Power quality early warning method and device for distributed power distribution network
By extracting the power supply and load waveform characteristics of the distributed distribution network, building a similarity queue, and outputting power quality warning messages, it solves the problem of abnormal power quality in the distribution network and is able to realize reliable early warning for small and medium-sized distribution networks , suitable for small and medium-sized distribution networks.
Patent Information
- Application Number
- CN202411180702.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-11-19
AI Technical Summary
Abnormal power quality in distributed distribution networks is not easy to be discovered and prevented in early stages. The existing artificial intelligence algorithms are not suitable for small distribution networks, and it is necessary to develop a power quality early warning method suitable for small and medium-sized distribution networks.
By obtaining multiple classification coefficients, standard frequency feature queues, power output waveforms and load waveforms, extracting the frequency characteristics of the waveform segment, building the first frequency feature queue, calculating the similarity queue, outputting the power quality warning message, and using the standard frequency feature queue early warning of the power quality characteristics.
It realizes reliable and reliable power quality warning in small and medium-sized distribution networks, reduces the demand for computing power, and is suitable for small and medium-sized distribution networks, and can find abnormalities that are not easy to detect by conventional methods.
Smart Images

Figure CN118983801A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of distributed distribution network operation monitoring, and in particular to a distributed distribution network power quality early warning method and device. Background Art
[0002] Distributed Generation is relative to the traditional centralized power supply mode, which means that the power generation system is arranged near the user in a small-scale, decentralized manner. Compared with centralized power supply stations, distributed power supply has the following advantages: very low transmission and distribution losses, remote control and monitoring of regional power quality and performance; very suitable for providing electricity to rural areas, pastoral areas, mountainous areas, developing areas, commercial areas and residential areas, greatly reducing environmental pressure.
[0003] While the distributed power distribution network brings the above advantages, it also brings complex power flow and power quality problems in the distribution network that are difficult to find and locate. For example, islanding problems, overvoltage problems, and harmonic pollution problems are all common problems in distributed power grids. Due to the complexity of distributed distribution networks, the above problems are difficult to achieve early detection and early prevention.
[0004] Some related technologies attempt to use artificial intelligence algorithms, such as artificial neural networks, to analyze the relationship between power grid quality and power flow from the power grid's power flow, thereby realizing the discovery and early warning of power grid quality problems. However, these algorithms usually require a large number of samples and strong computing power, and are not well suited for small distribution networks.
[0005] Based on this, it is necessary to develop a distributed distribution network power quality early warning method. Summary of the invention
[0006] The embodiments of the present invention provide a distributed distribution network power quality early warning method and device to solve the problem that the power quality anomaly of the distributed distribution network is not easy to be discovered in the prior art.
[0007] In a first aspect, an embodiment of the present invention provides a distributed distribution network power quality early warning method, comprising: Acquire multiple classification coefficients, multiple standard frequency characteristic queues, multiple power output waveforms, and multiple load waveforms; Extracting multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms for multiple times, extracting frequency features of the multiple waveform segments extracted each time, and constructing a first frequency feature queue according to the extracted features and the multiple classification coefficients to obtain multiple first frequency feature queues; Calculating the similarity between each standard frequency feature queue and the plurality of first frequency feature queues respectively, and constructing the obtained plurality of similarities into a similarity queue, thereby obtaining a plurality of similarity queues; A power quality warning message is outputted according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature.
[0008] In a possible implementation, the extracting multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms multiple times, extracting frequency features of the multiple waveform segments extracted each time, and constructing the extracted features into a first frequency feature queue includes: Obtain multiple classification coefficients, waveform segment lengths, and characteristic frequencies; Extracting the waveform of the waveform segment length from the positions corresponding to the first time nodes in the multiple power output waveforms and the multiple load waveforms, thereby obtaining multiple waveform segments; Extract frequency features from the multiple waveform segments according to the characteristic frequencies to obtain multiple first frequency features, select corresponding coefficients from the multiple classification coefficients according to the source waveforms and frequency orders of the first frequency features to perform multiplication processing on each first frequency feature, and use the multiple first frequency features after the multiplication processing as multiple second frequency features; Constructing the plurality of second frequency features into a first frequency feature queue according to the source waveform and frequency of the second frequency features; If the traversal extraction of the multiple power output waveforms and the multiple load waveforms is not completed, the position of the first time node is adjusted, and the step of extracting the waveform of the waveform segment length from the position corresponding to the first time node in the multiple power output waveforms and the multiple load waveforms is jumped to.
[0009] In a possible implementation, the extracting frequency features from the multiple waveform segments according to the characteristic frequencies to obtain multiple first frequency features, and selecting corresponding coefficients from the multiple classification coefficients to perform multiplication processing on each first frequency feature according to the source waveform and frequency of the first frequency feature, and using the multiple first frequency features after the multiplication processing as multiple second frequency features, includes: According to the first formula and the characteristic frequency, frequency features are extracted from the multiple waveform segments respectively to obtain multiple first frequency features, wherein the first formula is:
[0010] In the formula, For the The first waveform segment The first frequency feature, is the total number of data obtained after the waveform segment discretization processing, For the The first waveform segment is obtained after discretization data, is a natural constant, is an imaginary unit, is the characteristic frequency, is the total number of data obtained by discretizing the waveform segments corresponding to the characteristic frequency. is the ratio of pi; Selecting a target coefficient for each first frequency feature from the plurality of classification coefficients according to a source waveform and a frequency frequency of the first frequency feature; The product of each first frequency feature and the corresponding target coefficient is taken as the second frequency feature, thereby obtaining the multiple frequency features.
[0011] In a possible implementation, the construction process of the multiple classification coefficients and the multiple standard frequency feature queues includes: Acquire a plurality of undetermined coefficients and a plurality of first frequency characteristic queue samples, wherein each first frequency characteristic queue sample corresponds to a power quality state; Using the multiple undetermined coefficients to perform multiplication processing on multiple features of each first frequency feature queue sample, and using the first frequency feature queue sample after the multiplication processing as the second frequency feature queue sample, thereby obtaining multiple second frequency feature queue samples; Clustering the plurality of second frequency feature queue samples into a target number of sample classes, wherein the target number is the number of power quality state types; Determine the power quality state of each sample class, wherein the power quality state of the sample class is the power quality state of the majority of samples in the sample class; Filtering out a plurality of difference samples from the plurality of sample classes, wherein the difference samples are second frequency characteristic queue samples having a power quality state different from that of the sample class in which they are located; If the number of the plurality of difference samples is less than the number threshold, the plurality of difference samples are deleted from the plurality of sample classes, the class centers of the plurality of sample classes are used as the plurality of standard frequency feature queues, the power quality states of the plurality of sample classes are used as the power quality states of the plurality of standard frequency feature queues, and the plurality of undetermined coefficients are used as the plurality of classification coefficients; Otherwise, according to the class centers of the multiple sample classes and the distances between the multiple difference samples, the multiple unknown coefficients are adjusted, and the process jumps to the step of multiplying the multiple features of each first frequency feature queue sample using the multiple unknown coefficients, and using the multiplied first frequency feature queue sample as the second frequency feature queue sample, thereby obtaining multiple second frequency feature queue samples.
[0012] In a possible implementation manner, adjusting the multiple undetermined coefficients according to the class centers of the multiple sample classes and the distances between the multiple difference samples includes: Obtaining a target class center for each difference sample, wherein the target class center is a class center of a sample class having the same power quality state as the difference sample; Constructing a coefficient optimization equation that makes each difference sample approach the center of the target class and move away from the class center of the sample class; The multiple undetermined coefficients are adjusted according to the coefficient optimization equation.
[0013] In a possible implementation, the coefficient optimization equation is:
[0014] In the formula, For the The intermediate coefficients, is the difference sample Features, The first Features, is the class center of the sample class where the difference sample belongs Features, is the total number of difference sample features, After adjustment Undetermined coefficients, Before adjustment Undetermined coefficients.
[0015] In a possible implementation, the calculating the similarity between each standard frequency feature queue and the multiple first frequency feature queues respectively, and constructing the obtained multiple similarities into similarity queues, thereby obtaining multiple similarity queues, includes: Taking out a standard frequency characteristic queue from the plurality of standard frequency characteristic queues as a queue to be processed; According to the second formula, the similarities between the queue to be processed and the multiple first frequency feature queues are calculated respectively, so as to obtain multiple similarities, wherein the second formula is:
[0016] In the formula, For the The pending queue and The similarity of the first frequency feature queues, For the The first frequency feature queue Features, For the of the pending queue Features, is the total number of features in the first frequency feature queue; constructing the multiple similarities into a similarity queue corresponding to the queue to be processed; If the traversal of the plurality of standard frequency feature queues is not completed, the process jumps to the step of taking out a standard frequency feature queue from the plurality of standard frequency feature queues as a queue to be processed.
[0017] In a possible implementation, outputting a power quality warning message according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature includes: According to the multiple similarity queues and the first formula, multiple undetermined coefficients are determined, wherein the first formula is:
[0018] In the formula, For the The first in the similarity queue Similarity, is the number of similarities in the similarity queue, For the The sum of multiple similarities in a similarity queue, For the Undetermined coefficients; If there is an undetermined coefficient greater than the warning threshold among the multiple undetermined coefficients, the power quality state of the standard frequency characteristic queue with the largest value of the undetermined coefficient source is taken as the target power quality state, and a power quality warning message is output according to the target power quality state.
[0019] In a second aspect, an embodiment of the present invention provides a distributed distribution network power quality early warning device, comprising: A data acquisition module, used to acquire multiple classification coefficients, multiple standard frequency characteristic queues, multiple power output waveforms, and multiple load waveforms; a feature extraction module, configured to extract multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms for multiple times, extract frequency features of the multiple waveform segments extracted each time, and construct a first frequency feature queue according to the extracted features and the multiple classification coefficients to obtain multiple first frequency feature queues; A feature analysis module, used for calculating the similarity between each standard frequency feature queue and the plurality of first frequency feature queues, and constructing the obtained plurality of similarities into a similarity queue, thereby obtaining a plurality of similarity queues; as well as, The warning message output module is used to output a power quality warning message according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature.
[0020] In a possible implementation, the warning message output module includes: The undetermined coefficient determination module is used to determine a plurality of undetermined coefficients according to the plurality of similarity queues and a first formula, wherein the first formula is:
[0021] In the formula, For the The first in the similarity queue Similarity, is the number of similarities in the similarity queue, For the The sum of multiple similarities in a similarity queue, For the Undetermined coefficients; as well as, The early warning message module is used to take the power quality state of the standard frequency characteristic queue with the largest value of the undetermined coefficient as the target power quality state when there is an undetermined coefficient greater than the early warning threshold among the multiple undetermined coefficients, and output a power quality early warning message according to the target power quality state.
[0022] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows: The present invention provides a distributed distribution network power quality early warning method and device, firstly obtain multiple classification coefficients, multiple standard frequency feature queues, multiple power output waveforms and multiple load waveforms; then extract multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms for multiple times, extract frequency features for the multiple waveform segments extracted each time, and construct a first frequency feature queue according to the extracted features and the multiple classification coefficients to obtain multiple first frequency feature queues; then calculate the similarity between each standard frequency feature queue and the multiple first frequency feature queues, and construct the obtained multiple similarities into a similarity queue, thereby obtaining multiple similarity queues; finally, output a power quality early warning message according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature. The embodiment of the present invention extracts the output waveform and the load waveform of the distributed distribution network, extracts the feature queue of the waveform and compares the similarity with the standard frequency feature queue, and determines the early warning message. Since the standard frequency queue is classified and data processed. Therefore, the early warning message is more reliable and credible. In terms of standard frequency queue construction, the embodiments of the present invention do not require massive samples, nor do they require strong computing power when applied. The sample size required for construction is small, and it is suitable for small and medium-sized distribution networks. It does not require strong computing power when applied, and is easy to promote. In addition, since the standard frequency queue is constructed through samples, the samples have been verified by practical tests, and anomalies that are not easily discovered by conventional methods can be discovered through comparison. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0024] Figure 1 It is a flow chart of the implementation of the distributed distribution network power quality early warning method provided by the embodiment of the present invention; Figure 2 This is an application scenario diagram of the distributed distribution network power quality early warning method provided by an embodiment of the present invention; Figure 3 It is a schematic diagram of the first frequency characteristic queue construction process provided by an embodiment of the present invention; Figure 4 is a schematic diagram of a similarity queue construction process provided by an embodiment of the present invention; Figure 5 It is a schematic diagram of a process of multiple standard frequency characteristic queues provided by an embodiment of the present invention; Figure 6It is a structural schematic diagram of a distributed distribution network power quality early warning device provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0025] In the following description, specific details such as specific system structures and techniques are provided for the purpose of illustration rather than limitation so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary details.
[0026] In order to make the purpose, technical solutions and advantages of the present invention more clear, the following will be described through specific implementation methods in conjunction with the accompanying drawings.
[0027] Figure 1 The implementation flow chart of the distributed distribution network power quality early warning method provided by the embodiment of the present invention is detailed as follows: In step 101, a plurality of classification coefficients, a plurality of standard frequency characteristic queues, a plurality of power output waveforms, and a plurality of load waveforms are obtained.
[0028] In some embodiments, the construction process of the multiple classification coefficients and the multiple standard frequency feature queues includes: Acquire a plurality of undetermined coefficients and a plurality of first frequency characteristic queue samples, wherein each first frequency characteristic queue sample corresponds to a power quality state; Using the multiple undetermined coefficients to perform multiplication processing on multiple features of each first frequency feature queue sample, and using the first frequency feature queue sample after the multiplication processing as the second frequency feature queue sample, thereby obtaining multiple second frequency feature queue samples; Clustering the plurality of second frequency feature queue samples into a target number of sample classes, wherein the target number is the number of power quality state types; Determine the power quality state of each sample class, wherein the power quality state of the sample class is the power quality state of the majority of samples in the sample class; Filtering out a plurality of difference samples from the plurality of sample classes, wherein the difference samples are second frequency characteristic queue samples having a power quality state different from that of the sample class in which they are located; If the number of the plurality of difference samples is less than the number threshold, the plurality of difference samples are deleted from the plurality of sample classes, the class centers of the plurality of sample classes are used as the plurality of standard frequency feature queues, the power quality states of the plurality of sample classes are used as the power quality states of the plurality of standard frequency feature queues, and the plurality of undetermined coefficients are used as the plurality of classification coefficients; Otherwise, according to the class centers of the multiple sample classes and the distances between the multiple difference samples, the multiple unknown coefficients are adjusted, and the process jumps to the step of multiplying the multiple features of each first frequency feature queue sample using the multiple unknown coefficients, and using the multiplied first frequency feature queue sample as the second frequency feature queue sample, thereby obtaining multiple second frequency feature queue samples.
[0029] In some implementations, adjusting the plurality of undetermined coefficients according to the class centers of the plurality of sample classes and the distances between the plurality of difference samples comprises: Obtaining a target class center for each difference sample, wherein the target class center is a class center of a sample class having the same power quality state as the difference sample; Constructing a coefficient optimization equation that makes each difference sample approach the center of the target class and move away from the class center of the sample class; The multiple undetermined coefficients are adjusted according to the coefficient optimization equation.
[0030] In some embodiments, the coefficient optimization equation is:
[0031] In the formula, For the The intermediate coefficients, is the difference sample Features, The first Features, is the class center of the sample class where the difference sample belongs Features, is the total number of difference sample features, After adjustment Undetermined coefficients, Before adjustment Undetermined coefficients.
[0032] Step 102, extracting multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms multiple times, extracting frequency features of the multiple waveform segments extracted each time, and constructing a first frequency feature queue based on the extracted features and the multiple classification coefficients to obtain multiple first frequency feature queues.
[0033] In some embodiments, the extracting multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms multiple times, extracting frequency features of the multiple waveform segments extracted each time, and constructing the extracted features into a first frequency feature queue includes: Obtain multiple classification coefficients, waveform segment lengths, and characteristic frequencies; Extracting the waveform of the waveform segment length from the positions corresponding to the first time nodes in the multiple power output waveforms and the multiple load waveforms, thereby obtaining multiple waveform segments; Extract frequency features from the multiple waveform segments according to the characteristic frequencies to obtain multiple first frequency features, select corresponding coefficients from the multiple classification coefficients according to the source waveforms and frequency orders of the first frequency features to perform multiplication processing on each first frequency feature, and use the multiple first frequency features after the multiplication processing as multiple second frequency features; Constructing the plurality of second frequency features into a first frequency feature queue according to the source waveform and frequency of the second frequency features; If the traversal extraction of the multiple power output waveforms and the multiple load waveforms is not completed, the position of the first time node is adjusted, and the step of extracting the waveform of the waveform segment length from the position corresponding to the first time node in the multiple power output waveforms and the multiple load waveforms is jumped to.
[0034] In some embodiments, the extracting frequency features from the plurality of waveform segments according to the characteristic frequencies to obtain a plurality of first frequency features, and selecting a corresponding coefficient from the plurality of classification coefficients to perform a multiplication process on each first frequency feature according to the source waveform and the frequency of the first frequency feature, and using the plurality of first frequency features after the multiplication process as a plurality of second frequency features, comprises: According to the first formula and the characteristic frequency, frequency features are extracted from the multiple waveform segments respectively to obtain multiple first frequency features, wherein the first formula is:
[0035] In the formula, For the The first waveform segment The first frequency feature, is the total number of data obtained after the waveform segment discretization processing, For the The first waveform segment is obtained after discretization data, is a natural constant, is an imaginary unit, is the characteristic frequency, is the total number of data obtained by discretizing the waveform segments corresponding to the characteristic frequency. is the ratio of pi; Selecting a target coefficient for each first frequency feature from the plurality of classification coefficients according to a source waveform and a frequency frequency of the first frequency feature; The product of each first frequency feature and the corresponding target coefficient is taken as the second frequency feature, thereby obtaining the multiple frequency features.
[0036] For example, Figure 2 This is an application scenario diagram of the distributed distribution network power quality early warning method provided by the embodiment of the present invention. Figure 2 As shown, the figure shows an application scenario diagram of the distributed distribution network power quality early warning method provided by the embodiment of the present invention, the main power supply 201 and multiple distributed power supplies 204 respectively supply power to multiple loads 206 through the bus 202, wherein the output end of the main power supply 201 is directly connected to the bus 201, the distributed power supply 204 is connected to the bus 201 through the power feeder 203, and the load 206 is connected to the bus 201 through the load feeder 205. The power output waveform described in the embodiment of the present invention includes the voltage waveform and current waveform of the output end of the main power supply 201, and also includes the voltage waveform and current waveform output by the distributed power supply 204, and the load waveform includes: the input voltage waveform and input current waveform of the load 206.
[0037] The real-time method of the present invention is to cut the above-mentioned waveform according to the corresponding time period, and each time it is cut, the frequency of the waveform segment obtained by cutting is transformed, and the transformation result is constructed as a frequency feature queue. That is to say, the above-mentioned waveform can be obtained to construct multiple frequency feature queues, and these frequency feature queues are similarly calculated with the standard frequency feature queue, and the calculation results are constructed as a similarity queue. According to the quality characteristics and similarity independence of the standard frequency feature queue, the power quality warning message is output.
[0038] According to the above concept, the embodiment of the present invention first constructs a first frequency feature queue, such as Figure 3 As shown, this figure is a schematic diagram of the first frequency feature queue construction process provided by an embodiment of the present invention. Multiple waveforms 301 (including multiple power output waveforms and multiple load waveforms) are cut into multiple waveform segments 302 according to preset time lengths and time nodes. These waveform segments 302 are subjected to frequency feature extraction respectively to obtain multiple frequency features 303, wherein each waveform segment 302 corresponds to multiple frequency features 303, and then these frequency features 303 are connected end to end to construct a frequency feature queue 304.
[0039] In terms of frequency feature extraction, the present invention first applies the first formula to extract the first batch of frequency features. The first formula is:
[0040] In the formula, For the The first waveform segment The first frequency feature, is the total number of data obtained after the waveform segment discretization processing, For the The first waveform segment is obtained after discretization data, is a natural constant, is an imaginary unit, is the characteristic frequency, is the total number of data obtained by discretizing the waveform segments corresponding to the characteristic frequency. is the ratio of pi.
[0041] The extracted multiple first frequency features are multiplied by the corresponding classification coefficients respectively, and the obtained results are used as second frequency features. The frequency feature queue 304 is constructed using the second frequency features.
[0042] Here we have to talk about the construction process of multiple classification coefficients and multiple standard frequency characteristic queues. Multiple classification coefficients and multiple standard frequency characteristic queues are based on multiple first frequency characteristic queue samples. Through clustering, frequency characteristic queues with differences between the clustering results and the first frequency characteristic queue samples are found, and these frequency characteristic queues with differences are modified. Finally, multiple standard frequency characteristic queues with clustering results consistent with the power quality status of the standard frequency characteristic queues are obtained, and the undetermined coefficients used to adjust the multiple first frequency characteristic queue samples are constructed as standard frequency characteristic queues. The process of obtaining the first frequency characteristic queue samples here is consistent with the construction process of the above-mentioned first frequency characteristic queues, and both are obtained through waveform extraction. The only difference is that the first frequency characteristic queue samples have a certain power quality status, and the construction of the first frequency characteristic queue will not be repeated.
[0043] like Figure 5 As shown, after clustering multiple first frequency feature queue samples (clustering into classes with the same number of power quality status types), we obtain multiple sample classes 501, and the power quality status of the majority of samples 502 in each sample class 501 is taken as the power quality status of this class. At the same time, we can find that there are a few samples in these classes that are different from the power quality status of this class. We take these samples as difference samples 503. These difference samples 503 should be clustered to the target class to which they should belong, that is, the sample class with the same power quality status.
[0044] To achieve this goal, we use multiple undetermined coefficients to process these difference samples, so that these difference samples are far away from the current sample class and close to the target class. In order to determine a set of such multiple undetermined coefficients, we construct a coefficient optimization equation. The goal of this equation is to make each difference sample close to the target class and away from the original sample class. The coefficient optimization equation is:
[0045] In the formula, For the The intermediate coefficients, is the difference sample Features, The first Features, is the class center of the sample class where the difference sample belongs Features, is the total number of difference sample features, After adjustment Undetermined coefficients, Before adjustment Undetermined coefficients.
[0046] After the coefficient optimization is completed, each first frequency feature queue sample is coefficientized using multiple undetermined coefficients, and the processed first frequency feature queue samples are clustered again until the number of difference samples that appear after clustering is lower than the threshold. At this time, the multiple undetermined coefficients are used as multiple classification coefficients, and after eliminating the difference samples in the multiple sample classes, the class centers of the multiple sample classes are calculated as multiple standard frequency feature queues, and the power quality characteristics of the multiple sample classes are used as the power quality characteristics of the multiple standard frequency feature queues.
[0047] Step 103, calculating the similarity between each standard frequency feature queue and the multiple first frequency feature queues respectively, and constructing the obtained multiple similarities into a similarity queue, thereby obtaining multiple similarity queues.
[0048] In some implementations, the calculating the similarity between each standard frequency feature queue and the plurality of first frequency feature queues, and constructing the obtained plurality of similarities into a similarity queue, thereby obtaining a plurality of similarity queues, comprises: Taking out a standard frequency characteristic queue from the plurality of standard frequency characteristic queues as a queue to be processed; According to the second formula, the similarities between the queue to be processed and the multiple first frequency feature queues are calculated respectively, so as to obtain multiple similarities, wherein the second formula is:
[0049] In the formula, For the The pending queue and The similarity of the first frequency feature queues, For the The first frequency feature queue Features, For the of the pending queue Features, is the total number of features in the first frequency feature queue; constructing the multiple similarities into a similarity queue corresponding to the queue to be processed; If the traversal of the plurality of standard frequency feature queues is not completed, the process jumps to the step of taking out a standard frequency feature queue from the plurality of standard frequency feature queues as a queue to be processed.
[0050] For example, Figure 4 As shown, the figure shows a schematic diagram of the similarity queue construction process provided by an embodiment of the present invention. In terms of constructing a similarity queue, the embodiment of the present invention treats each standard frequency feature queue 401 as a queue to be processed, calculates the similarity 402 with each frequency feature queue 304 respectively, and constructs multiple similarities 402 into a similarity queue.
[0051] In one implementation, the similarity is determined using a second formula, which is:
[0052] In the formula, For the The pending queue and The similarity of the first frequency feature queues, For the The first frequency feature queue Features, For the of the pending queue Features, is the total number of features in the first frequency feature queue.
[0053] Step 104: output a power quality warning message according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature.
[0054] In some implementations, outputting a power quality warning message according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature includes: According to the multiple similarity queues and the first formula, multiple undetermined coefficients are determined, wherein the first formula is:
[0055] In the formula, For the The first in the similarity queue Similarity, is the number of similarities in the similarity queue, For the The sum of multiple similarities in a similarity queue, For the Undetermined coefficients; If there is an undetermined coefficient greater than the warning threshold among the multiple undetermined coefficients, the power quality state of the standard frequency characteristic queue with the largest value of the undetermined coefficient source is taken as the target power quality state, and a power quality warning message is output according to the target power quality state.
[0056] Exemplarily, after multiple similarity queues are constructed (each similarity queue corresponds to a standard frequency feature queue), a first formula is used to determine multiple undetermined coefficients. The first formula is:
[0057] In the formula, For the The first in the similarity queue Similarity, is the number of similarities in the similarity queue, For the The sum of multiple similarities in a similarity queue, For the Undetermined coefficients; Obviously, each undetermined coefficient corresponds to a standard frequency characteristic queue. If there is a coefficient greater than the warning threshold (for example, the warning threshold is 0.38) among the multiple undetermined coefficients, the largest undetermined coefficient is found from the multiple undetermined coefficients, the power quality state of the standard frequency characteristic queue from which the undetermined coefficient comes is used as the target power quality state, and a warning message is output according to this target power quality state.
[0058] For example, after calculation, among multiple undetermined coefficients, there is an undetermined coefficient that is greater than the warning threshold value of 0.38, then the undetermined coefficient with the largest value is found from the multiple undetermined coefficients. The power quality status of the standard frequency characteristic queue from which this undetermined coefficient comes indicates that the distributed distribution network is in a state of local overvoltage. At this time, a warning message of local overvoltage in the distributed distribution network is output.
[0059] The embodiment of the present invention first obtains multiple classification coefficients, multiple standard frequency feature queues, multiple power output waveforms and multiple load waveforms; then extracts multiple waveform segments corresponding to the same period from the multiple power output waveforms and the multiple load waveforms for multiple times, extracts frequency features for the multiple waveform segments extracted each time, and constructs a first frequency feature queue according to the extracted features and the multiple classification coefficients to obtain multiple first frequency feature queues; then calculates the similarity between each standard frequency feature queue and the multiple first frequency feature queues, and constructs the obtained multiple similarities into a similarity queue, thereby obtaining multiple similarity queues; finally, according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature, outputs a power quality warning message. The embodiment of the present invention extracts the output waveform and load waveform of the distributed distribution network, extracts the feature queue of the waveform and compares the similarity with the standard frequency feature queue, and determines the warning message. Since the standard frequency queue is classified and data processed. Therefore, the warning message is more reliable and credible. In terms of the construction of the standard frequency queue, the embodiment of the present invention does not require massive samples, and does not require strong computing power when applied. The sample size required for construction is small, which is suitable for small and medium-sized distribution networks. It does not require strong computing power when applied and is easy to promote. And because the standard frequency queue is constructed through samples, the samples have been verified through practical testing, and through comparison, anomalies that are not easy to find by conventional methods can be found.
[0060] It should be understood that the size of the serial numbers of the steps in the above implementation does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the implementation method of the present invention.
[0061] The following is an implementation of the device of the present invention. For details not described in detail, reference may be made to the corresponding method implementation described above.
[0062] Figure 6 The structure diagram of the distributed distribution network power quality early warning device provided by the embodiment of the present invention is shown. For the convenience of explanation, only the part related to the embodiment of the present invention is shown, which is described in detail as follows: like Figure 6 As shown, the distributed distribution network power quality early warning device 6 includes: a data acquisition module 601, a feature extraction module 602, a feature analysis module 603 and an early warning message output module 604, wherein: A data acquisition module 601 is used to acquire multiple classification coefficients, multiple standard frequency characteristic queues, multiple power output waveforms, and multiple load waveforms; A feature extraction module 602 is used to extract multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms for multiple times, extract frequency features of the multiple waveform segments extracted each time, and construct a first frequency feature queue according to the extracted features and the multiple classification coefficients to obtain multiple first frequency feature queues; The feature analysis module 603 is used to calculate the similarity between each standard frequency feature queue and the multiple first frequency feature queues, and construct the obtained multiple similarities into a similarity queue, thereby obtaining multiple similarity queues; The warning message output module 604 is used to output a power quality warning message according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature.
[0063] In a possible implementation, the warning message output module 604 includes: The undetermined coefficient determination module is used to determine a plurality of undetermined coefficients according to the plurality of similarity queues and a first formula, wherein the first formula is:
[0064] In the formula, For the The first in the similarity queue Similarity, is the number of similarities in the similarity queue, For the The sum of multiple similarities in a similarity queue, For the Undetermined coefficients; as well as, The early warning message module is used to take the power quality state of the standard frequency characteristic queue with the largest value of the undetermined coefficient as the target power quality state when there is an undetermined coefficient greater than the early warning threshold among the multiple undetermined coefficients, and output a power quality early warning message according to the target power quality state.
[0065] In the above-mentioned embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0066] Those skilled in the art will appreciate that the templates, units, and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.
[0067] If the module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned implementation method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of the above-mentioned distributed distribution network power quality early warning method implementation method. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory, random access memory, electrical carrier signal, telecommunication signal and software distribution medium.
[0068] The above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the protection scope of the present invention.
Claims
1. A distributed distribution network power quality early warning method, It is characterized in that include: Acquire multiple classification coefficients, multiple standard frequency characteristic queues, multiple power output waveforms, and multiple load waveforms; Extracting multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms for multiple times, extracting frequency features of the multiple waveform segments extracted each time, and constructing a first frequency feature queue according to the extracted features and the multiple classification coefficients to obtain multiple first frequency feature queues; Calculating the similarity between each standard frequency feature queue and the plurality of first frequency feature queues respectively, and constructing the obtained plurality of similarities into a similarity queue, thereby obtaining a plurality of similarity queues; A power quality warning message is outputted according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature.
2. The distributed distribution network power quality early warning method according to claim 1, It is characterized in that The method of extracting multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms for multiple times, extracting frequency features of the multiple waveform segments extracted each time, and constructing the extracted features into a first frequency feature queue includes: Obtain multiple classification coefficients, waveform segment lengths, and characteristic frequencies; Extracting the waveform of the waveform segment length from the positions corresponding to the first time nodes in the multiple power output waveforms and the multiple load waveforms, thereby obtaining multiple waveform segments; Extract frequency features from the multiple waveform segments according to the characteristic frequencies to obtain multiple first frequency features, select corresponding coefficients from the multiple classification coefficients according to the source waveforms and frequency orders of the first frequency features to perform multiplication processing on each first frequency feature, and use the multiple first frequency features after the multiplication processing as multiple second frequency features; Constructing the plurality of second frequency features into a first frequency feature queue according to the source waveform and frequency of the second frequency features; If the traversal extraction of the multiple power output waveforms and the multiple load waveforms is not completed, the position of the first time node is adjusted, and the step of extracting the waveform of the waveform segment length from the position corresponding to the first time node in the multiple power output waveforms and the multiple load waveforms is jumped to.
3. The distributed distribution network power quality early warning method according to claim 2, It is characterized in that The extracting frequency features from the plurality of waveform segments according to the characteristic frequencies to obtain a plurality of first frequency features, and selecting a corresponding coefficient from the plurality of classification coefficients to perform a multiplication process on each first frequency feature according to the source waveform and the frequency of the first frequency feature, and using the plurality of first frequency features after the multiplication process as a plurality of second frequency features, comprises: According to the first formula and the characteristic frequency, frequency features are extracted from the multiple waveform segments respectively to obtain multiple first frequency features, wherein the first formula is: In the formula, For the The first waveform segment The first frequency feature, is the total number of data obtained after the waveform segment discretization processing, For the The first waveform segment is obtained after discretization data, is a natural constant, is an imaginary unit, is the characteristic frequency, is the total number of data obtained by discretizing the waveform segments corresponding to the characteristic frequency. is pi; Selecting a target coefficient for each first frequency feature from the plurality of classification coefficients according to a source waveform and a frequency frequency of the first frequency feature; The product of each first frequency feature and the corresponding target coefficient is taken as the second frequency feature, thereby obtaining the multiple frequency features.
4. The distributed distribution network power quality early warning method according to claim 1, It is characterized in that The construction process of the multiple classification coefficients and the multiple standard frequency feature queues includes: Acquire a plurality of undetermined coefficients and a plurality of first frequency characteristic queue samples, wherein each first frequency characteristic queue sample corresponds to a power quality state; Using the multiple undetermined coefficients to perform multiplication processing on multiple features of each first frequency feature queue sample, and using the first frequency feature queue sample after the multiplication processing as the second frequency feature queue sample, thereby obtaining multiple second frequency feature queue samples; Clustering the plurality of second frequency feature queue samples into a target number of sample classes, wherein the target number is the number of power quality state types; Determine the power quality state of each sample class, wherein the power quality state of the sample class is the power quality state of the majority of samples in the sample class; Filtering out a plurality of difference samples from the plurality of sample classes, wherein the difference samples are second frequency characteristic queue samples having a power quality state different from that of the sample class in which they are located; If the number of the plurality of difference samples is less than the number threshold, the plurality of difference samples are deleted from the plurality of sample classes, the class centers of the plurality of sample classes are used as the plurality of standard frequency feature queues, the power quality states of the plurality of sample classes are used as the power quality states of the plurality of standard frequency feature queues, and the plurality of undetermined coefficients are used as the plurality of classification coefficients; Otherwise, according to the class centers of the multiple sample classes and the distances between the multiple difference samples, the multiple unknown coefficients are adjusted, and the process jumps to the step of multiplying the multiple features of each first frequency feature queue sample using the multiple unknown coefficients, and using the multiplied first frequency feature queue sample as the second frequency feature queue sample, thereby obtaining multiple second frequency feature queue samples.
5. The distributed distribution network power quality early warning method according to claim 4, It is characterized in that The adjusting the multiple undetermined coefficients according to the class centers of the multiple sample classes and the distances between the multiple difference samples comprises: Obtaining a target class center for each difference sample, wherein the target class center is a class center of a sample class having the same power quality state as the difference sample; Constructing a coefficient optimization equation that makes each difference sample approach the center of the target class and move away from the class center of the sample class; The multiple undetermined coefficients are adjusted according to the coefficient optimization equation.
6. The distributed distribution network power quality early warning method according to claim 5, It is characterized in that The coefficient optimization equation is: In the formula, For the The intermediate coefficients, is the difference sample Features, The first Features, is the class center of the sample class where the difference sample belongs Features, is the total number of difference sample features, After adjustment Undetermined coefficients, Before adjustment Undetermined coefficients.
7. The distributed distribution network power quality early warning method according to claim 1, It is characterized in that The calculating the similarity between each standard frequency feature queue and the plurality of first frequency feature queues respectively, and constructing the obtained plurality of similarities into a similarity queue, thereby obtaining a plurality of similarity queues, comprises: Taking out a standard frequency characteristic queue from the plurality of standard frequency characteristic queues as a queue to be processed; According to the second formula, the similarities between the queue to be processed and the multiple first frequency feature queues are calculated respectively, so as to obtain multiple similarities, wherein the second formula is: In the formula, For the The pending queue and The similarity of the first frequency feature queues, For the The first frequency feature queue Features, For the of the pending queue Features, is the total number of features in the first frequency feature queue; constructing the multiple similarities into a similarity queue corresponding to the queue to be processed; If the traversal of the plurality of standard frequency feature queues is not completed, the process jumps to the step of taking out a standard frequency feature queue from the plurality of standard frequency feature queues as a queue to be processed.
8. The distributed distribution network power quality early warning method according to any one of claims 1 to 7, It is characterized in that The outputting a power quality warning message according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature includes: According to the multiple similarity queues and the first formula, multiple undetermined coefficients are determined, wherein the first formula is: In the formula, For the The first in the similarity queue Similarity, is the number of similarities in the similarity queue, For the The sum of multiple similarities in a similarity queue, For the Undetermined coefficients; If there is an undetermined coefficient greater than the warning threshold among the multiple undetermined coefficients, the power quality state of the standard frequency characteristic queue with the largest value of the undetermined coefficient source is taken as the target power quality state, and a power quality warning message is output according to the target power quality state.
9. A distributed distribution network power quality early warning device, It is characterized in that Used to implement the distributed distribution network power quality early warning method according to any one of claims 1 to 8, the distributed distribution network power quality early warning device comprises: A data acquisition module, used to acquire multiple classification coefficients, multiple standard frequency characteristic queues, multiple power output waveforms, and multiple load waveforms; a feature extraction module, configured to extract multiple waveform segments corresponding to the same time period from the multiple power output waveforms and the multiple load waveforms for multiple times, extract frequency features of the multiple waveform segments extracted each time, and construct a first frequency feature queue according to the extracted features and the multiple classification coefficients to obtain multiple first frequency feature queues; A feature analysis module, used for calculating the similarity between each standard frequency feature queue and the plurality of first frequency feature queues, and constructing the obtained plurality of similarities into a similarity queue, thereby obtaining a plurality of similarity queues; as well as, The warning message output module is used to output a power quality warning message according to the multiple similarity queues and the power quality characteristics corresponding to each standard frequency feature.
10. The distributed distribution network power quality early warning device according to claim 9, It is characterized in that The warning message output module includes: The undetermined coefficient determination module is used to determine a plurality of undetermined coefficients according to the plurality of similarity queues and a first formula, wherein the first formula is: In the formula, For the The first in the similarity queue Similarity, is the number of similarities in the similarity queue, For the The sum of multiple similarities in a similarity queue, For the Undetermined coefficients; as well as, The early warning message module is used to take the power quality state of the standard frequency characteristic queue with the largest value of the undetermined coefficient as the target power quality state when there is an undetermined coefficient greater than the early warning threshold among the multiple undetermined coefficients, and output a power quality early warning message according to the target power quality state.
