, C4.5 decision-tree is particularly applicable to the condition of large amounts of mining data. The paper proposes a new approach based on C4.5 decision-tree for islanding detection in distributed generation system. Without any negative effect on the power quality, this novel method greatly reduces the damage to the utility resulting from the islanding running state, and also highly enhances the capability of detecting islands of the protection relay. How to construct C4.5 decision-tree on the basis of past operation data of an existed distributed generation system was introduced in detail firstly. And this method was tested on a typical distribution system with multiple distributed recourses by using Matlab/Simulink tools. The simulation results show that C4.5 decision-tree is effective and the island operating mode of DGs can be totally forecasted by this new algorithm.

Application of Intelligent Algorithm in Island Detection of Distributed Generation

In the grid-connection, the distributed generation system in the status of the islanding is liable for damage to equipment, affecting the performance of the utility. Seriously, island may create a hazard for utility line-worker. Islanding detection methods for distributed generation system are reviewed. Characterized by high efficiency and performance, C4.5 decision-tree is particularly applicable to the condition of large amounts of mining data. The paper proposes a new approach based on C4.5 decision-tree for islanding detection in distributed generation system. Without any negative effect on the power quality, this novel method greatly reduces the damage to the utility resulting from the islanding running state, and also highly enhances the capability of detecting islands of the protection relay. How to construct C4.5 decision-tree on the basis of past operation data of an existed distributed generation system was introduced in detail firstly. And this method was tested on a typical distribution system with multiple distributed recourses by using Matlab/Simulink tools. The simulation results show that C4.5 decision-tree is effective and the island operating mode of DGs can be totally forecasted by this new algorithm.

New distributed protection algorithm based on ANN and adaptive fault area search

DG has great negative effects on conventional relay protection in distribution networks. Decentralized protection strategy is more profitable mechanism to contain shortages and to satisfied fault selectivity of power system. It is known that voltage measure is difficult in distributed system (DS). The paper proposed an innovative way to decide fault direction without voltage measurement and developed an intelligent adaptive fault location by isolating DS. This locating algorithm can greatly reduce protection information collection requirement because of artificial neural network (ANN) introduction. With ANN, it can classify relevant fault area more easily and fault clearance will be more quick and sensitive. Case studies have been shown to support algorithm advantages.

A Framework of Intelligent Sampling Rate Conversion Algorithm for ECT Digital Power Plant

In the application of electronic current transformer (ECT), it is required to convert different sampling rates to the same rate, when the differential protection device is connected with the various ECTs of different sampling rates. In this paper, the sampling rate conversion algorithm based on decimation and interpolation is studied, which can realize the adaptive sampling rate conversion by a rational factor of L/M and provide an intelligent implementation structure for the algorithm - the time-varying filter structure with improved high-efficiency calculating ability. It can be shown from the examples that when converting the high sampling rate to low one, the algorithm not only has the sampling rate conversion function, but also has the ability to filter the high frequency element in it and then reduce the differential current.