Short-term load forecasting of powersystem is not only the basis of scheduling of generating sets, but also the basis of working out the transaction schedule in electricity market. This paper proposes a short-term lo...
详细信息
Short-term load forecasting of powersystem is not only the basis of scheduling of generating sets, but also the basis of working out the transaction schedule in electricity market. This paper proposes a short-term load forecasting method based on combination of radial basis function (RBF) neural network and fuzzy control, uses on-line self-modify factor fuzzy control to eliminate forecast error on the basis of RBF neural network forecasting. The practical examples show that the accuracy of short-term load forecasting and training speed can be improved and gains the very satisfactory results by the proposed method
A ion migration testing for DC ceramic support insulators was designed and conducted in this paper. After the long-term ion migration testing, the change of Na + distribution in the tested samples was analyzed, and t...
详细信息
A ion migration testing for DC ceramic support insulators was designed and conducted in this paper. After the long-term ion migration testing, the change of Na + distribution in the tested samples was analyzed, and the mechanical performance of the tested samples was verified. It is indicated from our results that the ion migration testing in our laboratory doesn't influence the mechanical performance for the ceramic support insulators
To avoid the mal-operation of transformer differential protection when out-zone fault occurs under current transformer (CT) saturation, a new algorithm based on the mathematical morphology is put forward in this paper...
详细信息
To avoid the mal-operation of transformer differential protection when out-zone fault occurs under current transformer (CT) saturation, a new algorithm based on the mathematical morphology is put forward in this paper. Unlike the traditional time-differential method, which is based on the time difference between fault occurrence and the differential current increase, this algorithm only needs detecting the time of fault occurring, while the time of differential current increasing is no more needed. A new current waveform is obtained by a symmetrical transformation of the initial part of differential current after fault occurred, and then its characteristic is extracted by morphological gradient. In the case of CT's serious saturation, a discrimination of out-zone fault and inner-zone fault can be clearly made by this algorithm, which needs a little calculation and a short data window. Simulation results of EMTP validate its feasibility and reliability.
The reflection and refraction of fault-generated wave on EHV lines were investigated and the equations of transient current detected by relay were deduced. Theoretical analysis shows that the protection based on the e...
详细信息
The reflection and refraction of fault-generated wave on EHV lines were investigated and the equations of transient current detected by relay were deduced. Theoretical analysis shows that the protection based on the energy ratio of higher frequency components is more adaptive to various faults and can distinguish between the internal and external faults near the remote busbar more effectively. Such protection also has similar performance when used on EHV lines and series compensated EHV lines. The wavelet packet transform technique was employed to determine the energy ratio. Simulation results show that the proposed conclusions are correct.
A method of powersystem cascading outages assessment based on support vector machine (SVM) was provided in this paper. Taking full advantages of the ability of SVM to solve the problem with small sample, nonlinear an...
详细信息
A method of powersystem cascading outages assessment based on support vector machine (SVM) was provided in this paper. Taking full advantages of the ability of SVM to solve the problem with small sample, nonlinear and high dimension, this method constructs an assessment mode which uses resampling classification technique based on boosting method, pretreatment technique based on distance criterion and information fusion technique. This mode can help to improve the precision and stability of assessment, hence better generalization ability of assessment is guaranteed. At the end of this paper, an example is provided to demonstrate the validity of the proposed approach.
Attributes are require to be discrete when using data mining methods to establish forecasting model, for example, rough set, ANN and decision tree. But most discrete methods used before can't consider the relation...
详细信息
Attributes are require to be discrete when using data mining methods to establish forecasting model, for example, rough set, ANN and decision tree. But most discrete methods used before can't consider the relationship of conditional and decision attributes well, so the discrete results are not very satisfaction. This paper presents a new discrete method which combines clusters and entropy. First, cluster is adopted to get the waiting breakpoints. Then finally results are decided by their rate of entropy. According to clustering, the boundary value is greatly reduced. Meanwhile, through entropy which considers the condition attributes and decision attributes, attributes can be discreted more exactly. At last, the discrete results compared with the method based on the importance of breakpoints show that the proposed method assurance the quality of classing. The space of sample can be compressed effectively, which lightens the training load of model and the gotten rules are with good adaptability and useful value. The correctness and effectiveness are validated by the practical application and the accuracy of forecasting will be increased.
A novel method is put forward, which the power meter can be read based on compute vision, and the architecture of remote meter auto reading system is designed. Different with the normal automatic meter reading system ...
详细信息
A novel method is put forward, which the power meter can be read based on compute vision, and the architecture of remote meter auto reading system is designed. Different with the normal automatic meter reading system that must obtain the value from digital communication interface, the new method recognizes the meter reading utilizing the computer vision technique, it is a kind of non-contact measurement. In the meter automatic reading, the computer vision discrimination, incorporate human's ability of visual and apprehensive models. Not only the digital display power meter can be read, but also the direct-reading instrument can be fetched. In this paper, the scheme of remote automatic reading is introduced, the local meter reading information of distributed substation can be transmitted to center using computer network and then the pretreatment, segmentation and pattern matching, are detailed description in the meter image recognition process. Via energy meter automatic recognition experiment, the results based computer vision is accurate enough to monitor the remote substation running parameters
Oriented to wye-delta connected three phase transformer, a new configuration of current transformer (CT) in transformer differential protection is proposed. In this configuration, only one single phase CT connected in...
详细信息
Oriented to wye-delta connected three phase transformer, a new configuration of current transformer (CT) in transformer differential protection is proposed. In this configuration, only one single phase CT connected in any phase of the delta side is more needed than the traditional scheme. Worked with line currents measured by other current transformers connected in both sides of the transformer, any phase current in the delta side can be calculated based on the Kirchhoff's Current Law. This new scheme ensures the accurate calculation of the equivalent instantaneous inductance, which is immune to the loop current existed in the delta side of transformer, as well as no dead zone in transformer differential protection. A new algorithm based on the distribution of the normalized equivalent instantaneous inductance for the discrimination between inrush current and fault current is proposed. The normalization of equivalent instantaneous inductance makes the setting-value have little relation with the parameter of transformers, which has more universal meaning in the determination of setting values. Utilizing the distributive characteristic of the normalized equivalent instantaneous inductance, the variety of the normalized equivalent instantaneous inductance is effectively revealed. At last, experiment data obtained from HYBRISIM (Hybrid Simulator) validates the reliability and sensitivity of this algorithm. This algorithm has good prospect of engineering application in the EHV transformer protection.
Unbalanced element brings a big problem to fault analysis of symmetrical components method. According with fault analysis of phase components, a residual voltage transformation method for short-circuit current calcula...
详细信息
Unbalanced element brings a big problem to fault analysis of symmetrical components method. According with fault analysis of phase components, a residual voltage transformation method for short-circuit current calculation is presented with matrix translation in the place of classical sequence networks connection, which is suitable for both phase coordinates and symmetrical components coordinates. Calculation formulas are proposed with matrices calculation to deal with unsymmetrical faults and parameter matrices for various faults are listed. Analysis of simultaneous faults with unbalanced parameter element is done without calculation qualities increased. Simple examples have shown that the method is practical and suitable for programming.
暂无评论