Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Electrical signal parameter measurement system
Reexamination Certificate
2000-01-06
2002-07-09
Hilten, John S. (Department: 2863)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Electrical signal parameter measurement system
C702S182000, C706S015000, C706S020000
Reexamination Certificate
active
06418385
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to a method for determining the location of a partial discharge in a high-voltage installation, in which signals produced by a partial discharge are detected and supplied to an evaluation device.
BACKGROUND INFORMATION
A method for investigating partial discharges is described in for example, European Patent No. 0 241 764 A1. This patent describes that high frequency measurement values of electrical partial discharges (PD) in a high-voltage installation are picked up by several sensors and are compared with particular values, which for their part arise through simulation of partial discharges in a simulated network that corresponds, with regard to the modeling, to the present high-voltage installation with respect to the high-frequency behavior. By comparison of the measured values with the values obtained in the simulation, the location and type of partial discharge can be inferred.
For the execution of the described method, a precise analysis of the present high-voltage installation is necessary with respect to the high-frequency behavior and the transmission characteristics. This requires the creation of a new simulation network for each installation. The operational expense for this is considerable.
Am the article entitled “Artificial self-organizing neural network for partial discharge source recognition,” Archiv fur Electrotechnik, 76 (1993), p. 149-154, describes for the investigation of electrical partial discharges with the aid of a previously trained neural network. The neural network is trained beforehand on the basis of actual partial discharges that are detected by a sensor. Using this method, causes corresponding to occurring partial discharges can be determined.
German Patent No. 43 33 259 describes a method for locating electrical disturbances in the form of short-circuit currents, which likewise works with the aid of a neural network trained with measurement data obtained by a single sensor. The result of the method specified there is a directional indication that relates to the direction in which a short circuit can be found, seen from a point of detection.
SUMMARY
An object of, the present invention is to provide a method for determining the location of a partial discharge in a high-voltage installation that is simple and can be carried out at low expense and with high precision, and that can be adapted to different designs of the high-voltage installation at low expense, yet nonetheless provides individually optimized results.
This object is achieved according to the present invention in that the evaluation device includes a neural network that has been trained beforehand by the input of partial discharge measurement data, measurement data detected by at least two sensors being supplied to the neural network for training purposes, with simultaneous stipulation of the locations of occurrence of the corresponding partial discharges, and measurement data of the at least two sensors likewise being supplied to the evaluation device for the investigation of a partial discharge.
Due to the use of a neural network for the evaluation, the high-voltage installation need no longer be analyzed with regard to its high-frequency behavior. It is no longer necessary, to determine a transmission coefficient, an attenuation coefficient, and a reflection coefficient for individual elements of the high-voltage installation. The adaptation of the evaluation device, i.e., the training of the neural network, is thus possible at an expense for each switching installation to be constructed that is considerably lower than is the case for the corresponding analysis of the high-voltage installation and the creation of a network simulation.
The precision and reliability of the determination of the results by the neural network is dependent primarily on the number and spatial distribution of the sensors, the quality of the training, and the number of training measurements. Standards can be determined for these, so that the method can also be carried out by non-qualified personnel.
The use of at least two sensors simplifies the location procedure considerably, and makes the results more precise. The analyzable database is significantly improved not only by the number of available data, but also by the different perspective of different sensors with respect to the location to be determined of the partial discharge.
The training of the neural network can also include disturbing signals that originate from genuine partial discharges in the high-voltage installation, to be distinguished from events that represent either regular occurrences (actuation of switches) or disturbances radiated in from the outside.
An advantageous embodiment of the present invention provides that before being input into the neural network, the measurement data is pre-processed in a processing device that forms a part of the evaluation device.
In this way, the measurement data can be prepared in such a way that the data has a high degree of significance with respect to the location of partial discharges, and that the result after passing through the neural network thus becomes still more reliable and precise.
Another advantageous embodiment of the present invention provides that more than two sensors are provided, the evaluation device being supplied only with the measurement data of the two sensors yielding the greatest signal strength of the partial discharge signals.
Through such a selection of the measurement data supplied to the evaluation device, a better degree of reliability is likewise achieved in the processing by the neural network.
Moreover, the present invention can advantageously be designed so that the measurement data is analyzed in the processing device with respect to frequency.
In this way, signals that do not originate from genuine partial discharges are not taken into account in the analysis, since actual partial discharge processes have a high correlation with the network frequency.
It is also possible to construct the present invention in such a way that the correlation of the measurement data with the network frequency, or with whole-number multiples thereof, is determined, and that only measurement data whose correlation value exceeds a determined value are supplied to the neural network.
Correspondingly pre-processed signals must then be supplied to the neural network already during the training thereof.
It is likewise possible to construct the present invention such that particular phase angles of the partial discharge signals to the useful frequency, and particular signal strengths, are respectively assigned the frequency with which these values occur in a partial discharge measurement, and that these frequency values are supplied to the neural network.
An additional advantageous embodiment of the present invention provides that the impulse curve of a partial discharge signal in the time domain is detected and is supplied to the neural network.
In this case as well, the neural network must be trained with correspondingly pre-processed signals.
REFERENCES:
patent: 5305235 (1994-04-01), Izui et al.
patent: 5419197 (1995-05-01), Ogi et al.
patent: 5537327 (1996-07-01), Snow et al.
patent: 5627941 (1997-05-01), Takahashi et al.
patent: 5838881 (1998-11-01), Nelson et al.
patent: 43 33 259 (1995-05-01), None
patent: 0 241 764 (1987-10-01), None
patent: 0 572 767 (1993-12-01), None
Patent Abstracts of Japan, vol. 107, No. 076 (P-1487), Feb. 16, 1993 & JP 04 279870 (Hitachi Cable Ltd.) Oct. 5, 1992.
A. Schnettler et al., “Artificial self-organizing neural network for partial discharge source recognition,” Archiv für Elektrotechnik 76 (1993) pp. 149-154.
Gorablenkow Jörg
Hücker Thomas
Bui Bryan
Hilten John S.
Kenyon & Kenyon
Siemens Aktiengesellschaft
LandOfFree
Method for determining the location of a partial discharge does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for determining the location of a partial discharge, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for determining the location of a partial discharge will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2885408