Integrated intelligent power management system and harmonic control method

Abstract

The invention discloses an integrated intelligent power management system and a harmonic control method. The method comprises the following steps: acquiring analog signals of voltage and current; digital signal sequences of voltages and currents in the N power distribution nodes are generated; obtaining a K-order frequency domain harmonic component of each power distribution node; judging abnormal harmonic nodes; if the abnormal harmonic nodes exist in the N power distribution nodes, constructing a harmonic description matrix for abnormal harmonic source positioning; inputting the harmonic description matrix into a pre-trained abnormal harmonic source positioning model, and outputting an abnormal node identifier where an abnormal harmonic source is located; according to the method, the judgment of the abnormal harmonic nodes is converted from single-index passive alarm to multi-feature driven active identification, and the management level of a power system is improved.

Description

Technical Field

[0001] This invention relates to power management, specifically to an integrated intelligent power management system and a harmonic control method. Background Technology

[0002] In modern power distribution systems, with the widespread access of numerous nonlinear loads such as frequency converters, switching power supplies, and arc equipment, the problem of power grid harmonic pollution is becoming increasingly prominent. It is urgent to identify nodes that generate or significantly amplify harmonics in a timely manner to ensure power quality and equipment safety.

[0003] Patent document CN108110760B discloses a harmonic coordination control method for microgrid systems that considers both grid harmonics and load harmonics, enabling collaborative governance of grid harmonics. However, existing technologies have significant limitations in the identification of abnormal harmonic nodes: traditional methods generally use fixed thresholds (such as total harmonic distortion (THD) > 5%) to independently determine each node, ignoring the differences in the types of equipment connected to different nodes, operating conditions, and system impedance characteristics. This leads to high-impedance line ends being misjudged as abnormal sources due to natural voltage distortion, or small-capacity harmonic sources with propagation risks being missed because their harmonic content is below the threshold under light load. Furthermore, current judgment logic relies almost entirely on amplitude indicators (such as THD or the content of a certain harmonic), completely ignoring harmonic phase information—which is precisely the key basis for distinguishing between local harmonic injection and upstream harmonic propagation. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an integrated intelligent power management system and a harmonic control method, which solves the technical problems mentioned in the background by introducing anomaly node identifiers for abnormal harmonic sources.

[0006] To achieve the above objectives, the present invention provides the following technical solution: An integrated intelligent power management system and harmonic control method, comprising the following steps: S1. Simultaneously acquire the analog signals of voltage and current of N power distribution nodes at M sampling times on the time axis; S2. Perform analog-to-digital conversion and normalization on the analog signals at M sampling times to generate digital signal sequences of voltage and current in N distribution nodes; S3. Perform Fast Fourier Transform on the digital signal sequences of voltage and current of each distribution node to obtain the Kth frequency domain harmonic components of each distribution node. S4. Determine abnormal harmonic nodes based on the Kth frequency domain harmonic components; S5. If there are abnormal harmonic nodes among the N distribution nodes, then construct a harmonic description matrix for locating abnormal harmonic sources. S6. Input the harmonic description matrix into the pre-trained abnormal harmonic source localization model and output the abnormal node identifier where the abnormal harmonic source is located. S7. Determine the dynamic abnormal harmonic suppression strategy based on the abnormal node identifier.

[0007] In some embodiments, analog signals of voltage and current at N distribution nodes are simultaneously acquired on the time axis at M sampling times, including: S1-1. Anchor a sampling time on the time axis and configure a spatial sampling window at the sampling time; S1-2. Anchor N power distribution nodes within the target area in the spatial sampling window and simultaneously collect their respective analog voltage and current signals; S1-3. Slide the spatial sampling window M times at equal intervals with a preset sliding step size to simultaneously collect the analog signals of voltage and current of N power distribution nodes at M sampling times.

[0008] In some embodiments, digital signal sequences of voltage and current are generated for each of the N distribution nodes, including: S2-1, Anchor any target distribution node among N distribution nodes; S2-2. Obtain the analog signals of voltage and current of the target power distribution node at M sampling times; S2-3. Perform synchronous analog-to-digital conversion on the analog signals of voltage and current at M sampling times to obtain the original digital sample values ​​of voltage and current respectively; S2-4. Proportionally calculate the original digital sample values ​​of voltage and current to their rated signals to obtain normalized digital signals of each original digital sample value. S2-5. Arrange the M digital signals of voltage and current respectively in the order of sampling time to generate the digital signal sequences of voltage and current of the target power distribution node. S2-6. Traverse N distribution nodes until the digital signal sequences of voltage and current in each of the N distribution nodes are generated.

[0009] In some embodiments, the determination of anomalous harmonic nodes is based on the Kth frequency domain harmonic component, including: S4-1. Extract the complex frequency domain components corresponding to the 2nd to Kth harmonics from the Kth frequency domain harmonic components; S4-2. Calculate the amplitude of each complex frequency domain component, where the amplitude represents the signal strength of the harmonic. S4-3. Based on the amplitude of each complex frequency domain component, calculate the harmonic content of each harmonic component in the frequency domain. Total Harmonic Distortion ; S4-4. Compare the harmonic content of each frequency domain harmonic component and the total harmonic distortion rate with their set thresholds. S4-5. If the harmonic content or total harmonic distortion rate of any first-order harmonic component exceeds its set threshold, then it is determined that there is an abnormal harmonic node among the N nodes.

[0010] In some embodiments, based on N distribution nodes with anomalous harmonic nodes, a harmonic description matrix is ​​constructed containing the amplitude and phase of the 2nd to Kth harmonics of the N distribution nodes, including: S5-1. Obtain the node identifiers of N power distribution nodes and assign unique node serial numbers according to preset rules; S5-2. Construct a binary harmonic description sequence for the distribution node based on the unique node number; S5-3. Obtain the binary harmonic description sequence for each of the N node numbers; S5-4. Sort the binary harmonic description sequences of the N node numbers in ascending and descending order of the node numbers to generate an N-row × (K-1)-column harmonic description matrix; where each matrix element is represented as a binary harmonic descriptor.

[0011] In some embodiments, a binary harmonic description sequence is constructed for each distribution node based on a unique node number, including: S5-2-1. Anchor the Kth harmonic component corresponding to each node number, and extract the K-1 complex frequency domain components of the 2nd to Kth harmonic components. S5-2-2 Calculate the amplitude and phase of each of the K-1 complex frequency domain components; S5-2-3. Arrange the amplitudes and phases of the K-1 complex frequency domain components in order of harmonic order to form an amplitude sequence and a phase sequence with the same node number. S5-2-4. Align the amplitude and phase sequences of the same node number according to the harmonic order, and lock the amplitude and phase at the same sequence position; S5-2-5. Combine the amplitude and phase of the same sequence position to construct a binary harmonic descriptor that can characterize the frequency domain characteristics of the distribution node at a specific harmonic order. S5-2-6. Traverse the amplitude sequence and phase sequence of the same node number until a binary harmonic description sequence consisting of K-1 binary harmonic descriptors is obtained.

[0012] In some embodiments, the pre-training step of the abnormal harmonic source localization model includes: A1. Inject harmonic sources of known location and type into the simulation platform; A2. After injecting the harmonic source, obtain the analog signals of voltage and current of N power distribution nodes at M sampling times on the simulation platform; A3. Construct the harmonic description matrix by taking the analog signals of voltage and current of N power distribution nodes at M sampling times; A4. Using the harmonic description matrix as input and the identifier of the distribution node where the harmonic source is located as a label, construct several batches of training samples. A5. Input several batches of training samples into the neural network model for supervised training to obtain an abnormal harmonic source localization model that can predict the node identifier of the abnormal harmonic source based on the input harmonic description matrix.

[0013] In some embodiments, the supervised training step includes: B1. Select the first batch of harmonic description matrices from several batches of training samples and input them into the neural network model; B2. Extract features from the first batch of harmonic description matrices to generate node-level frequency domain feature vectors; B3. Propagate the node-level frequency domain feature vectors forward to generate node identifier prediction values; B4. Calculate the cross-entropy loss by comparing the predicted node identifier with the distribution node identifier used as a label; B5. If the cross-entropy loss is not minimized, calculate the gradient of the cross-entropy loss with respect to the current model parameters. B6. Update the model parameters with a predefined learning rate based on the gradient, and then use the updated model parameters to iteratively train the next batch of harmonic description matrices until the cross-entropy loss is minimized. B7. The neural network model that minimizes the cross-entropy loss is derived as the abnormal harmonic source localization model.

[0014] In some embodiments, a dynamic anomalous harmonic suppression strategy is determined based on the anomalous node identifier, including: S7-1. Obtain the device type, real-time operating status, and system impedance parameters corresponding to the abnormal node identifier, and characterize them as device tag, operating status tag, and impedance tag, respectively. S7-2. Use the device tag, operating status tag, and impedance tag as the query keys for the harmonic suppression strategy library, and match their corresponding device adaptation sub-strategy, status response sub-strategy, and impedance coordination sub-strategy respectively. S7-3. Combine the equipment adaptation sub-strategy, state response sub-strategy, and impedance coordination sub-strategy into a dynamic abnormal harmonic suppression strategy.

[0015] This invention provides a harmonic control method for an integrated intelligent power management system, which has the following advantages: This invention focuses on identifying abnormal harmonic nodes. By simultaneously acquiring voltage and current signals from multiple nodes and performing Fast Fourier Transform (FFT) on each, the complex frequency domain components of the 2nd to Kth harmonics are extracted. The amplitude and phase of each harmonic are then calculated, and the harmonic content and total harmonic distortion (THD) are derived based on the amplitude. Combined with preset thresholds, this significantly enhances the ability to detect real harmonic injection points. Furthermore, this identification process is executed in parallel on N distribution nodes in batches, achieving efficient screening of abnormal nodes. This transforms the identification of abnormal harmonic nodes from passive alarm based on a single indicator to proactive identification driven by multiple features, thereby improving the management level of the power system.

[0016] In a second aspect, the present invention provides an integrated intelligent power management system for executing the harmonic control method described in the first aspect, the system comprising: The analog signal acquisition module is used to synchronously acquire the analog signals of voltage and current of N power distribution nodes at M sampling times on the time axis; The signal sequence construction module is used to perform analog-to-digital conversion and normalization on the analog signals at M sampling times to generate digital signal sequences of voltage and current in N power distribution nodes. The harmonic component extraction module is used to perform fast Fourier transform on the digital signal sequences of voltage and current of each distribution node to obtain the Kth frequency domain harmonic components of each distribution node. The abnormal harmonic determination module is used to determine abnormal harmonic nodes based on the Kth frequency domain harmonic components. The harmonic matrix construction module is used to construct a harmonic description matrix for locating abnormal harmonic sources if there are abnormal harmonic nodes among N distribution nodes. The anomaly identification and localization module is used to input the harmonic description matrix into the pre-trained abnormal harmonic source localization model and output the anomaly node identification where the abnormal harmonic source is located. The suppression strategy determination module determines the dynamic abnormal harmonic suppression strategy based on the abnormal node identifier.

[0017] Compared with the prior art, the beneficial effects of the integrated intelligent power management system of the present invention are the same as those of the harmonic control method of the integrated intelligent power management system described above, so they will not be repeated here. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a harmonic control method for an integrated intelligent power management system according to the present invention. Figure 2 This is a schematic diagram of the process for determining abnormal harmonic nodes according to the present invention; Figure 3This is a schematic diagram illustrating the construction process of the binary harmonic description sequence of the present invention; Figure 4 A structural block diagram of an integrated intelligent power management system. Detailed Implementation

[0020] 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.

[0021] Example 1: Please refer to Figures 1 to 3 This invention provides a harmonic control method for an integrated intelligent power management system, comprising the following steps: S1. Simultaneously acquire the analog signals of voltage and current of N power distribution nodes at M sampling times on the time axis; S2. Perform analog-to-digital conversion and normalization on the analog signals at M sampling times to generate digital signal sequences of voltage and current in N distribution nodes; S3. Perform Fast Fourier Transform on the digital signal sequences of voltage and current of each distribution node to obtain the Kth frequency domain harmonic components of each distribution node. Where K is the preset maximum harmonic order for analysis, and the sampling parameters are configured to achieve a frequency resolution of 50Hz to ensure that each harmonic is accurately aligned with the FFT frequency point.

[0022] Specifically, the expression for the Fast Fourier Transform is: ; in, A digital signal sequence representing a voltage or current of length M. This represents the complex frequency domain component corresponding to the k-th frequency point. The imaginary unit; Specifically, the Fast Fourier Transform calculates the complex response at each frequency point and extracts the frequency points corresponding to the 2nd to Kth harmonics. The complex values ​​are the frequency domain components of the corresponding harmonics, which include amplitude and phase information.

[0023] S4. Determine abnormal harmonic nodes based on the Kth frequency domain harmonic components; S5. If there are abnormal harmonic nodes among the N distribution nodes, then construct a harmonic description matrix for locating abnormal harmonic sources. S6. Input the harmonic description matrix into the pre-trained abnormal harmonic source localization model and output the abnormal node identifier where the abnormal harmonic source is located. Specifically, a node identifier represents a uniquely identifiable number or address information of a power distribution node within the system.

[0024] S7. Determine the dynamic abnormal harmonic suppression strategy based on the abnormal node identifier.

[0025] In this embodiment, by synchronously acquiring voltage and current signals from multiple nodes, extracting harmonic frequency domain components through Fourier transform, and sequentially executing anomaly judgment, harmonic description matrix construction, model localization, and suppression strategy generation, a closed-loop control from harmonic perception to source control is achieved, avoiding the lag of traditional methods that rely on manual inspection or single threshold alarms.

[0026] Specifically, in this embodiment, step S1 includes: S1-1. Anchor a sampling time on the time axis and configure a spatial sampling window at the sampling time; S1-2. Anchor N power distribution nodes within the target area in the spatial sampling window and simultaneously collect their respective analog voltage and current signals; S1-3. Slide the spatial sampling window M times at equal intervals with a preset sliding step size to simultaneously collect the analog signals of voltage and current of N power distribution nodes at M sampling times.

[0027] In this embodiment, by configuring a sliding spatial sampling window on the time axis, M equally spaced synchronous samplings are performed on N power distribution nodes, ensuring strict alignment of the signals of each node in the time dimension.

[0028] Specifically, in this embodiment, step S2 includes: S2-1, Anchor any target distribution node among N distribution nodes; S2-2. Obtain the analog signals of voltage and current of the target power distribution node at M sampling times; S2-3. Perform synchronous analog-to-digital conversion on the analog signals of voltage and current at M sampling times to obtain the original digital sample values ​​of voltage and current respectively; S2-4. Proportionally calculate the original digital sample values ​​of voltage and current to their rated signals to obtain normalized digital signals of each original digital sample value. S2-5. Arrange the M digital signals of voltage and current respectively in the order of sampling time to generate the digital signal sequences of voltage and current of the target power distribution node. S2-6. Traverse N distribution nodes until the digital signal sequences of voltage and current in each of the N distribution nodes are generated.

[0029] In this embodiment, by synchronously converting the analog signals of each node to digital and normalizing them according to the rated value, the amplitude difference between devices with different ranges is eliminated, so that the generated digital signal sequence has a uniform numerical scale.

[0030] Specifically, in this embodiment, step S4 includes: S4-1. Extract the complex frequency domain components corresponding to the 2nd to Kth harmonics from the Kth frequency domain harmonic components; S4-2. Calculate the amplitude of each complex frequency domain component, where the amplitude represents the signal strength of the harmonic. For example, the amplitude is obtained by squaring the real and imaginary parts of the complex frequency domain components, adding them together, and then taking the square root of the result. The formula is: ; in, This represents the amplitude of the h-th harmonic. Represents the complex frequency domain component of the h-th harmonic. The real part, Represents the complex frequency domain component of the h-th harmonic. The imaginary part; S4-3. Based on the amplitude of each complex frequency domain component, calculate the harmonic content of each harmonic component in the frequency domain. Total Harmonic Distortion ; Specifically, the harmonic content represents the percentage of a certain harmonic amplitude to the fundamental frequency amplitude; the formula for calculating the harmonic content is: ; in, The fundamental amplitude.

[0031] Furthermore, the total harmonic distortion (THD) represents the degree of influence of all harmonic components on the fundamental frequency; the formula for calculating the THD is: ; S4-4. Compare the harmonic content of each frequency domain harmonic component and the total harmonic distortion rate with their set thresholds. S4-5. If the harmonic content or total harmonic distortion rate of any first-order harmonic component exceeds its set threshold, then it is determined that there is an abnormal harmonic node among the N nodes.

[0032] In this embodiment, by calculating the amplitude of each harmonic and deriving the harmonic content and total harmonic distortion rate, and then comparing them with a set threshold, abnormal nodes that exceed the power quality standard can be objectively identified, avoiding misjudgment or omission caused by relying on only a single indicator.

[0033] Specifically, in this embodiment, step S5 includes: S5-1. Obtain the node identifiers of N power distribution nodes and assign unique node serial numbers according to preset rules; For example, the preset rules can be sorted in ascending order according to the physical installation location number of the power distribution node address.

[0034] S5-2. Construct a binary harmonic description sequence for the distribution node based on the unique node number; S5-3. Obtain the binary harmonic description sequence for each of the N node numbers; S5-4. Sort the binary harmonic description sequences of the N node numbers in ascending and descending order of the node numbers to generate an N-row × (K-1)-column harmonic description matrix; where each matrix element is represented as a binary harmonic descriptor.

[0035] In this embodiment, by assigning a unique serial number to each distribution node and combining the amplitude and phase into a binary harmonic descriptor and arranging them in sequence into a matrix, the frequency domain characteristics of multiple nodes and multiple frequency points are expressed in a structured form, which is convenient as a standardized input for machine learning models.

[0036] Furthermore, step S5-2 also includes: S5-2-1. Anchor the Kth harmonic component corresponding to each node number, and extract the K-1 complex frequency domain components of the 2nd to Kth harmonic components. S5-2-2 Calculate the amplitude and phase of each of the K-1 complex frequency domain components; For example, the amplitude is calculated by taking the modulus of the complex number, i.e., calculating its distance from the origin in the complex plane; the phase is calculated by taking the argument of the complex number, i.e., determining its angle relative to the real axis in the complex plane. The complex plane represents a two-dimensional coordinate system with the real part as the horizontal axis and the imaginary part as the vertical axis, used to characterize the amplitude and phase information of the complex number.

[0037] S5-2-3. Arrange the amplitudes and phases of the K-1 complex frequency domain components in order of harmonic order to form an amplitude sequence and a phase sequence with the same node number. S5-2-4. Align the amplitude and phase sequences of the same node number according to the harmonic order, and lock the amplitude and phase at the same sequence position; S5-2-5. Combine the amplitude and phase of the same sequence position to construct a binary harmonic descriptor that can characterize the frequency domain characteristics of the distribution node at a specific harmonic order. S5-2-6. Traverse the amplitude sequence and phase sequence of the same node number until a binary harmonic description sequence consisting of K-1 binary harmonic descriptors is obtained.

[0038] In this embodiment, by separating the amplitude and phase from the complex frequency domain components and aligning them according to the harmonic order to form a binary descriptor, the frequency domain information of each harmonic is completely preserved.

[0039] In this embodiment, the pre-training step of the abnormal harmonic source localization model includes: A1. Inject harmonic sources of known location and type into the simulation platform; A2. After injecting the harmonic source, obtain the analog signals of voltage and current of N power distribution nodes at M sampling times on the simulation platform; A3. Construct the harmonic description matrix by taking the analog signals of voltage and current of N power distribution nodes at M sampling times; A4. Using the harmonic description matrix as input and the identifier of the distribution node where the harmonic source is located as a label, construct several batches of training samples. A5. Input several batches of training samples into the neural network model for supervised training to obtain an abnormal harmonic source localization model that can predict the node identifier of the abnormal harmonic source based on the input harmonic description matrix.

[0040] Specifically, the neural network model can be a multilayer perceptron (MLP) or a graph convolutional neural network (GCN).

[0041] In this embodiment, by injecting harmonic sources at known locations into the simulation platform and generating corresponding harmonic description matrices and label pairs, a training sample set with real physical meaning is constructed, so that the model learning process is consistent with the actual harmonic propagation characteristics of the power grid.

[0042] In this embodiment, the supervised training step includes: B1. Select the first batch of harmonic description matrices from several batches of training samples and input them into the neural network model; B2. Extract features from the first batch of harmonic description matrices to generate node-level frequency domain feature vectors; B3. Propagate the node-level frequency domain feature vectors forward to generate node identifier prediction values; B4. Calculate the cross-entropy loss by comparing the predicted node identifier with the distribution node identifier used as a label; B5. If the cross-entropy loss is not minimized, calculate the gradient of the cross-entropy loss with respect to the current model parameters. B6. Update the model parameters with a predefined learning rate based on the gradient, and then use the updated model parameters to iteratively train the next batch of harmonic description matrices until the cross-entropy loss is minimized. B7. The neural network model that minimizes the cross-entropy loss is derived as the abnormal harmonic source localization model.

[0043] In this embodiment, by using cross-entropy loss as the optimization objective, the neural network parameters are updated with gradients, so that the predicted node identifiers output by the model gradually approach the true labels, and finally a localization model that can stably identify the location of anomaly sources from the harmonic description matrix is ​​obtained.

[0044] Specifically, in this embodiment, step S7 includes: S7-1. Obtain the device type, real-time operating status, and system impedance parameters corresponding to the abnormal node identifier, and characterize them as device tag, operating status tag, and impedance tag, respectively. The featureization is achieved by mapping the original parameters to preset discrete categories or encoding vectors.

[0045] S7-2. Use the device tag, operating status tag, and impedance tag as the query keys for the harmonic suppression strategy library, and match their corresponding device adaptation sub-strategy, status response sub-strategy, and impedance coordination sub-strategy respectively. Specifically, the harmonic suppression strategy library represents a structured knowledge base that stores recommended suppression measures corresponding to different equipment types, operating states, and impedance conditions; while the specific matching method is based on predefined rules between tags and strategies.

[0046] For example, the device adaptation sub-strategy includes active filter switching, passive filter configuration, and PWM carrier frequency adjustment; the state response sub-strategy includes improving transient tolerance during startup and dynamically adjusting thresholds when the load changes; and the impedance coordination sub-strategy includes avoiding the connection of parallel capacitors at the resonant frequency and using damping resistors to suppress resonance.

[0047] S7-3. Combine the equipment adaptation sub-strategy, state response sub-strategy, and impedance coordination sub-strategy into a dynamic abnormal harmonic suppression strategy.

[0048] In this embodiment, by characterizing the device type, operating status, and system impedance of abnormal nodes as tags, and matching corresponding sub-strategies from the strategy library for combination, the suppression measures can be adapted to the current operating conditions, avoiding overcompensation or undercompensation problems caused by using fixed strategies.

[0049] Example 2: See Figure 4 The technical solution of Embodiment 2 differs from Embodiment 1 in that this embodiment of the invention also provides an integrated intelligent power management system. This system is used to implement the above-described method embodiments, and details already described will not be repeated. The terms "module," "unit," and "subunit" used below refer to combinations of software and / or hardware that achieve a predetermined function. Although the system described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0050] like Figure 4 As shown, Figure 4 This is a structural block diagram of an integrated intelligent power management system according to the present invention. The system includes: The analog signal acquisition module is used to synchronously acquire the analog signals of voltage and current of N power distribution nodes at M sampling times on the time axis; The signal sequence construction module is used to perform analog-to-digital conversion and normalization on the analog signals at M sampling times to generate digital signal sequences of voltage and current in N power distribution nodes. The harmonic component extraction module is used to perform fast Fourier transform on the digital signal sequences of voltage and current of each distribution node to obtain the Kth frequency domain harmonic components of each distribution node. The abnormal harmonic determination module is used to determine abnormal harmonic nodes based on the Kth frequency domain harmonic components. The harmonic matrix construction module is used to construct a harmonic description matrix for locating abnormal harmonic sources if there are abnormal harmonic nodes among N distribution nodes. The anomaly identification and localization module is used to input the harmonic description matrix into the pre-trained abnormal harmonic source localization model and output the anomaly node identification where the abnormal harmonic source is located. The suppression strategy determination module determines the dynamic abnormal harmonic suppression strategy based on the abnormal node identifier.

[0051] In the aforementioned device, an integrated intelligent power management system solves the problem that anomaly detection relies solely on amplitude.

[0052] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A harmonic control method for an integrated intelligent power management system, characterized in that, include: S1. Simultaneously acquire the analog signals of voltage and current of N power distribution nodes at M sampling times on the time axis; S2. Perform analog-to-digital conversion and normalization on the analog signals at M sampling times to generate digital signal sequences of voltage and current in N distribution nodes; S3. Perform Fast Fourier Transform on the digital signal sequences of voltage and current of each distribution node to obtain the Kth frequency domain harmonic components of each distribution node. S4. Determine abnormal harmonic nodes based on the Kth frequency domain harmonic components; S5. If there are abnormal harmonic nodes among the N distribution nodes, then construct a harmonic description matrix for locating abnormal harmonic sources. S6. Input the harmonic description matrix into the pre-trained abnormal harmonic source localization model and output the abnormal node identifier where the abnormal harmonic source is located. S7. Determine the dynamic abnormal harmonic suppression strategy based on the abnormal node identifier.

2. The integrated intelligent power management system and harmonic control method according to claim 1, characterized in that, Simultaneously acquire analog signals of voltage and current from N distribution nodes at M sampling times along the time axis, including: S1-1. Anchor a sampling time on the time axis and configure a spatial sampling window at the sampling time; S1-2. Anchor N power distribution nodes within the target area in the spatial sampling window and simultaneously collect their respective analog voltage and current signals; S1-3. Slide the spatial sampling window M times at equal intervals with a preset sliding step size to simultaneously collect the analog signals of voltage and current of N power distribution nodes at M sampling times.

3. The integrated intelligent power management system and harmonic control method according to claim 1, characterized in that, Generate digital signal sequences for voltage and current in N distribution nodes, including: S2-1, Anchor any target distribution node among N distribution nodes; S2-2. Obtain the analog signals of voltage and current of the target power distribution node at M sampling times; S2-3. Perform synchronous analog-to-digital conversion on the analog signals of voltage and current at M sampling times to obtain the original digital sample values ​​of voltage and current respectively; S2-4. Proportionally calculate the original digital sample values ​​of voltage and current to their rated signals to obtain normalized digital signals of each original digital sample value. S2-5. Arrange the M digital signals of voltage and current respectively in the order of sampling time to generate the digital signal sequences of voltage and current of the target power distribution node. S2-6. Traverse N distribution nodes until the digital signal sequences of voltage and current in each of the N distribution nodes are generated.

4. The integrated intelligent power management system and harmonic control method according to claim 2, characterized in that, Based on the Kth frequency domain harmonic components, the determination of anomalous harmonic nodes is performed, including: S4-1. Extract the complex frequency domain components corresponding to the 2nd to Kth harmonics from the Kth frequency domain harmonic components; S4-2. Calculate the amplitude of each complex frequency domain component, where the amplitude represents the signal strength of the harmonic. S4-3. Based on the amplitude of each complex frequency domain component, calculate the harmonic content of each harmonic component in the frequency domain. Total Harmonic Distortion ; S4-4. Compare the harmonic content of each frequency domain harmonic component and the total harmonic distortion rate with their set thresholds. S4-5. If the harmonic content or total harmonic distortion rate of any first-order harmonic component exceeds its set threshold, then it is determined that there is an abnormal harmonic node among the N nodes.

5. The integrated intelligent power management system and harmonic control method according to claim 1, characterized in that, Based on N distribution nodes with anomalous harmonic nodes, a harmonic description matrix is ​​constructed, containing the amplitude and phase of the 2nd to Kth harmonics of the N distribution nodes, including: S5-1. Obtain the node identifiers of N power distribution nodes and assign unique node serial numbers according to preset rules; S5-2. Construct a binary harmonic description sequence for the distribution node based on the unique node number; S5-3. Obtain the binary harmonic description sequence for each of the N node numbers; S5-4. Sort the binary harmonic description sequences of the N node numbers in ascending and descending order of the node numbers to generate an N-row × (K-1)-column harmonic description matrix; where each matrix element is represented as a binary harmonic descriptor.

6. The integrated intelligent power management system and harmonic control method according to claim 1, characterized in that, Based on the unique node number, a binary harmonic description sequence is constructed for each distribution node, including: S5-2-1. Anchor the Kth harmonic component corresponding to each node number, and extract the K-1 complex frequency domain components of the 2nd to Kth harmonic components. S5-2-2 Calculate the amplitude and phase of each of the K-1 complex frequency domain components; S5-2-3. Arrange the amplitudes and phases of the K-1 complex frequency domain components in order of harmonic order to form an amplitude sequence and a phase sequence with the same node number. S5-2-4. Align the amplitude and phase sequences of the same node number according to the harmonic order, and lock the amplitude and phase at the same sequence position; S5-2-5. Combine the amplitude and phase of the same sequence position to construct a binary harmonic descriptor that can characterize the frequency domain characteristics of the distribution node at a specific harmonic order. S5-2-6. Traverse the amplitude sequence and phase sequence of the same node number until a binary harmonic description sequence consisting of K-1 binary harmonic descriptors is obtained.

7. The integrated intelligent power management system and harmonic control method according to claim 1, characterized in that, The pre-training steps of the abnormal harmonic source localization model include: A1. Inject harmonic sources of known location and type into the simulation platform; A2. After injecting the harmonic source, obtain the analog signals of voltage and current of N power distribution nodes at M sampling times on the simulation platform; A3. Construct the harmonic description matrix by taking the analog signals of voltage and current of N power distribution nodes at M sampling times; A4. Using the harmonic description matrix as input and the identifier of the distribution node where the harmonic source is located as a label, construct several batches of training samples. A5. Input several batches of training samples into the neural network model for supervised training to obtain an abnormal harmonic source localization model that can predict the node identifier of the abnormal harmonic source based on the input harmonic description matrix.

8. The integrated intelligent power management system and harmonic control method according to claim 1, wherein the supervised training step includes: B1. Select the first batch of harmonic description matrices from several batches of training samples and input them into the neural network model; B2. Extract features from the first batch of harmonic description matrices to generate node-level frequency domain feature vectors; B3. Propagate the node-level frequency domain feature vectors forward to generate node identifier prediction values; B4. Calculate the cross-entropy loss by comparing the predicted node identifier with the distribution node identifier used as a label; B5. If the cross-entropy loss is not minimized, calculate the gradient of the cross-entropy loss with respect to the current model parameters. B6. Update the model parameters with a predefined learning rate based on the gradient, and then use the updated model parameters to iteratively train the next batch of harmonic description matrices until the cross-entropy loss is minimized. B7. The neural network model that minimizes the cross-entropy loss is derived as the abnormal harmonic source localization model.

9. The integrated intelligent power management system and harmonic control method according to claim 1, wherein a dynamic abnormal harmonic suppression strategy is determined based on the abnormal node identifier, comprising: S7-1. Obtain the device type, real-time operating status, and system impedance parameters corresponding to the abnormal node identifier, and characterize them as device tag, operating status tag, and impedance tag, respectively. S7-2. Use the device tag, operating status tag, and impedance tag as the query keys for the harmonic suppression strategy library, and match their corresponding device adaptation sub-strategy, status response sub-strategy, and impedance coordination sub-strategy respectively. S7-3. Combine the equipment adaptation sub-strategy, state response sub-strategy, and impedance coordination sub-strategy into a dynamic abnormal harmonic suppression strategy.

10. An integrated intelligent power management system for executing the harmonic control method according to any one of claims 1 to 9, characterized in that, include: The analog signal acquisition module is used to synchronously acquire the analog signals of voltage and current of N power distribution nodes at M sampling times on the time axis; The signal sequence construction module is used to perform analog-to-digital conversion and normalization on the analog signals at M sampling times to generate digital signal sequences of voltage and current in N power distribution nodes. The harmonic component extraction module is used to perform fast Fourier transform on the digital signal sequences of voltage and current of each distribution node to obtain the Kth frequency domain harmonic components of each distribution node. The abnormal harmonic determination module is used to determine abnormal harmonic nodes based on the Kth frequency domain harmonic components. The harmonic matrix construction module is used to construct a harmonic description matrix for locating abnormal harmonic sources if there are abnormal harmonic nodes among N distribution nodes. The anomaly identification and localization module is used to input the harmonic description matrix into the pre-trained abnormal harmonic source localization model and output the anomaly node identification where the abnormal harmonic source is located. The suppression strategy determination module is used to determine the dynamic abnormal harmonic suppression strategy based on the abnormal node identifier.

Images