KR20230043370A - Server, method and computer program for managing power quality of data center - Google Patents

Abstract

The present invention relates to a server, method, and computer program for acquiring a plurality of voltage data from a plurality of voltage sensors installed near an electric power system, a power supply device, and a power distribution device that constitute a power system. The server that manages the power quality of a data center comprises: a voltage data acquisition unit that acquires a plurality of voltage data from a plurality of voltage sensors installed near an electric power system, a power supply device and a power distribution device that constitute a power system; a simulation unit that simulates power quality based on whether the harmonics occur, the degree of voltage fluctuation, and the degree of voltage drop; and a diagnostic unit that diagnoses whether an abnormality in the power quality occurs based on the results of the simulation, wherein the simulation unit comprises: a harmonic generation determination unit that determines whether harmonics are generated through a first power quality prediction engine based on the first voltage data measured at an input terminal of the power supply device among the plurality of voltage data; a voltage change detection unit that detects the degree of voltage change through a second power quality prediction engine based on the second voltage data measured at an output terminal of the power supply device; and a voltage drop detection unit that detects the degree of voltage drop through a third power quality prediction engine based on third voltage data measured at an output terminal of the power distribution device. The simulation unit functions as an environment of a deep reinforcement learning model, and the diagnosis unit functions as an agent of the deep reinforcement learning model.

Description

Server, method and computer program for managing power quality of data center {SERVER, METHOD AND COMPUTER PROGRAM FOR MANAGING POWER QUALITY OF DATA CENTER}

The present invention relates to a server, method and computer program for managing power quality in a data center.

A data center is a facility that gathers equipment necessary for providing IT services, such as servers, networks, and storage, in one building and integrates and manages them. A data center consists of servers, storage, and network devices for providing computing services, and generators, uninterruptible power supplies (UPS), thermo-hygrostats, backup systems, and security systems necessary to maintain these devices.

These data centers require large-scale power as a core infrastructure for storing and distributing big data, and in relation to a technology for supplying power to the data center, prior art Korea Patent Publication No. 2021-0076340 discloses a data center power supply system and A power supply method of a data center power supply system is disclosed.

In the case of a data center, it becomes difficult to provide stable and continuous IT services when an electric system power outage or an unexpected failure of network equipment occurs. Hereinafter, a cause of deterioration in power quality when power is supplied to a data center will be briefly described with reference to FIG. 1 .

1 is a diagram illustrating causes of deterioration in power quality occurring in a conventional power system. Referring to FIG. 1, a conventional power system 1 is composed of a power system 100 and a UPS system 110 including a power supply unit 111 (UPS) and a power distribution unit 112 (PDU).

In the conventional power system 1, harmonics (120 ) had the problem of increasing the capacity.

In addition, in the conventional power system 1, the power used by the power supply device 111 in three phases is changed by the characteristics of the CPU used in one phase (single phase), so that the voltage value for each single phase is different. It fluctuates so that the voltage fluctuation 130 is generated.

In addition, the conventional power system 1 has a problem that a voltage drop 140 is generated due to insulation and leakage current due to a decrease in equipment efficiency of the distribution board 113 in the UPS system 110, so that aging power equipment Retrofit of the distribution panel 113 to be replaced with a new facility was required.

When using such a conventional power system 1 to detect whether or not there is an abnormality in power quality, as the regulations of the power system 100 are applied to the UPS system 110 as it is, It was judged that the voltage fluctuation of 207V to 233V corresponds to the normal range. Due to this, as it becomes impossible to diagnose the safe state of the voltage values measured in real time in the power system 100 and the UPS system 110, it is possible to diagnose the occurrence of abnormal symptoms such as voltage fluctuations, voltage drops, power outages, and the like. I have a problem that I can't.

It is intended to provide a server, method, and computer program for acquiring a plurality of voltage data from a plurality of voltage sensors installed near a power system constituting a power system, a power supply device, and a power distribution device.

Based on the first voltage data measured at the input terminal of the power supply device among the plurality of voltage data, it is determined whether harmonics are generated through the first power quality prediction engine, and based on the second voltage data measured at the output terminal of the power supply device A server, method, and computer program for detecting a degree of voltage change through a second power quality prediction engine and detecting a degree of voltage drop through a third power quality prediction engine based on third voltage data measured at an output terminal of a power distribution device want to provide

It is intended to provide a server, method, and program for diagnosing whether an abnormality occurs in power quality based on simulation results.

However, the technical problem to be achieved by the present embodiment is not limited to the technical problems described above, and other technical problems may exist.

As a means for achieving the above-described technical problem, an embodiment of the present invention is a voltage for obtaining a plurality of voltage data from a plurality of voltage sensors installed near a power system constituting a power system, a power supply device, and a power distribution device. a data acquisition unit, a harmonic generation determination unit determining whether harmonics are generated through a first power quality prediction engine based on first voltage data measured at an input terminal of the power supply device among the plurality of voltage data, and a harmonic generation determination unit of the power supply device. A voltage fluctuation detection unit for detecting a degree of voltage fluctuation through a second power quality prediction engine based on the second voltage data measured at the output terminal and a third power quality prediction based on the third voltage data measured at the output terminal of the power distribution device A simulation unit including a voltage drop detection unit that detects a degree of voltage drop through an engine, and a simulation unit that simulates power quality based on whether the harmonics occur, the degree of voltage fluctuation, and the degree of voltage drop, and the power based on a result of the simulation A power quality comprising a diagnosis unit for diagnosing whether or not there is an abnormality in quality, wherein the simulation unit functions as an environment of the deep reinforcement learning model, and the diagnosis unit functions as an agent of the deep reinforcement learning model. A management server can be provided.

Another embodiment of the present invention is a step of acquiring a plurality of voltage data from a plurality of voltage sensors installed near the power system constituting the power system, the power supply device, and the power distribution device by the voltage data acquisition unit, by the simulation unit Determining whether harmonics are generated through a first power quality prediction engine based on first voltage data measured at an input terminal of the power supply device among the plurality of voltage data; Detecting a degree of voltage variation through a second power quality prediction engine based on the measured second voltage data, and third power quality based on the third voltage data measured at the output end of the power distribution device by the simulation unit. Detecting the degree of voltage drop through a prediction engine, simulating power quality based on whether the harmonics occur, the degree of voltage fluctuation, and the degree of voltage drop by the simulation unit, and determining the result of the simulation by the diagnosis unit. and diagnosing whether an abnormality occurs in the power quality based on the above, wherein the simulation unit functions as an environment of the deep reinforcement learning model, and the diagnosis unit functions as an agent of the deep reinforcement learning model. It is possible to provide a power quality management method.

In another embodiment of the present invention, when the computer program is executed by a computing device, a plurality of voltage sensors installed near a power system constituting a power system by a voltage data acquisition unit, a power supply device, and a power distribution device are stored. Obtaining voltage data, determining whether harmonics occur through a first power quality prediction engine based on first voltage data measured at an input terminal of the power supply device among the plurality of voltage data by a simulation unit, and supplying the power A degree of voltage fluctuation is detected through a second power quality prediction engine based on the second voltage data measured at the output terminal of the device, and a third power quality prediction engine based on the third voltage data measured at the output terminal of the power distribution device Detects the degree of voltage drop, simulates power quality based on whether the harmonics occur, the degree of voltage fluctuation, and the degree of voltage drop, and generates an abnormality in the power quality based on the result of the simulation by the diagnosis unit. A computer program stored in a medium containing a sequence of instructions for diagnosing whether or not, the simulation unit functions as an environment of the deep reinforcement learning model, and the diagnosis unit functions as an agent of the deep reinforcement learning model. can provide

The above-described means for solving the problems is only illustrative and should not be construed as limiting the present invention. In addition to the exemplary embodiments described above, there may be additional embodiments described in the drawings and detailed description.

According to any one of the above-described problem solving means of the present invention, voltage data related to power quality is collected in real time in a power system, and whether harmonics are generated, voltage change information, and voltage drop information using the voltage data collected in real time It is possible to provide a server, a method and a computer program for managing power quality for diagnosing power quality in real time based on the above.

A server, method, and computer program for predicting power quality according to load fluctuations of IT network equipment based on whether harmonics occur, voltage fluctuations, and voltage drop, and optimizing power quality based on voltage waves and harmonics according to load fluctuations can provide.

In order to manage the power quality of the data center in real time, based on the measurement and data analysis of voltage data, it is possible to detect in real time where the power load (IT network equipment) has an abnormality, thereby reducing downtime of the computer room due to electrical disasters. It is possible to provide a server, method, and computer program that can minimize (Downtime).

It is possible to provide a server, a method, and a computer program that can provide stable and continuous data center services by generating UPS system management indicators for electrical safety.

1 is a diagram illustrating causes of deterioration in power quality occurring in a conventional power system.
2 is a configuration diagram of a power system according to an embodiment of the present invention.
3 is a configuration diagram of a power quality management server according to an embodiment of the present invention.
4 is an exemplary diagram illustrating prediction data output through past data and a deep reinforcement learning model according to an embodiment of the present invention.
5 is an exemplary diagram for explaining a process of diagnosing whether an abnormality in power quality has occurred according to an embodiment of the present invention.
6 is a flowchart of a method for managing power quality of a data center in a power quality management server according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail so that those skilled in the art can easily practice the present invention with reference to the accompanying drawings. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case where it is "directly connected" but also the case where it is "electrically connected" with another element interposed therebetween. . In addition, when a part "includes" a certain component, this means that it may further include other components, not excluding other components, unless otherwise stated, and one or more other characteristics. However, it should be understood that it does not preclude the possibility of existence or addition of numbers, steps, operations, components, parts, or combinations thereof.

In this specification, a "unit" includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Further, one unit may be realized using two or more hardware, and two or more units may be realized by one hardware.

In this specification, some of the operations or functions described as being performed by a terminal or device may be performed instead by a server connected to the terminal or device. Likewise, some of the operations or functions described as being performed by the server may also be performed in a terminal or device connected to the corresponding server.

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

2 is a configuration diagram of a power system according to an embodiment of the present invention. Referring to FIG. 2 , the power system 2 may include a power system 200 , a UPS system 210 and a power quality management server 220 . Here, the UPS system 210 may include a power supply unit 211 and a power distribution unit 212 .

The power system 200, the UPS system 210, and the power quality management server 220 illustrate components that can be controlled by the power system 2 by way of example.

Each component of the power system 2 of FIG. 2 is generally connected through a network. For example, as shown in FIG. 2 , the power quality management server 220 may be connected to the power system 200 or the UPS system 210 simultaneously or at intervals.

A network refers to a connection structure capable of exchanging information between nodes such as terminals and servers, such as a local area network (LAN), a wide area network (WAN), and the Internet (WWW: World Wide Web), wired and wireless data communication network, telephone network, and wired and wireless television communication network. Examples of wireless data communication networks include 3G, 4G, 5G, 3rd Generation Partnership Project (3GPP), Long Term Evolution (LTE), World Interoperability for Microwave Access (WIMAX), Wi-Fi, Bluetooth communication, infrared communication, ultrasonic communication, visible light communication (VLC: Visible Light Communication), LiFi, and the like, but are not limited thereto.

The power system 200 may receive power and perform primary transformation 202 on the supplied power. For example, the power system 200 may transform a voltage of 154 kV or 22.9 Kv to 6.6 kV. Thereafter, the power system 200 may perform a secondary transformation 203 on power to be supplied to the UPS system 210 from the distribution panel. For example, the power system 200 may transform a voltage of 6.6 kV to 380 V.

The power system 200 may include a harmonic reduction device 201 .

The harmonic reduction device 201 is a filter composed of a series circuit of an inductor and a capacitor having a low impedance value for a related frequency to prevent a specific frequency from entering a facility or system in order to remove one or more harmonic currents and voltages. can be

When receiving the harmonic control signal from the power quality management server 220, the harmonic reduction device 201 may reduce the harmonic based on the harmonic control value included in the received harmonic control signal.

The UPS system 210 may include a power supply unit 211 (Uninterrupted Power Supply (UPS)) and a power distribution unit (212) (Power Distribution Unit (PDU)).

The power supply device 211 may refer to a device that supplies power from a battery for a predetermined period of time during a power outage, and then replenishes power to the battery in a standby (Stand-By) state when power is supplied.

The power supply device 211 may receive power from the power system 200 . Here, the voltage used in the power supply 211 may be 380V.

The power distribution device 212 may serve to provide local and remote power management for the IT rack.

The power distribution device 212 may receive power from the power supply device 211 . Here, the voltage used in the power distribution device 212 may be 220V.

The power quality management server 220 obtains a plurality of voltage data from a plurality of voltage sensors installed near the power system 200 constituting the power system 2, the power supply device 211, and the power distribution device 212. can For example, the power quality management server 220 may obtain a plurality of voltage data from a plurality of sensors installed near input/output terminals of the power system 200, the power supply device 211, and the power distribution device 212. .

The power quality management server 220 may determine whether harmonics occur through the first power quality estimation engine based on first voltage data measured at an input terminal of the power supply 211 among a plurality of voltage data. At this time, the power quality management server 220 may input the first voltage data to the first power quality prediction engine to derive a first management indicator value in which harmonic generation is predicted.

The power quality management server 220 may detect the degree of voltage variation through the second power quality estimation engine based on the second voltage data measured at the output terminal of the power supply device 211 . At this time, the power quality management server 220 may input the second voltage data to the second power quality prediction engine to derive a second management indicator value in which the degree of voltage variation is predicted.

The power quality management server 220 may detect the degree of voltage drop through the third power quality estimation engine based on the third voltage data measured at the output terminal of the power distribution device 212 . At this time, the power quality management server 220 may input the third voltage data to the third power quality prediction engine to derive a third management index value by which the degree of voltage drop is predicted.

The power quality management server 220 may simulate power quality based on whether harmonics occur, the degree of voltage fluctuation, and the degree of voltage drop.

The power quality management server 220 may diagnose whether an abnormality in power quality has occurred based on the result of the simulation.

Here, the power quality management server 220 includes the functions of an environment and an agent of a deep reinforcement learning model, and provides a first management index value, a second management index value, and a third management index value as an agent. State information is given, and the agent can take action to detect whether or not an abnormality in power quality has occurred based on the state information so as to satisfy a preset reward function.

The power quality management server 220 may control at least one of the voltage or harmonics of the power system 2 based on at least one of the first management index value, the second management index value, and the third management index value.

For example, the power quality management server 220 calculates a harmonic control value for the harmonic reduction device 201 included in the power system 200 based on the first management index value, and includes the calculated harmonic control value. A harmonic control signal to be transmitted to the harmonic reduction device 201.

For another example, the power quality management server 220 calculates a voltage control value for the power supply 211 based on the second management index value, and supplies a voltage control signal including the calculated voltage control value to power. device 211.

This power quality management server 220 is equipped with a power quality stabilization solution for operating the data center with optimal power quality, and based on digital transformation (DX) of power quality, it is possible to determine whether or not an abnormality in the power system 2 has occurred in real time. diagnosis can be performed.

3 is a configuration diagram of a power quality management server according to an embodiment of the present invention. Referring to FIG. 3 , the power quality management server 220 may include a voltage data acquisition unit 300 , a simulation unit 310 , a diagnosis unit 320 and a control unit 330 .

The voltage data acquisition unit 300 acquires a plurality of voltage data from a plurality of voltage sensors installed near the power system 200 constituting the power system 2, the power supply device 211, and the power distribution device 212. can For example, the voltage data acquisition unit 300 may obtain a plurality of voltage data from a plurality of sensors in units of distribution boards in order to remotely control all distribution boards of the power system 2 in the data center in real time.

The simulation unit 310 may include a harmonic generation determination unit 311 , a voltage fluctuation detection unit 312 and a voltage drop detection unit 313 .

The harmonic generation determination unit 311 may determine whether harmonics occur through the first power quality prediction engine based on first voltage data measured at an input terminal of the power supply 211 among a plurality of voltage data. To this end, the harmonic generation determination unit 311 may predict whether or not harmonics occur in real time in order to diagnose variability of harmonics generated in the UPS system 210 and introduced into the power system 200 .

The harmonic generation determining unit 311 may input the first voltage data to the first power quality prediction engine to derive a first management indicator value in which harmonic generation is predicted.

The voltage fluctuation detection unit 312 may detect the degree of voltage fluctuation through the second power quality estimation engine based on the second voltage data measured at the output terminal of the power supply 211 . To this end, the voltage fluctuation detection unit 312 measures in real time for all distribution boards in the UPS system 210 in order to manage the power quality of the input terminal (IN) / output terminal (OUT) of the power supply 211 used at 380V. Based on this, the degree of voltage fluctuation can be predicted based on 380V.

The voltage fluctuation detection unit 312 may input the second voltage data to the second power quality prediction engine to derive a second management index value at which the degree of voltage fluctuation is predicted.

The voltage drop detection unit 313 may detect the degree of voltage drop through the third power quality estimation engine based on the third voltage data measured at the output terminal of the power distribution device 212 . To this end, the voltage fluctuation detection unit 312 measures in real time for all distribution boards in the UPS system 210 to manage the power quality of the input terminal (IN) / output terminal (OUT) of the power distribution device 212 used as 220V. Based on this, the degree of voltage drop can be predicted based on 220V.

The voltage drop detection unit 313 may input the third voltage data to the third power quality prediction engine to derive a third management index value by which the degree of voltage drop is predicted.

The simulation unit 310 may simulate power quality based on whether harmonics occur, the degree of voltage fluctuation, and the degree of voltage drop. The simulation unit 310 may function as an environment for a deep reinforcement learning model.

The diagnosis unit 320 may diagnose whether an abnormality in power quality has occurred based on a simulation result. A process of diagnosing whether an abnormality in power quality has occurred based on simulation results will be described in detail with reference to FIGS. 4 and 5 .

4 is an exemplary diagram illustrating prediction data output through past data and a deep reinforcement learning model according to an embodiment of the present invention. Referring to FIG. 4 , the simulation unit 310 may function as an environment 401 of the deep reinforcement learning model 400, and the diagnosis unit 320 may function as an agent 402 of the deep reinforcement learning model 400. there is.

Here, state 403 information including the first management metric value, the second management metric value, and the third management metric value is given to the agent 402, and the agent 402 satisfies the preset reward 404 function. Action 405 may be taken to detect whether an abnormality in power quality has occurred based on the state information. At this time, the agent 402 may directly learn the action probabilities through the deep reinforcement learning model 400 and search for a policy for what action to take. When the agent 402 learns the action probability, the stability of learning can be increased by using the value function 406 (Value).

The simulation unit 310 provides current power quality measurement data and past power quality, such as the transformer power factor, harmonics, IT equipment load factor, and primary/secondary voltages of the input/output terminals of the power supply unit 211 measured for power quality management. Diagnostic data can be used to simulate power quality.

The simulation unit 310 may use the deep reinforcement learning model 410 to simulate power quality based on whether harmonics occur, the degree of voltage fluctuation, and the degree of voltage drop.

Here, the deep reinforcement learning model 410 may include an Actor network and a Critic network. The Actor network determines the action when the agent 402 is given a state, and the Critic network can evaluate the value of the state. Through this process, the simulation unit 310 may output the π(s, a) value derived through the Actor network and the V(s) value derived through the Critic network as the simulation result 413 . Here, V(s) may mean a value function, and π(s, a) may mean a probability of taking a specific action in a certain state.

Through this process, the simulation unit 310 may output past data 420 and prediction data 421 using the current power quality measurement value 411 and the past power quality diagnosis value 412 . Here, the prediction data 421 may represent at least one of a first management index value in which harmonic generation is predicted, a second management index value in which voltage fluctuation degree is predicted, and a third management index value in which voltage drop degree is predicted. .

Thereafter, the simulation unit 310 refers to the difference between the real-time current power quality measurement value 411 and the past power quality diagnosis value 412, and the first to third management index values that are predicted results through deep reinforcement learning A condition evaluation can be performed on To this end, the simulation unit 310 performs state evaluation on load fluctuations and power quality fluctuation values that fluctuate in real time through deep reinforcement learning, so that the corrected real-time prediction data 421 is used as a basis for determining whether an anomaly has occurred and It can be set as a management target value of power quality. At this time, the deep reinforcement learning model 400 may evaluate the state of the predicted result by referring to the difference between the setting of the rule set and the past state.

5 is an exemplary diagram for explaining a process of diagnosing whether an abnormality in power quality has occurred according to an embodiment of the present invention. Referring to FIG. 5 , the diagnosis unit 320 may diagnose whether an abnormality 510 has occurred in power quality based on a simulation result. For example, a reference point for occurrence of abnormal power quality may be generated based on a criterion for determining an abnormality in power quality and a target value for power quality management. At this time, it is possible to diagnose whether an abnormality 510 has occurred in power quality using the measurement data 501 and the prediction data 500, not the rule set.

Conventionally, power quality was inspected once a year by using a rule set method, which was only a formal periodic inspection, but the present invention determines harmonics and voltage-related anomalies in real time, interlocking with the harmonic reduction device 201 As a result, it is possible to block facility burnout in advance and to predict major items of power quality degradation of the power supply device 211 .

Conventionally, as the regulations of the power system 100 are applied to the UPS system 110 as it is for diagnosis, it is difficult to diagnose whether or not an abnormality has occurred in real time for the voltage measured in the UPS system 110, and the power system 100 Although the fluctuation of 207V to 233V was considered a normal range according to the regulations on the use of dedicated electric facilities, the present invention uses predicted data to establish a power quality standard suitable for the UPS system 210 to diagnose whether or not an abnormality in power quality has occurred can do.

That is, the present invention diagnoses the occurrence of abnormalities in power quality by utilizing data predicted through the solution of the power quality management server 220 while collectively applying the regulations of the power system 100, which is a problem of the prior art, and further New indicators for control can be created. Through this, the present invention diagnoses the power supply unit 211 in the UPS system 210 and the distribution panel in which problems related to voltage and harmonics are expected to occur in real time, and uses the prediction data to diagnose power quality abnormalities. can create new indicators for the diagnosis of

Returning to FIG. 3 , the controller 330 may control at least one of the voltage or harmonics of the power system 2 based on at least one of the first management metric value, the second management metric value, and the third management metric value. there is. At this time, when constructing the control model, the control unit 330 calculates the optimal control predicted operation point value in order to prevent excessive control of the harmonic reduction device 201 and the power supply device 211 by applying a hunting prevention processor for each time. And, the harmonic reduction device 201 or the power control device 211 of the power system 2 can be finally controlled based on the calculated optimal control predicted operating point value. Here, the optimal control predicted operating point value may be calculated based on the first to third management index values and the controlled history.

For example, the control unit 330 calculates a harmonic control value for the harmonic reduction device 201 included in the power system 200 based on the first management index value, and harmonic control including the calculated harmonic control value. A signal may be transmitted to the harmonic reduction device 201 .

For another example, the controller 330 may calculate a voltage control value for the power supply 211 based on the second management index value. For example, the controller 330 calculates a voltage control value for the power supply 211 as 2.2V based on 220V for a voltage step-up of 1%, and based on 220V for a voltage step-up of 2%. The voltage control value for the power supply 211 is calculated as 4.4V, and for a voltage step-up of 3%, the voltage control value for the power supply 211 is calculated as 6.6V based on 220V, and 4 For a voltage step-up of %, the voltage control value for the power supply 211 relative to 220V is calculated as 8.8V, and for a voltage step-up of 5%, the voltage for the power supply 211 relative to 220V The control value can be calculated as 11V.

The controller 330 may transmit a voltage control signal including the calculated voltage control value to the power supply 211 .

The control unit 330 may perform real-time control within an expected range to respond to instantaneous IT equipment turn-off and load factor fluctuations. Here, if a power quality problem occurs and the corresponding state value is out of the predicted control range, even though power quality stabilization through control is required, control may not be performed. This is to minimize breakdowns and electrical hunting by excessively repeating ON/OFF operations of switching elements of power conversion devices due to temporary power quality fluctuations.

6 is a flowchart of a method for managing power quality of a data center in a power quality management server according to an embodiment of the present invention. The method for managing power quality of a data center in the power quality management server 220 shown in FIG. 6 includes steps processed time-sequentially by the power system 2 according to the embodiment shown in FIGS. 2 to 5 do. Therefore, even if the details are omitted below, they are also applied to the method for managing power quality of the data center in the power quality management server 220 according to the embodiments shown in FIGS. 2 to 5 .

In step S610, the power quality management server 220 is installed near the power system 200, the power supply unit 211, and the power distribution unit 212 constituting the power system 2 by the voltage data acquisition unit 300. A plurality of voltage data may be obtained from the plurality of voltage sensors 220 to 223 .

In step S620, the power quality management server 220 generates harmonics through the first power quality prediction engine based on the first voltage data measured at the input terminal of the power supply 211 among the plurality of voltage data by the simulation unit 310. occurrence can be determined.

In step S630, the power quality management server 220 detects the degree of voltage variation through the second power quality prediction engine based on the second voltage data measured at the output terminal of the power supply 211 by the simulation unit 310. can

In step S640, the power quality management server 220 detects the degree of voltage drop through a third power quality prediction engine based on the third voltage data measured at the output terminal of the power distribution device 212 by the simulation unit 310. can

In step S650, the power quality management server 220 may simulate power quality based on the occurrence of harmonics, the degree of voltage fluctuation, and the degree of voltage drop by means of the simulation unit 310. Here, the simulation unit 310 may function as an environment for a deep reinforcement learning model.

In step S660, the power quality management server 220 may diagnose whether an abnormality in power quality has occurred based on the result of the simulation by the diagnosis unit 320. Here, the diagnosis unit 320 may function as an agent of a deep reinforcement learning model.

In the foregoing description, steps S610 to S660 may be further divided into additional steps or combined into fewer steps, depending on an embodiment of the present invention. Also, some steps may be omitted if necessary, and the order of steps may be switched.

The method for managing power quality of a data center in a power quality management server described with reference to FIGS. 2 to 6 may be in the form of a computer program stored in a medium executed by a computer or a recording medium including instructions executable by a computer. can be implemented Also, the method for managing power quality of a data center in a power quality management server described with reference to FIGS. 2 to 6 may be implemented in the form of a computer program stored in a medium executed by a computer.

Computer readable media can be any available media that can be accessed by a computer and includes both volatile and nonvolatile media, removable and non-removable media. Also, computer readable media may include computer storage media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.

The above description of the present invention is for illustrative purposes, and those skilled in the art can understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, the embodiments described above should be understood as illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined form.

The scope of the present invention is indicated by the following claims rather than the detailed description above, and all changes or modifications derived from the meaning and scope of the claims and equivalent concepts should be construed as being included in the scope of the present invention. do.

200: power system
201: harmonic reduction device
210: UPS grid
211: power supply
212: power distribution unit
220: power quality management server
300: voltage data acquisition unit
310: simulation unit
311: harmonic generation determination unit
312: voltage fluctuation detection unit
313: voltage drop detection unit
320: diagnosis unit
330: control unit

Claims (17)

In a server managing power quality of a data center,
a voltage data obtaining unit acquiring a plurality of voltage data from a plurality of voltage sensors installed near a power system constituting the power system, a power supply device, and a power distribution device;
A harmonic generation determination unit for determining whether harmonics are generated through a first power quality prediction engine based on first voltage data measured at an input terminal of the power supply device among the plurality of voltage data, A voltage fluctuation detection unit detecting a degree of voltage fluctuation through a second power quality prediction engine based on second voltage data and a voltage through a third power quality prediction engine based on third voltage data measured at an output terminal of the power distribution device a simulation unit including a voltage drop detection unit that detects a degree of drop, and simulating power quality based on whether the harmonics are generated, the degree of voltage fluctuation, and the degree of voltage drop; and
A diagnosis unit for diagnosing whether or not an abnormality has occurred in the power quality based on the result of the simulation.
including,
The simulation unit functions as an environment for a deep reinforcement learning model,
The diagnosis unit functions as an agent of the deep reinforcement learning model, the power quality management server.
According to claim 1,
The power quality management server, wherein the harmonic generation determination unit inputs the first voltage data to the first power quality prediction engine to derive a first management index value in which the occurrence of the harmonics is predicted.
According to claim 2,
The power quality management server, wherein the voltage fluctuation detection unit inputs the second voltage data to the second power quality prediction engine to derive a second management index value by which the degree of voltage fluctuation is predicted.
According to claim 3,
The power quality management server, wherein the voltage drop detection unit inputs the third voltage data to the third power quality prediction engine to derive a third management index value by which the degree of voltage drop is predicted.
According to claim 4,
State information including the first management metric value, the second management metric value, and the third management metric value is given to the agent, and the agent satisfies a preset reward function. A power quality management server that takes action to detect whether or not an abnormality occurs in the power quality based on the information.
According to claim 5,
Further comprising a controller for controlling at least one of voltage or harmonics of the power system based on at least one of the first management index value, the second management index value, and the third management index value. server.
According to claim 6,
The control unit calculates a harmonic control value for a harmonic reduction device included in the power system based on the first management index value,
To transmit a harmonic control signal including the calculated harmonic control value to the harmonic reduction device, the power quality management server.
According to claim 6,
The control unit calculates a voltage control value for the power supply based on the second management index value,
To transmit a voltage control signal including the calculated voltage control value to the power supply device, the power quality management server.
A method for managing power quality of a data center in a power quality management server,
obtaining a plurality of voltage data from a plurality of voltage sensors installed near a power system constituting a power system, a power supply device, and a power distribution device by a voltage data obtaining unit;
determining whether harmonics are generated through a first power quality prediction engine based on first voltage data measured at an input terminal of the power supply device among the plurality of voltage data by a simulation unit;
detecting a degree of voltage variation through a second power quality prediction engine based on second voltage data measured at an output terminal of the power supply by the simulation unit;
detecting a degree of voltage drop through a third power quality prediction engine based on third voltage data measured at an output terminal of the power distribution device by the simulation unit;
simulating power quality based on whether the harmonics are generated, the degree of voltage fluctuation, and the degree of voltage drop by the simulation unit; and
Diagnosing, by a diagnostic unit, whether an abnormality occurs in the power quality based on a result of the simulation.
including,
The simulation unit functions as an environment for a deep reinforcement learning model,
Wherein the diagnosis unit functions as an agent of the deep reinforcement learning model.
According to claim 9,
The power quality management method of claim 1, further comprising deriving a first management index value by which the occurrence of harmonics is predicted by inputting the first voltage data to the first power quality prediction engine.
According to claim 10,
The power quality management method further comprising the step of inputting the second voltage data to the second power quality prediction engine to derive a second management index value by which the degree of voltage fluctuation is predicted.
According to claim 11,
The power quality management method of claim 1, further comprising deriving a third management index value by which the degree of voltage drop is predicted by inputting the third voltage data to the third power quality prediction engine.
According to claim 12,
State information including the first management metric value, the second management metric value, and the third management metric value is given to the agent, and the agent satisfies a preset reward function. The power quality management method of taking an action to detect whether or not an abnormality in the power quality has occurred based on the information.
According to claim 13,
Controlling at least one of the voltage or harmonics of the power system based on at least one of the first management metric value, the second management metric value, and the third management metric value method.
15. The method of claim 14,
The control step is
Calculating a harmonic control value for a harmonic reduction device included in the power system based on the first management index value; and
And transmitting a harmonic control signal including the calculated harmonic control value to the harmonic reduction device.
According to claim 15,
The control step is
calculating a voltage control value for the power supply based on the second management metric value; and
And transmitting a voltage control signal including the calculated voltage control value to the power supply device.
A computer program stored on a medium containing a sequence of instructions for managing power quality of a data center, comprising:
When the computer program is executed by a computing device,
Acquiring a plurality of voltage data from a plurality of voltage sensors installed near a power system constituting a power system, a power supply device, and a power distribution device by a voltage data acquisition unit;
Based on the first voltage data measured at the input terminal of the power supply device among the plurality of voltage data by the simulation unit, it is determined whether harmonics are generated through a first power quality prediction engine, and the measured at the output terminal of the power supply device The degree of voltage change is detected through a second power quality prediction engine based on the second voltage data, and the degree of voltage drop is determined through a third power quality prediction engine based on the third voltage data measured at the output terminal of the power distribution device. Detect and simulate power quality based on whether the harmonics occur, the degree of voltage fluctuation, and the degree of voltage drop,
Diagnosing whether or not an abnormality in the power quality has occurred based on a result of the simulation by a diagnosis unit;
The simulation unit functions as an environment for a deep reinforcement learning model,
The diagnostic unit includes a sequence of instructions to function as an agent of the deep reinforcement learning model, a computer program stored in a medium.

Images