窃电用户窃电量的精确估算对挽回电力企业的经济损失和依法处置窃电用户具有重要的实际意义.为实现用户窃电量精确估算,通过对时间序列回归算法进行优化改进,提出了一种新的窃电量估算方法.该方法以用户当前用电量时间序列样本和历史同期时间序列样本为基础,通过引入最大均值差异(maximum mean difference,MMD)得到基于最大均值差异-最小二乘支持向量回归(MMD-least square support vector regression,MMD-LSSVR)的半监督学习回归算法,以提高回归的准确度和数据的利用度、降低时间复杂度;同时,通过引入交叉变异人工蜂群算法(artificial bee colony based on crossover mutation,CMABC)对算法关键参数进行最佳适应度约束,以提高估算结果精度和收敛速度.在此基础上,提出了基于MMD-CMABC-LSSVR的窃电量估算方法.算法验证结果表明,采用所提方法,其估算电量与实际用电量的相对误差仅为2％,精度远优于传统方法;实际应用案例表明,所提方法可有效恢复窃电时间区段内窃电用户负荷和计量曲线,并精确估算出窃电量.所提方法为反窃电稽查工作提供了一种新的有效手段,具有良好的应用前景.

It is the key to realize reliable identification and diagnosis of discharge fault in distribution transformers in the study of acoustic characteristics of discharge fault noise and the noise of the distribution transformers itself and identification methods. We proposed a method for the diagnosis of discharge fault in distribution transformers based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN). Firstly, CEEMDAN was used to decompose the discharge fault acoustic signals, thus several intrinsic mode functions (IMF) could be obtained. The kurtosis of each IMF was sought to select the suitable IMF components for signal reconstruction and to extract the discharge acoustic signal. Then the CEEMDAN decomposition of the discharge acoustic signal was carried out again to obtain the marginal spectrum entropy, the center of gravity frequency, the band energy entropy and the singular spectrum entropy, which constitute the eigenvector. Finally, the support vector data description (SVDD) was used to classify and identify different typical discharge acoustic signals. The experimental results show that the proposed method can achieve more than 90％ of the recognition rate when considering the superimposed noise of the distribution transformer, which can be used for the identification and diagnosis of the discharge fault of distribution transformers.

The overheat fault of gas insulated switch-gear (GIS) disconnector contacts severely damages the reliable operation of equipment.In order to monitor the temperature rise and the overheat fault of GIS disconnector contacts,this paper establishes a 3D numerical coupled multi-physics model of eddy current-fluid-temperature of GIS with contacts.The resistance of contactor is replaced by a conductor with variable resistivity,and the current density and power loss of GIS disconnector are calculated through analyzing the electromagnetic field.The temperature of conductor and tank are calculated by using the finite element model whose heat source is power loss.The inversion law and influence factors of the temperature rise of GIS disconnector contacts are studied based on the coupled multi-physics model.The results show that the relationship between the contact of phase A,B and the temperature rise of tank is 1 ∶ 0.215,and the relationship between the contact C and the temperature rise of tank is 1 ∶ 0.165.The research results can provide theoretical basis and technical support for monitoring the temperature rise of GIS.

As an improved algorithm of LS-TSVM, RELS-TSVM has a high training speed, good robustness and strong generalization ability. This paper introduces it to the fault diagnosis of transformer and proposes a fault diagnosis model based on RELS-TSVM optimized by CSO algorithm. In the model, it constructs a multi-class classifier for transformer fault type identification by combining binary tree and RELS-TSVM, and uses CSO that has strong search performance to optimize the parameters of RELS-TSVM, which would upgrade the performance of the model to the best state. The example of transformer fault diagnosis based on DGA shows that the fault diagnosis model in this paper is of high accuracy and works better than PSO-SVM model.

Temperature field distribution of dry-type air-core shunt reactor is of much importance for the equipment safety and reliable operation.In order to improve the calculation precision of the temperature field of dry-type air-core reactor,based on the multi-field coupling finite element theory,an electromagnetic-fluid-temperature multi-field coupling numerical calculation model of dry-type air-core shunt reactor is established.The model is under the constraint of external circuit and the spacers between encapsulations are considered as well as the rainhat on the reactor.The loss of encapsulation calculated in electromagnetic field is imported into fluid-temperature field calculation as a heat source.Because the conductivity of the conductor is related to temperature,the encapsulation loss is iteratively calculated in the electromagnetic-fluid-temperature field.In this method,temperature distribution of dry-type air-core reactor is obtained.Compared with the infrared measuring temperature rise,the temperature rise error of the model without spacers nor rainhat is as high as 9％,however the accuracy of the calculation model considering spacers and rainhat is much better,with temperature rise error less than 3％.The temperature distribution analysis of the reactor encapsulation in the axial and radial on the top can provide a theoretical basis for dry-type air-core shunt reactor in product design and temperature rising online monitoring.

To overcome disadvantages of the conventional electric energy measurement system (CEEMS) in 10 kV distribution network, the study presents a high-voltage electrical energy meter (HVEEM) comprised of signal sensors, two measurement units, a synthesis unit, power supplies, and extended communication devices. With measurement chip circuit boards floating at 10 kV potentials, HVEEM accomplishes electrical energy measurement at the primary side. With the whole structure, HVEEM can be calibrated as a whole, and the synchronisation error makes no difference on its whole error. As the core component, the capacitive voltage divider (CVD) acts as not only the voltage sensor but also the power supply. The stability of CVDs, which is almost not affected by temperature, can be improved by the manner of voltage accelerated ageing for a certain time. CVDs for power supplies are able to output enough power for measurement chip circuit boards even if a single-phase ground or an opening fault occurs. The tests both in the laboratory and the field prove that HVEEM has substantial advantages over CEEMS especially in accuracy and reliability.

On the purpose of optimal design and online monitoring of three phase enclosure gas insulated bus (GIB), a 3-D circuit-field coupling FEM model has been developed. The current constriction effects in plug-in connectors are simulated by modeling contact bridges between contact surfaces and the influence of conductor gravity on contact resistance has been taken into account. The distributions of current which is constrained by external circuit are obtained from field-circuit coupling calculation and electromagnetic force, which is derived from electromagnetic field calculation, is used as load inputs in mechanical field analysis. The validity of calculation model is demonstrated by comparing with vibration experiments. The dynamic current distribution and electromagnetic force behaviors of three phase enclosure GIB under steady state and different short circuit conditions have been analyzed using the calculation model. Analysis results show that the uneven heating of contact fingers due to current distributions under steady state and contact fingers with smaller contact forces are seriously ablated by large short currents under short circuit conditions, and are the main contact degradation mechanism of plug-in connector. Conductor electromagnetic forces under single phase short circuit condition are larger than those of two phase and three phase short circuit conditions and the electromagnetic force peak moments under different fault conditions are not the same.

The acoustic emission signals under different discharge steps were decomposed by ensemble empirical mode decomposition (EEMD), the intrinsic mode function was obtained, and the instantaneous frequency of acoustic emission signals was gained by Hilbert transform on intrinsic mode function. On this basis, the marginal spectrum of acoustic emission signals was calculated, and the marginal spectrum entropy and gravity frequency of acoustic emission signals were used as characteristic value to realize pattern recognition of contamination discharge. Combining large number of contamination discharge experiments of insulators, the characteristic value of acoustic emission signal was used to analyze the contamination discharge pattern. The results show that the proposed method can effectively distinguish the three discharge steps in contamination discharge of insulator. This provides technical support to judge the insulation status of contamination insulators and realize flashover warning.

A 3-D electromagnetic-thermal-mechanical coupling finite element model of a gas-insulated bus plug-in connector is proposed in this paper with a subsequent analysis for the contact fatigue mechanism. Special attention has been paid to the contact degradation mechanism under normal cyclic operation conditions, which is important for the optimal designing and condition monitoring of the equipment. The current constriction effect and the contact resistance are considered through modeling the equivalent contact bridges between contact interfaces, and the current densities derived from the electromagnetic field are applied as load inputs in a electrical-thermal-mechanical analysis where varying of friction coefficient is considered. The validity of the calculation model has been demonstrated by temperature rise and relative motion experiments, and the influence of daily changing current and environmental temperature on the thermal and mechanical characteristics of plug-in connector has been analyzed. Analysis results show that the temperature rise among different contact fingers of plug-in connector is not the same with different contact forces, and the plastic deformation can be induced on contact spots. The insert depth of connector can be changed under the action of alternating thermal loading induced by daily change of current and environmental temperature, leading to contact degradation and overheating fault of plug-in connector.

基于瞬态场路耦合有限元理论建立了外部电压源激励下的三相共箱气体绝缘母线电路-电磁-结构场数值计算模型.采用顺序耦合方法将电磁场分析得到的节点电磁力作为载荷施加到结构场有限元模型中计算外壳的振动特性.通过将计算结果与镜像电流法和外壳振动测试结果进行对比验证该计算模型的有效性.基于该模型分析计算了工频稳态和不同短路故障条件下三相共箱气体绝缘母线电动力空间分布特性和时变特性,计算结果表明由于三相导体位于同一金属壳体内,短路故障类型将直接影响导体和外壳上电动力的分布特性.单相短路故障条件下气体绝缘母线导体短路电动力在幅值上要大于相间短路和三相短路且在时间上要早于相间短路和三相短路.计算模型和分析结果可用于三相共箱气体绝缘母线结构优化设计和短路故障监测.

This paper deals with the scale modeling method to investigate the overheat failure mechanism and the processes of bus contacts in gas-insulated switchgears (GIS). Mathematical models of the physical phenomenon in the overheat failure process are summarized to derive the coupled eddy current-fluid-heat field scaling relationships. In pursuit of better availability of the model, the scaling relationships are then simplified and a partial scale model is further designed with the dimension parameters, physical parameters, and boundary conditions presented. Temperature distributions and current densities of the scale model are compared with those of the prototype to verify the effectiveness of the scale model by 3-D finite-element method. The test scale model is fabricated, and the temperature rise tests are conducted to validate the correctness of the scale modeling and the simulation calculation.

Experimental studies have shown that Acoustic Emission(AE) signals generated by the polluted-insulator discharge contain information of discharge energy. To extract the frequency characteristics of AE signals in polluted-insulator discharge, algorithm combining the empirical mode decomposition(EMD) and fast fourier transform(FFT) is used. Through a great many of artificial contamination experiments, the AE signals in different contamination discharge stages are collected, and the frequency characteristics can be extracted by the method presented in this article. The results show that the frequency characteristics of AE signals can be effectively extracted by the proposed method, which gives the right corresponding relationship between frequency characteristics and the polluted-insulator corona, partial and arc discharge. It also provides technical support for monitoring the intensity of polluted-insulator discharge and the change of external insulation status. The method for extracting AE signal frequency characteristics proposed in this paper has been applied to on-line monitoring of polluted-insulator external insulation status and good results have been achieved.

Knowledge of the heat dissipation ability of gas-insulated bus bars (GIB) is paramount in the design stage. To reduce the capital cost, a scale model which has the identical electromagnetic-thermal characteristics of a full scale GIB is designed in this paper. The scaling relationships of the power losses, convection heat transfer, radiant heat transfer and thermal equilibrium are analyzed based on the governing equations and non-dimensional correlations. Current densities, power losses, convective heat transfer coefficients and temperature distributions in conductor and tank of the prototype and the scale models under different load currents are compared by FEM (Finite Element Method). The effectiveness of scale models is validated by the comparison between calculated and test results.

Least square support vector machine (LS-SVM) can solve small sample, high-dimensional and non-linear multi-classification problem well, so it is applicable to the power transformer fault diagnosis. However, the parameters of LS-SVM have significant effect on the classification results.In this paper, the adaptive differential evolution algorithm (ADE) is applied to optimize the parameters of LS-SVM. The scaling factor and crossover rate are adjusted dynamically in the whole evolution process, so the robustness of the algorithm is improved greatly. The optimized LS-SVM is applied to fault diagnosis of power transformer, the results obtained demonstrate superiority of the proposed approach.

Since it is difficult to determine the BPA(Basic Probability Assignment) in the transformer information fusion diagnosis method,an identification model based on the multi-SVM(Support Vector Machine) and D-S evidence theory is proposed for the interior fault position of electric transformer.The BPA is objectively realized based on the one-versus-one multi-class SVM posterior probability estimation;the complementary information of DGA(Dissolved Gas Analysis) data and routine electrical test data is fully utilized to detect the possible interior fault positions of electric transformer.Practical examples show that,the proposed model identifies the fault positions effectively and is better than the mono-SVM in both accuracy and generalization.

At present, most intelligent fault diagnostic methods of power transformer are based on dissolved gas analysis in oil to make diagnosis on fault property, which lacks of a quantitative diagnosis on inner fault position. To solve the problem, a novel probability estimation model of interior fault position for power transformer based on support vector machine (SVM) is proposed. The model takes advantages of SVM and probability modeling to make probability estimation on interior possible fault position of power transformer by fully utilizing dissolved gas analysis and routine electrical testing data. It overcomes drawbacks in hard-decision outputs of the traditional support vector machine and gives a probabilistic conclusion by using posterior probability support vector machine. Finally, fault diagnosis examples are used to illustrate the performance of the proposed model. The diagnostic results show that the proposed model has high recognition rate and better probability distribution, which proves its effectiveness and usefulness.

External insulation strength assessment of contaminated insulator is proposed in this paper using the acoustic emission signals generated when the polluted insulator flashover discharges. Systematic artificial contamination experiments were done. The acoustic emission signals generated from the polluted insulator were monitored by the sound monitoring device with high sensitivity. And the acoustic emission signals were analyzed. It shows that there is a relationship between the strength of filthy discharge and acoustic emission signals. The 14 features that can reflect acoustic emission signals of polluted insulator were extracted. Then a method of principal feature selection based on algorithm ReliefF is utilized. By using least squares support machine (LS-SVM), the classification model of is built.. After analysis and comparison, LS-SVM model has a higher accuracy in t classification of different external insulation strength stages.

目前变压器智能故障诊断大多是以油中溶解气体为特征对故障性质的诊断,缺乏对内部故障部位的分析及量化的诊断结果。针对上述问题,提出一种基于SVM的电力变压器内部故障部位的概率估计模型。该模型结合SVM与概率建模的优点,充分利用油中溶解气体和电气试验数据的互补信息,运用SVM后验概率理论,对变压器内部可能发生故障的部位进行概率估计,克服了标准SVM硬判决输出的缺陷,以概率的形式给出诊断结论。通过实例分析表明,该模型不仅故障识别率较高,还具有良好的概率分布形态,具有较好的实用性和推广性。

External insulation status assessment of contaminated insulator is proposed in this paper using the acoustic emission signals generated when the polluted insulator flashover discharges to improve the accuracy of diagnosis on insulator's external insulation.Through artificial contamination experiments,the acoustic emission signals generated from a polluted insulator were monitored by sound monitoring devices with high sensitivity.The acoustic emission signals were analyzed by using KPCA(kernel principle component analysis),so as to increase the number of features.Then random forests were constructed to get classifier groups in high dimensional kernel space.Finally,according to the voting result of the classifier groups,the final classification of the testing sample was gained.The analysis and experimental results show that,by converting the 3original features of the acoustic emission signals to 65kernel features through KPCA,the difference among classifier groups was improved effectively,and results of the state diagnosis based on KPCA and random forests had a higher accuracy.By using the acoustic emission signals generated from the polluted insulator discharge,the discharge phase can be distinguished,which realizes the monitoring of the external insulation status of insulators.

This paper constructs a feed-forward wavelet network by combining the wavelet analysis with BP neural network.The wavelet network takes Gauss wavelet and its telescopic translation system as the hidden layer of the wavelet basic function,and the wavelet neural network is improved.300 measured content data of dissolved gas in oil are selected as the samples for training the feed-forward wavelet network and faults recognition,and the simulation results are compared and analyzed.Experimental results demonstrate that the proposed improved wavelet network adapts to transformer fault diagnosis with better performance than the IEC,BPNN and the BP neural network based on principal component analysis.This method has been used in the actual transformer fault diagnosis project.