Abnormality cause identifying method, abnormality cause identifying device, power converter and power conversion system

Patent metadata

Field	Value
Publication number	US-11275124-B2
Application number	US-201916726442-A
Country	US
Kind code	B2
Filing date	Dec 24, 2019
Priority date	Feb 12, 2019
Publication date	Mar 15, 2022
Grant date	Mar 15, 2022

Abstract

An abnormality cause identifying method that is applied to a computer is provided. The abnormality cause identifying method includes: outputting a control parameter that is calculated based on a detection value detected from a power converter that converts power supplied from a power supply and supplies the converted power to a load; plotting, on coordinates having at least two axes, a value that is calculated using the detection value and the control parameter; and identifying an abnormality cause based on a quadrant of the coordinates on which the value is plotted.

First claim

What is claimed is:

1. An abnormality cause identifying method that is applied to a computer, the abnormality cause identifying method comprising:
outputting a control parameter that is calculated based on a detection value detected from a power converter that converts power supplied from a power supply and supplies the converted power to a load where in the power convert includes a rectifier, a chopper and an inverter;
plotting, on coordinates having at least two axes, a value that is calculated using the detection value and the control parameter;
identifying an abnormality cause based on a quadrant of the coordinates on which the value is plotted,
wherein the identifying includes:
determining whether target data plotted on the coordinates is within a determination area preset according to a predetermined condition;

calculating a vector value based on each of an X-axis component value of the value and a Y-axis component value of the value in the case where the value is out of the determination area;
multiplying the vector value with
a normalization parameter to normalize the value,
a mode parameter determined by any of an operation state and an abnormal event of the power converter, and
a weight factor;

subtracting a bias value used for preventing an incorrect determination of an abnormality cause from the multiplied value;
automatically correcting the mode parameter, weight factor, and the bias value by a processor that is configured to run a trained neural network that is trained by the mode parameter, weight factor, and the bias value that are adjusted by a user, and
identifying, by the trained neural network, as the abnormality cause, an abnormal event corresponding to a greatest value among values exceeding zero in the case where the subtracted value exceeds zero, and
generating a message including at least one of the abnormality cause or the time when the abnormal event occurred and causing a terminal that is connected via network to display the message.

2. The abnormality cause identifying method according to claim 1 , further comprising:
applying a sampling process at a constant interval to the value that is calculated based on the detection value and the control parameter to obtain sampled values;
generating a first value obtained by calculating a square of each of the sampled values and by calculating a summation of the calculated squares included in a constant time period;
generating a second value obtained by calculating a square of each difference between consecutive sampled values at the constant interval and by calculating a summation of the calculated squares included in the constant time period; and
plotting the first value and the second value on the coordinates.

3. The abnormality cause identifying method according to claim 2 , further comprising:
applying a sampling process to the detection value at a constant interval to obtain sampled values,
generating a third value obtained by calculating a deviation of each of the sampled values from the control parameter and by calculating a summation of the deviations included in a constant time period; and
generating a fourth value obtained by calculating each difference between consecutive calculated deviations at the constant interval and by calculating a summation of the calculated differences included in the constant time period; and
plotting the third value and the fourth value on the coordinates.

4. The abnormality cause identifying method according to claim 3 , further comprising:
calculating a fifth value by applying Fourier transformation to a voltage value detected by the power converter;
calculating a sixth value by applying Fourier transformation to a current value detected by the power converter;
plotting the fifth value and the sixth value on the coordinates.

5. The abnormality cause identifying method according to claim 4 , further comprising:
generating a seventh value using an alternating current (AC) output power detected by the power converter;
generating an eighth value using a charge and discharge power from a chopper unit included in the power converter;
plotting the seventh value and the eighth value on the coordinates.

6. The abnormality cause identifying method according to claim 5 , further comprising:
applying a sampling process to the detection value at a constant interval to obtain sampled values;
generating a tenth value by calculating a square of each of the sampled values, by calculating, as a ninth value, a summation of the calculated squares included in a constant time period, and by calculating, as the tenth value, a difference between the ninth value and an average value of the ninth value;
generating a twelfth value by calculating a square of each difference between consecutive sampled values at the constant interval, by calculating, as an eleventh value, a summation of the calculated squares included in the constant time period, and by calculating, as the twelfth value, a ratio between the eleventh value and the ninth value; and
plotting the tenth value and the twelfth value on the coordinates.

7. The abnormality cause identifying method according to claim 4 , further comprising:
performing plotting in such a way that an average value calculated using the detection value and the control parameter is plotted at a zero point of the coordinates.

8. An abnormality cause identifying device for identifying an abnormality of a power converter that converts power supplied from a power supply and supplies the converted power to a load, the power converter including a rectifier, a chopper and an inverter, the abnormality cause identifying device comprising:
an operation mode control adjustment unit configured to output a control parameter calculated based on a detection value detected from the power converter;
a data analysis unit configured to plot, on coordinates having at least two axes, a value that is calculated using the detection value and the control parameter; and
an identifying unit configured to identify an abnormality cause based on a quadrant of the coordinates on which the value is plotted,
wherein
the data analysis unit determines that an abnormality has occurred in the case where the plotted value exceeds a predetermined range,
the data analysis unit calculates a distance indicating how far the plotted value deviates from a predetermined range in the case where the plotted value exceeds the predetermined range, and outputs the calculated distance to the identifying unit,
the operation mode control adjustment unit outputs an operation mode information indicating an operation state of the power converter determined based on the detection value,
the identifying unit selects an operation mode parameter corresponding to any of abnormal events in addition to the operation mode information, and identifies an abnormality cause based on the operation mode parameter, the vector value and a preset weight factor, and
the identifying unit multiplies the operation mode parameter by the weight factor, subtracts a bias value used for preventing an incorrect determination of an abnormality cause from the multiplied value, and run a trained neural network that is trained by the mode parameter, weight factor, and the bias value that are adjusted by a user so as to automatically correct the mode parameter, the weight factor, and the bias value and identify an abnormal event corresponding to a largest value among values exceeding zero as an abnormality cause in the case where the subtracted value exceeds zero, and wherein
the identifying unit is further configured to generate a message including at least one of the abnormality cause or the time when the abnormal event occurred and cause a terminal that is connected

Assignees

• Fuji Electric Co Ltd

Inventors

• Umezawa Kazuyoshi
• Ishida Satoshi
• Tsukuta Motohiro
• Noto Yasuyuki

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• H02H7/26

  Sectionalised protection of cable or line systems, e.g. for disconnecting a section on which a short-circuit, earth fault, or arc discharge has occured (locating faults in cables G01R31/08) · CPC title

• H02H3/00

  Emergency protective circuit arrangements for automatic disconnection directly responsive to an undesired change from normal electric working condition with or without subsequent reconnection (specially adapted for specific types of electric machines or apparatus or for sectionalised protection of cable or line systems H02H7/00; systems for change-over to standby supply H02J9/00 ){; integrated protection (for motors H02H7/0822)} · CPC title

• H02H1/0092

  concerning the data processing means, e.g. expert systems, neural networks · CPC title