A voltage sag recognition system for a closed loop power grid vessel

Abstract

The application relates to a voltage sag identification system for a closed-loop power grid ship, relates to the field of the closed-loop power grid ship, and comprises the following modules: a voltage spectrum extraction module, which is inputted with a time-domain voltage signal, transmits frequency-domain amplitudes to an identifier training module and a voltage judgment module; the identifier training module divides a waveform into a certain number of equidistant signals, facilitates input of a voltage sag identifier, and can completely cover an original signal; the voltage judgment module is inputted with single voltage time-domain waveform data, divides the data, puts each segment into the voltage spectrum extraction module, and finally puts the data into a trained voltage sag identifier to perform identification and judgment; and a cycle judgment module, which is inputted with voltage waveform data, cyclically inputs all voltage time-domain waveform data to the voltage judgment module, and sequentially identifies whether voltage sag occurs. The application provides a data-driven voltage sag method, and provides a train of thought and experience for actual operation and maintenance of the closed-loop power grid ship.

Description

Technical Field

[0001] This invention relates to the field of closed-loop power grid vessels, and more particularly to a voltage sag identification system for closed-loop power grid vessels. Background Technology

[0002] In recent years, with the increasing costs of offshore oil and gas development and increasingly stringent environmental emission requirements, the adoption of intelligent closed-loop power grid technology on drilling vessels has become an important technological direction, considering factors such as reducing fuel consumption costs, improving load anti-interference capabilities, reducing the probability of power interference, minimizing downtime due to unit maintenance, and improving power supply continuity. A closed-loop power grid allows the feed circuits of the propeller and its related systems to form a loop, improving the reliability of the dynamic positioning system. However, due to the special nature of the ring network structure, the actual power flow distribution is difficult to determine, fault currents are complex, and there are many large load devices, leading to frequent voltage sag events in the system, posing a significant challenge to voltage sag localization.

[0003] Closed-loop operating vessels are characterized by multiple power sources, multiple loops, standardized structures, and strong scalability. Due to the large number of devices, many of which are large motors, voltage sag events occur frequently in the system. Locating voltage sag events on vessels with closed-loop medium-voltage distribution networks powered by multiple power sources increases computational complexity due to the non-single power supply path. This patent proposes a method for locating voltage sags on vessels with closed-loop power grids. Different sources of sags exhibit distinct electrical characteristics. By utilizing the differences in voltage sag waveforms, the type of voltage sag source can be distinguished, thereby identifying which type of equipment caused the voltage sag. Summary of the Invention

[0004] The purpose of this invention is to provide a voltage sag identification system for ships with closed-loop power grids, and to propose a data-driven voltage sag method, providing ideas and experience for the operation and maintenance of ships with actual closed-loop power grids.

[0005] To achieve the above objectives, the present invention provides a voltage sag identification system for ships with closed-loop power grids, comprising:

[0006] The voltage spectrum extraction module takes a time-domain voltage signal as input, obtains the frequency domain amplitude and frequency domain phase angle through discrete Fourier transform, discards the frequency domain phase angle, and transmits the frequency domain amplitude to the recognizer training module and the voltage judgment module.

[0007] The recognizer training module trains the time-domain voltage signal at intervals of N. w Each cycle takes 2N. w The data from each cycle divides the waveform into a certain number of equidistant signals, which facilitates input to the sag detector and can completely cover the original signal;

[0008] The voltage judgment module takes a single voltage time-domain waveform as input and processes the data at intervals of N. w Each cycle takes 2N. w The waveform is segmented into individual cycles, and each segment is then fed into a voltage spectrum extraction module. Finally, the signal is fed into a trained voltage sag detector for identification and judgment. The output of this module is whether a voltage sag has occurred in the input voltage time-domain waveform.

[0009] The loop judgment module takes the voltage waveform data of all PMU measurement units as input, and cyclically inputs all voltage time-domain waveform data to the voltage judgment module to determine whether a voltage dip occurs at the PMU measurement node.

[0010] Preferably, let the time-domain voltage waveform be f(n), n = 0, 1, ..., N-1, then the discrete Fourier transform of the signal f(n) is:

[0011]

[0012] Frequency domain amplitude and frequency domain phase angle are

[0013] F mag (m)=mag(F(m)), m=0, 1,…,N-1 (2)

[0014] F angle (m)=angle(F(m)),m=0,1,…,N-1 (3)

[0015] Where mag(·) is the amplitude calculation function; angle(·) is the phase angle calculation function.

[0016] Preferably, for a time-domain voltage signal f(n), n = 0, 1, ..., N-1, the voltage spectrum extraction module only outputs the frequency-domain phase angle information F. mag (m), m=0, 1, ..., N-1.

[0017] Preferably, let the time-domain voltage waveform be f(n), n = 0, 1, ..., N-1, and the sampling time be Δt, then the segmented waveform is...

[0018]

[0019] Preferably, using equation (4), f(n), n = 0, 1, ..., N-1 can be divided into... An equidistant signal.

[0020] Preferably, N is generated according to the mathematical model formula of the IEEE standard power quality disturbance signal. g The sampling time is Δt and the length is 2N. w The training data for each cycle is then fed into the voltage spectrum extraction module to obtain the input for the voltage sag identifier.

[0021]

[0022] Preferably, the disturbance behavior of the above signal is marked using positive integer encoding to obtain the output of the voltage sag identifier.

[0023] Y = {flag i |i = 1, ..., N g} (6)

[0024] The flag encoding voltage disturbance behavior indicates that when flag=0, it represents normal voltage; when flag=1, it represents a voltage dip.

[0025] Preferably, the development steps of the voltage sag detector include:

[0026] Time-domain voltage data are generated according to the EEE standard, and X and Y are obtained according to equations (5) and (6);

[0027] Training was performed using XGB00ST;

[0028] If the accuracy requirements are met after training, the XGBOOST model is saved; otherwise, XGBOOST training continues.

[0029] Preferably, the voltage determination module includes the following steps:

[0030] Import the pre-trained voltage sag detector;

[0031] Determine whether the voltage of the current signal segment is disturbed;

[0032] If the judgment result is yes, output the judgment result; otherwise, advance N from the starting point. W One cycle, then take 2N. W The process takes one cycle length and then returns to the step of judging the current signal segment.

[0033] Preferably, the process steps of the loop judgment module include:

[0034] Import PMU data;

[0035] Take a set of data;

[0036] Take one phase of data;

[0037] The voltage judgment module described above is used for disturbance identification;

[0038] Determine whether the three-phase data identification is complete. If yes, proceed to the next step; otherwise, return to the step of retrieving one data item.

[0039] Determine if all nodes have been identified. If so, end the process; otherwise, return to the step of retrieving a set of data.

[0040] In summary, the present invention has the following beneficial technical effects:

[0041] Compared to existing technologies, this patent system describes a voltage sag positioning system for closed-loop shipboard power grids and proposes a data-driven voltage sag method, providing ideas and experience for the operation and maintenance of ships with practical closed-loop power grids. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating the development process of the voltage sag detector in this invention.

[0043] Figure 2 This is a flowchart of the voltage determination module in this invention;

[0044] Figure 3 This is a flowchart of the loop judgment module in this invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] This invention discloses a voltage sag identification system for ships in closed-loop power grids, which includes a voltage spectrum extraction module, an identifier training module, a voltage judgment module, and a loop judgment module.

[0047] I. Voltage Spectrum Extraction Module

[0048] The input to this module is a time-domain voltage signal. The frequency domain amplitude and frequency domain phase angle are obtained through discrete Fourier transform. The frequency domain phase angle is discarded, and the frequency domain amplitude is transmitted to the recognition training module and the voltage judgment module.

[0049] Let the time-domain voltage waveform be f(n), n = 0, 1, ..., N-1, then the discrete Fourier transform of the signal f(n) is:

[0050]

[0051] Frequency domain amplitude and frequency domain phase angle are

[0052] F mag (m)=mag(F(m)), m=0, 1,…,N-1 (2)

[0053] F angle(m)=angle(F(m)),m=0,1,…,N-1 (3)

[0054] Where mag(·) is the amplitude calculation function; angle(·) is the phase angle calculation function.

[0055] For a time-domain voltage signal f(n), n = 0, 1, ..., N-1, this module only outputs the frequency-domain phase angle information F. mag (m), m=0, 1, ..., N-1.

[0056] II. Recognizer Training Module

[0057] The inconsistent data lengths of PMU measurement units typically pose challenges to the design of the input interface for voltage sag detectors. Therefore, the time-domain voltage signal is divided into N intervals. w Each cycle takes 2N. w Data for one cycle. Let the time-domain voltage waveform be f(n), n = 0, 1, ..., N-1, and the sampling time be Δt. Then the segmented waveform is...

[0058]

[0059] Using equation (4), f(n), n = 0, 1, ..., N-1 can be divided into... The design of the equidistant signal facilitates the input of the voltage sag detector and can completely cover the original signal, ensuring that voltage sag detection is accurate and complete.

[0060] N is generated according to the mathematical model formula of the IEEE standard power quality disturbance signal. g The sampling time is Δt and the length is 2N. w The training data for each cycle is then fed into the voltage spectrum extraction module to obtain the input for the voltage sag identifier.

[0061]

[0062] The disturbance behavior of the above signals is then labeled using positive integer encoding to obtain the output of the voltage sag detector.

[0063] Y = {flag i |i = 1, ..., N g} (6)

[0064] The flag encoding voltage disturbance behavior indicates that when flag=0, it represents normal voltage; when flag=1, it represents a voltage dip.

[0065] The voltage sag detector can be obtained by fitting X and Y using the XGBOOST algorithm. The Python third-party library XGBOOST provides a simple and easy-to-use interface for the XGBOOST algorithm. The development process for the voltage sag detector is as follows: Figure 1 As shown.

[0066] III. Voltage Judgment Module

[0067] The input to this module is a single voltage time-domain waveform data, which is then processed at intervals of N. w Each cycle takes 2N. w The waveform is segmented into several cycles, and each segment is then fed into a voltage spectrum extraction module. Finally, the signal is fed into a trained voltage sag detector for identification and judgment. The output of this module is whether a voltage sag has occurred in the input voltage time-domain waveform. The process of this module is as follows: Figure 2 As shown.

[0068] IV. Loop Decision Module

[0069] This module takes voltage waveform data from all PMU measurement units as input and cyclically inputs all voltage time-domain waveform data to the voltage judgment module. The module then sequentially determines whether a voltage dip has occurred at each PMU measurement node. The module's process is as follows: Figure 3 As shown.

[0070] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A voltage sag identification system for ships with closed-loop power grids, characterized in that, include: The voltage spectrum extraction module takes a time-domain voltage signal as input, obtains the frequency domain amplitude and frequency domain phase angle through discrete Fourier transform, discards the frequency domain phase angle, and transmits the frequency domain amplitude to the recognizer training module and the voltage judgment module. The recognizer training module trains the time-domain voltage signal at intervals. Zhou Bo took The data from each cycle divides the waveform into a certain number of equidistant signals, which facilitates input to the sag detector and can completely cover the original signal; The voltage judgment module takes a single voltage time-domain waveform as input and processes the data at intervals. Zhou Bo took The waveform is divided into segments, and each segment is fed into the voltage spectrum extraction module. Finally, it is fed into the trained voltage sag detector for identification and judgment. The output of this module is whether the input voltage time domain waveform has experienced a voltage sag. as well as The loop judgment module takes the voltage waveform data of all PMU measurement units as input, and cyclically inputs all voltage time-domain waveform data to the voltage judgment module to determine whether a voltage dip occurs at the PMU measurement node in turn. In the recognition module, the time-domain voltage waveform is set as follows: Sampling time is If the voltage signal period is T, then the segmented waveform is: , then Divided into An equidistant signal.

2. The voltage sag identification system for ships with closed-loop power grids according to claim 1, characterized in that, Let the time-domain voltage waveform be Where n is the sampling point number and N is the total number of sampling points, then the signal The discrete Fourier transform is (1) Frequency domain amplitude and frequency domain phase angle are (2) (3) in, This is the amplitude calculation function; This is the phase angle calculation function.

3. A voltage sag identification system for ships with closed-loop power grids according to claim 2, characterized in that, For time-domain voltage signals The voltage spectrum extraction module only outputs frequency domain amplitude information. .

4. A voltage sag identification system for ships with closed-loop power grids according to claim 1, characterized in that, Generated according to the mathematical model formula of IEEE standard power quality disturbance signal The sampling time is , length is The training data for each cycle is then fed into the voltage spectrum extraction module to obtain the input for the voltage sag detector. (5)。 5. A voltage sag identification system for ships with closed-loop power grids according to claim 4, characterized in that, The disturbance behavior of the above signals is labeled using positive integer encoding to obtain the output of the voltage sag identifier. (6) in, Voltage disturbance behavior encoding, when When represents normal voltage; when The time indicates that a voltage dip has occurred.

6. A voltage sag identification system for ships with closed-loop power grids according to claim 5, characterized in that, The development steps for the voltage sag detector include: Time-domain voltage data is generated according to IEEE standards, and X and Y are obtained according to equations (5) and (6); Training was performed using XGBoost; If the accuracy requirements are met after training, the XGBOOST model is saved; otherwise, XGBOOST training continues.

7. A voltage sag identification system for ships with closed-loop power grids according to claim 1, characterized in that, The voltage determination module includes the following steps: Import the pre-trained voltage sag detector; Determine whether the voltage of the current signal segment is disturbed; If the judgment result is yes, output the judgment result; otherwise, move forward from the starting point. One cycle, then take 2 more. The process takes one cycle length and then returns to the step of judging the current signal segment.

8. A voltage sag identification system for ships with closed-loop power grids according to claim 1, characterized in that, The process steps of the loop judgment module include: Import PMU data; Take a set of data; Take one phase of data; The voltage judgment module described above is used for disturbance identification; Determine whether the three-phase data has been identified. If yes, proceed to the next step; otherwise, return to the step of retrieving one-phase data. Determine if all nodes have been identified. If so, end the process; otherwise, return to the step of retrieving a set of data.

Images