KR20210133635A - Power conversion system - Google Patents

Abstract

An objective of the present invention is to provide a power conversion system capable of self-diagnosing failure of a power converter. The power conversion system comprises: a human machine interface (HMI); a power converter to convert DC voltage into AC voltage or convert the AC voltage into the DC voltage; and a controller to control the power converter according to a first command input to the HMI. The controller comprises: a learning model generator for applying a plurality of first fault data corresponding to a plurality of faults caused due to a real fault experiment with respect to the power converter and a plurality of second fault data corresponding to a plurality of faults caused due to a simulation experiment to learn, and to generate a fault diagnosis module for diagnose a plurality of faults; and a diagnosis determining unit for applying sensor data input from a plurality of sensors to the fault diagnosis model during an operation of the power converter to diagnose faults in the power converter.

Description

Power conversion system

The present invention relates to a power conversion system, and more particularly, to a power conversion system capable of self-diagnosing a failure of a power conversion unit.

In general, a power conversion system (Power Conversion System) is a system for converting power, a system for supplying a voltage output from a renewable energy source or other power source to a load according to the voltage and frequency level of the load.

Such a power conversion system is being developed as a power storage device (ESS), a solar power generation device, and the like, and is being developed to convert large-capacity power ranging from several hundred KW to several tens of MW.

The power conversion system includes a power conversion unit that converts DC voltage to AC voltage or converts AC voltage to DC voltage, a Human Machine Interface (HMI) that displays the operation status of the power conversion unit, and a control unit that controls the power conversion unit and HMI. may include

Here, the control unit may determine a failure of the power conversion unit based on sensor data input from a plurality of sensors, and output a failure alarm to the HMI.

In general, the power conversion system does not determine whether an internal failure or an external failure (eg, renewable energy source, load, etc.) have.

In recent years, research has been conducted to reduce the time required for diagnosis of a power conversion system for each failure.

An object of the present invention is to provide a power conversion system capable of self-diagnosing a failure of the power conversion unit.

In addition, an object of the present invention is to provide a power conversion system that can reduce the time required for failure diagnosis in real time by applying a failure diagnosis model generated by learning a plurality of failure data obtained through actual failure tests and simulation experiments for each failure. is in providing.

The objects of the present invention are not limited to the above-mentioned objects, and other objects and advantages of the present invention not mentioned may be understood by the following description, and will be more clearly understood by the examples of the present invention. Moreover, it will be readily apparent that the objects and advantages of the present invention may be realized by the means and combinations thereof indicated in the claims.

The power conversion system according to the present invention includes a Human Machine Interface (HMI), a power converter for converting a DC voltage to an AC voltage or converting an AC voltage to a DC voltage, and the power conversion according to a first command input to the HMI a control unit for controlling a unit, wherein the control unit is configured to respond to a plurality of first failure data corresponding to a plurality of faults induced by an actual failure test for the power conversion unit and the plurality of failures caused by a simulation experiment When a plurality of corresponding second failure data is applied to a set machine learning algorithm to learn, and a learning model generating unit that generates a failure diagnosis model for failure diagnosis for the plurality of failures and the power conversion unit operate, a plurality of sensors It may include a diagnosis determination unit for diagnosing a failure of the power conversion unit by applying the sensor data input from the to the failure diagnosis model.

The learning model generation unit divides the plurality of first and second failure data into learning data, test data and confirmation data for each failure, and applies the learning data to the machine learning algorithm to generate the failure diagnosis model A model evaluation unit that calculates an evaluation hit for the failure diagnosis model by applying the test data to the failure diagnosis model, and if the evaluation hit level is greater than or equal to a set reference hit level, applying the confirmation data to the failure diagnosis model It may include a model confirmation unit for confirming a normal prediction for the fault diagnosis of the power conversion unit.

The machine learning algorithm may be at least one of Python, Numpy, Pandas, and Keras.

The model evaluator may compare the plurality of failures with corresponding failures diagnosed by applying the test data to the failure diagnosis model to calculate the evaluation hit degree for each failure.

The model confirmation unit may control the model generation unit to regenerate the failure diagnosis model when the evaluation hit degree is less than the reference hit degree.

The model generator may regenerate the failure diagnosis model by re-classifying the plurality of first and second failure data into learning data, test data, and confirmation data for each failure under the control of the model confirmation unit.

The diagnosis determination unit, by applying the sensor data to the failure diagnosis model, diagnoses whether the power conversion unit is faulty, and additional fault information derived according to the failure information and the failure information of the power conversion unit when the power conversion unit is faulty diagnosed can create

The diagnosis determination unit may transmit the failure information and the additional failure information to an external device displayed and set on the HMI.

The failure information may include a failure alarm and a failure occurrence time for informing the content of the failure, and the additional failure information may include an additional failure alarm for notifying the additional failure content generated by the failure alarm.

The power conversion system according to the present invention has the advantage of reducing the time required for fault diagnosis by applying sensor data input from a plurality of sensors to a set fault diagnosis model.

In addition, the power conversion system according to the present invention has the advantage of improving the reliability of the fault diagnosis by being able to update the fault diagnosis model using the sensor data input during fault diagnosis.

In addition to the above-described effects, the specific effects of the present invention will be described together while describing specific details for carrying out the invention below.

1 is a system diagram schematically showing a power conversion system according to the present invention.
2 is a control block diagram showing a control configuration of the power conversion device shown in FIG.
3 is a flowchart illustrating a method of operating a power conversion system according to the present invention.

It should be noted that, in the following description, only parts necessary for understanding the embodiments of the present invention are described, and descriptions of other parts will be omitted so as not to obscure the gist of the present invention.

The terms or words used in the present specification and claims described below should not be construed as being limited to their ordinary or dictionary meanings, and the inventors have appropriate concepts of terms in order to best describe their inventions. It should be interpreted as meaning and concept consistent with the technical idea of the present invention based on the principle that it can be defined in Accordingly, the embodiments described in this specification and the configurations shown in the drawings are only preferred embodiments of the present invention, and do not represent all of the technical spirit of the present invention, so various equivalents that can be substituted for them at the time of the present application It should be understood that there may be variations and variations.

Hereinafter, embodiments of the present invention will be described in more detail with reference to the accompanying drawings.

1 is a system diagram schematically showing a power conversion system according to the present invention.

Referring to FIG. 1 , the power conversion system 100 may include a DC voltage source 110 , a power conversion device 120 , and a load 130 .

Here, the power converter 120 may convert the DC voltage output from the DC voltage source 110 to generate an AC voltage and supply it to the load 130 .

In this case, the power converter 120 may include an inverter for converting a DC voltage into an AC voltage or a converter for converting an AC voltage into a DC voltage, but is not limited thereto.

The power converter 120 may include a plurality of sensors for outputting sensor data by sensing a DC voltage, a voltage frequency, a temperature, and a humidity.

Here, the power conversion device 120 can self-diagnose internal failures and external failures of the power conversion device 120 by applying the sensor data to a set failure diagnosis model.

Thereafter, the power conversion device 120 may display fault information and additional fault information derived according to the fault information as an alarm when diagnosing a fault.

As described above, the power conversion system 100 can perform fault diagnosis in real time using sensor data, thereby having the advantage of securing reliability for fault diagnosis.

2 is a control block diagram showing a control configuration of the power conversion device shown in FIG.

Referring to FIG. 2 , the power conversion device 120 may include a Human Machine Interface (HMI) 210 , a power conversion unit 220 , a sensor unit 230 , and a control unit 240 .

The HMI 210 may input a command for controlling the operation of the power conversion device 120 to the controller 240 .

The HMI 210 may display result information corresponding to the measured sensor data on the sensor unit 230 , and may display failure information and additional failure information input from the control unit 240 .

The power converter 220 may convert the input DC voltage into an AC voltage, or convert the AC voltage into a DC voltage and output it.

For example, the power converter 220 may include at least one of an inverter and a converter. In general, an inverter may convert a DC voltage into an AC voltage by turning on and off a plurality of switch elements.

The sensor unit 230 may include a plurality of sensors, and may output sensor data s_data including each phase voltage and each phase current of DC voltage, DC current, and AC voltage to the control unit 240 .

The controller 240 may include a learning model generator 250 and a diagnosis determiner 260 .

First, the learning model generation unit 250 may include a model generation unit 252 , a model evaluation unit 254 , and a model verification unit 256 .

The model generation unit 252 includes a plurality of first failure data corresponding to a plurality of failures induced by an actual failure test for the power conversion unit 220 and a plurality of failures corresponding to the plurality of failures induced by a simulation experiment. It is possible to learn by applying the second failure data of , to a set machine learning algorithm, and generate a failure diagnosis model for failure diagnosis for the plurality of failures.

Here, the plurality of first failure data may be sensor data measured by the sensor unit 230 after varying the DC voltage supplied to the power conversion unit 220 for each set time.

In addition, the plurality of second failure data is simulated by a computer device, and may be measured data obtained by varying the DC voltage for each set time.

That is, the first and second failure data may be back data for generating a failure diagnosis model, and may include sensor data measured by the sensor unit 230 when the power conversion unit 220 is operated, and there is no limitation to this. do not leave

The machine learning algorithm may be at least one of Python, Numpy, Pandas, and Keras, but is not limited thereto.

The model generator 252 divides the plurality of first and second failure data into learning data, test data, and confirmation data for each failure, and applies the learning data to the machine learning algorithm to generate the failure diagnosis model. .

For example, the learning data, the test data, and the confirmation data may be described as follows.

Each of the first and second failure data may be data including each phase voltage and each phase current of a DC voltage, a DC current, and an AC voltage, and a measurement time.

Here, the model generator 252 may extract the learning data by increasing the first and second failure data by a first reference time in which a measurement time for each failure is set.

Also, the model generator 252 may extract test data that does not overlap with the training data by increasing the first and second failure data by a second reference time in which a measurement time for each failure is set.

Finally, the model generator 252 may increase the first and second failure data by a third reference time in which a measurement time for each failure is set to extract confirmation data that does not overlap with the learning data and the test data.

The model generator 252 may generate the failure diagnosis model by learning the learning data using a set machine learning algorithm.

The model evaluation unit 254 may apply the test data to the failure diagnosis model generated by the model generation unit 252 to calculate an evaluation hit for the failure diagnosis model.

That is, the model evaluation unit 254 may compare the plurality of failures with the corresponding failures diagnosed by applying the test data to the failure diagnosis model to calculate the evaluation hit for the predicted results for each failure.

The model check unit 256 may determine whether the evaluation hit level calculated by the model evaluation unit 254 is equal to or greater than a set reference hit level.

In this case, the model check unit 256 may confirm the normal prediction for the failure diagnosis of the power converter 220 by applying the confirmation data to the failure diagnosis model if the evaluation hit rating is equal to or greater than the reference hit rating.

Here, the model check unit 256 may control the model generation unit 252 to regenerate the failure diagnosis model when the evaluation hit rating is less than the reference hit rating.

The model generation unit 252 may regenerate the failure diagnosis model by re-classifying the plurality of first and second failure data into learning data, test data and confirmation data for each failure under the control of the model confirmation unit 256. have.

The diagnosis determination unit 260 may diagnose the failure of the power conversion unit 220 by applying sensor data input from a plurality of sensors to the failure diagnosis model when the power conversion unit 220 operates.

That is, the diagnosis determination unit 260 applies the sensor data to the failure diagnosis model to diagnose whether the power conversion unit 220 has a failure, and when the power conversion unit 220 diagnoses a failure of the power conversion unit 220 , Failure information and additional failure information derived according to the failure information may be generated.

Thereafter, the diagnosis determination unit 260 may transmit the failure information and the additional failure information to an external device (not shown) displayed and set on the HMI 210 .

The failure information may include a failure alarm and a failure occurrence time for informing the content of the failure, and the additional failure information may include an additional failure alarm for notifying the additional failure content generated by the failure alarm.

3 is a flowchart illustrating a method of operating a power conversion system according to the present invention.

Referring to FIG. 3 , the power conversion device 120 of the power conversion system 100 includes a plurality of first failure data corresponding to a plurality of failures induced by an actual failure experiment and the plurality of failure data induced by a simulation experiment. A plurality of second failure data corresponding to the failure may be collected (S310).

That is, the learning model generation unit 250 of the power conversion device 120 is a plurality of first failure data and simulation experiments corresponding to a plurality of failures induced by an actual failure experiment for the power conversion unit 220. A plurality of second failure data corresponding to the plurality of induced failures may be collected.

The power conversion device 120 divides the plurality of first and second failure data into learning data, test data and confirmation data for each failure (S320), and applies the learning data to the set machine learning algorithm to generate a failure diagnosis model. can be (S330).

That is, the learning model generating unit 250 is based on each phase voltage and each phase current of the DC voltage, DC current, and AC voltage included in each of the first and second failure data, and the measurement time, learning data, test data and confirmation data can be separated.

Here, the learning model generating unit 250 may extract the learning data by increasing the first and second failure data by a first reference time in which a measurement time for each failure is set.

Also, the learning model generating unit 250 may extract test data that does not overlap with the learning data by increasing the first and second failure data by a second reference time in which a measurement time for each failure is set.

Finally, the learning model generator 250 may increase the first and second failure data by a third reference time in which a measurement time for each failure is set, thereby extracting confirmation data that does not overlap with the learning data and the test data.

The power conversion device 120 may apply the test data to the failure diagnosis model to calculate an evaluation hit for the failure diagnosis model (S340).

That is, the learning model generating unit 250 may calculate the evaluation hit for the predicted results for each failure by comparing the plurality of failures with the corresponding failures diagnosed by applying the test data to the failure diagnosis model.

The power conversion device 120 determines whether the evaluation hit degree is equal to or greater than the set reference hit degree (350), and if the evaluation hit degree is greater than the reference hit degree, the confirmation data is applied to the failure diagnosis model to diagnose the failure of the power conversion unit 220 It is possible to confirm the normal prediction for the s360 (S360).

When the power conversion unit 220 is operated, the power conversion device 120 may apply sensor data input from a plurality of sensors to the failure diagnosis model to diagnose the failure of the power conversion unit 220 (S370).

That is, the diagnosis determination unit 260 of the power conversion device 120 applies the sensor data to the failure diagnosis model to diagnose whether the power conversion unit 220 has a failure, and when the power conversion unit 220 is diagnosed with a failure. Failure information of the power conversion unit 220 and additional failure information derived according to the failure information may be generated.

The power conversion device 120 may display the fault information and additional fault information generated during the fault diagnosis of the power conversion unit 220 on the HMI 210 (S380).

Here, the failure information may include a failure alarm and a failure occurrence time informing of the failure, and the additional failure information may include an additional failure alarm informing the additional failure information generated by the failure alarm.

After step (S350), if the evaluation hit rating is less than the set reference hit rating, the learning model generation unit 250 divides the plurality of first and second failure data into learning data, test data and confirmation data for each failure and re-classifies the failure. A diagnostic model may be regenerated (S390).

Features, structures, effects, etc. described in the above embodiments are included in at least one embodiment of the present invention, and are not necessarily limited to only one embodiment. Furthermore, features, structures, effects, etc. illustrated in each embodiment can be combined or modified for other embodiments by those of ordinary skill in the art to which the embodiments belong. Accordingly, the contents related to such combinations and modifications should be interpreted as being included in the scope of the present invention.

In addition, although the embodiment has been described above, it is only an example and does not limit the present invention, and those of ordinary skill in the art to which the present invention pertains are exemplified above in a range that does not depart from the essential characteristics of the present embodiment. It can be seen that various modifications and applications that have not been made are possible. For example, each component specifically shown in the embodiment may be implemented by modification. And differences related to such modifications and applications should be construed as being included in the scope of the present invention defined in the appended claims.

Claims (9)

Human Machine Interface (HMI);
a power converter for converting a DC voltage into an AC voltage or converting an AC voltage into a DC voltage; and
A control unit for controlling the power conversion unit according to the first command input to the HMI,
The control unit is
Machine learning in which a plurality of first failure data corresponding to a plurality of faults induced by an actual failure experiment for the power conversion unit and a plurality of second failure data corresponding to the plurality of failures induced by a simulation experiment are set a learning model generation unit that applies to an algorithm to learn, and generates a failure diagnosis model for failure diagnosis for the plurality of failures; and
Comprising a diagnostic determination unit for diagnosing a failure of the power conversion unit by applying sensor data input from a plurality of sensors to the failure diagnosis model when the power conversion unit is operated,
power conversion system. The method of claim 1,
The learning model generation unit,
a model generator for dividing the plurality of first and second failure data into learning data, test data, and confirmation data for each failure, and applying the learning data to the machine learning algorithm to generate the failure diagnosis model;
a model evaluation unit that applies the test data to the failure diagnosis model to calculate an evaluation hit for the failure diagnosis model; and
When the evaluation hit is greater than or equal to the set reference hit, including a model confirmation unit for applying the confirmation data to the failure diagnosis model to confirm a normal prediction for the failure diagnosis of the power conversion unit,
power conversion system. 3. The method of claim 2,
The machine learning algorithm is
At least one of Python, Numpy, Pandas and Keras,
power conversion system. 3. The method of claim 2,
The model evaluation unit,
Comparing the plurality of failures with the corresponding failures diagnosed by applying the test data to the failure diagnosis model, calculating the evaluation hit degree for each failure,
power conversion system. 3. The method of claim 2,
The model confirmation unit,
If the evaluation hit is less than the reference hit, controlling the model generator to regenerate the failure diagnosis model,
power conversion system. 6. The method of claim 5,
The model generation unit,
Re-generating the failure diagnosis model by re-classifying the plurality of first and second failure data into learning data, test data and confirmation data for each failure under the control of the model confirmation unit,
power conversion system. The method of claim 1,
The diagnostic determination unit,
By applying the sensor data to the failure diagnosis model to diagnose the failure of the power conversion unit, and to generate additional failure information derived according to the failure information and the failure information of the power conversion unit when diagnosing the failure of the power conversion unit,
power conversion system. 8. The method of claim 7,
The diagnostic determination unit,
Transmitting the failure information and the additional failure information to an external device displayed and set in the HMI,
power conversion system. 8. The method of claim 7,
The fault information is
Including the failure alarm and failure occurrence time to inform the details of the failure,
The additional fault information is
Including an additional failure alarm informing of additional failures occurring as the failure alarms,
power conversion system.

Images