Load identification method and device and storage medium

Abstract

The invention relates to a load identification method and device and a storage medium, and the method comprises the steps: obtaining the electrical parameters of a plurality of pieces of to-be-tested power equipment when a load is connected; inputting the electrical parameters into an integrated learning model, and performing identification processing based on the electrical parameters through at least two types of identification algorithm models included in the integrated learning model to obtain at least two types of corresponding initial identification results; and performing voting processing based on the at least two types of initial identification results, and generating a load identification result of the to-be-tested power equipment. According to the invention, the problem of low accuracy of load identification is solved.

Description

Technical Field

[0001] The present application relates to the field of load identification, and in particular to a load identification method, device and storage medium. Background Art

[0002] With the rise and development of smart grids and smart IoT, non-intrusive load monitoring technology has received widespread attention because it can provide power users with real-time information on the power consumption of electrical appliances. The effective application of non-intrusive load monitoring technology can significantly reduce energy consumption and improve energy efficiency. Non-intrusive load monitoring mainly collects, processes and analyzes the power circuit signals in the building to obtain the load usage inside the building, thereby effectively supervising whether dangerous electrical appliances are used inside the building on the information side.

[0003] Non-intrusive load monitoring has four main components: event detection, data processing, load decomposition, and load identification. Among them, load identification, as the core task of non-intrusive load monitoring, requires the algorithm to accurately identify the type of load based on human input or features extracted from electrical signals. In related technologies, load identification methods include malicious load detection technology based on load identification, and with the introduction of deep learning, the accuracy of load identification has been further improved. However, the load identification method has a relatively single way to determine the load type. When the load identification method based on deep learning fails, it cannot effectively determine the type of load, and misidentification will occur, thereby affecting the accuracy of load identification.

[0004] Currently, no effective solution has been proposed for the problem of low accuracy of load identification in related technologies. Summary of the invention

[0005] The embodiments of the present application provide a load identification method, device and storage medium to at least solve the problem of low accuracy of load identification in the related art.

[0006] In a first aspect, an embodiment of the present application provides a load identification method, the method comprising: Obtaining electrical parameters of the power equipment under test when the load is connected; Inputting the electrical parameters into an integrated learning model, and performing identification processing based on the electrical parameters through at least two types of identification algorithm models included in the integrated learning model to obtain corresponding at least two types of initial identification results; Voting is performed based on the at least two types of initial identification results to generate a load identification result for the power equipment to be tested.

[0007] In some embodiments, the at least two types of recognition algorithm models included in the integrated learning model are used to perform recognition processing based on the electrical parameters to obtain corresponding at least two types of initial recognition results, including: Based on the electrical parameters, power parameters of the power equipment to be tested are obtained; based on the first recognition algorithm model, the power parameters are clustered to obtain a first initial recognition result; Obtaining a preset dangerous device characteristic quantity; based on a second recognition algorithm model, comparing the electrical device characteristic quantity of the electric equipment with the dangerous device characteristic quantity to obtain a second initial recognition result; Based on the electrical parameters, an electric parameter trajectory diagram of the electric power equipment to be tested is generated; the electric parameter trajectory diagram is input into a third recognition algorithm model, and a third initial recognition result is output.

[0008] In some embodiments, obtaining the power parameter of the power device to be tested based on the electrical parameter includes: Based on the electrical parameters, determining a periodic voltage signal and a periodic current signal of the electrical equipment within a sampling period; Calculating an angle difference between the periodic voltage signal and the periodic current signal; The power parameter is calculated according to the angle difference.

[0009] In some embodiments, calculating the angle difference between the periodic voltage signal and the periodic current signal includes: Performing frequency domain transformation processing on the periodic voltage signal to obtain an initial voltage signal phase, and performing unwinding processing on the initial voltage signal phase to obtain a target voltage signal phase; Performing frequency domain change processing on the periodic current signal to obtain an initial current signal phase, and performing unwinding processing on the initial current signal phase to obtain a target current signal phase; The angle difference is calculated according to a phase difference between the target voltage signal phase and the target current signal phase.

[0010] In some embodiments, the power parameters include reactive power and distortion power; the clustering process is performed on the power parameters based on the first recognition algorithm model to obtain the first initial recognition result, including: Determining clustering features based on the reactive power and the distortion power; Based on the first recognition algorithm model, calculating the distance between the cluster feature and the initial cluster center; iteratively updating the initial cluster center to a new cluster center according to the distance until the distance between the cluster feature and the new cluster center converges; Based on the new cluster center, the first initial recognition result is obtained.

[0011] In some embodiments, the comparing the electrical equipment characteristic quantity of the electric equipment with the characteristic quantity of the dangerous equipment based on the second identification algorithm model to obtain the second initial identification result includes: Based on the second identification algorithm model, calculating the similarity characteristics between the electrical equipment characteristic quantity of the electric equipment and the characteristic quantity of the dangerous equipment; The second initial recognition result is obtained according to the similarity feature.

[0012] In some embodiments, the step of inputting the power parameter trajectory diagram into a third recognition algorithm model and outputting a third initial recognition result includes: According to the electric power parameter trajectory diagram, generating a corresponding grayscale diagram; A pixel value matrix is ​​calculated based on the grayscale image, and the pixel value matrix is ​​input into the third recognition algorithm model, and the third initial recognition result is output.

[0013] In some embodiments, the performing of voting based on the at least two types of initial identification results to generate a load identification result for the power equipment to be tested includes: Performing voting based on the at least two types of initial identification results to obtain a first vote result indicating that the corresponding power equipment to be tested is a dangerous device, and a second vote result indicating that the power equipment to be tested is a non-dangerous device; The load identification result is generated based on the first voting result and the second voting result.

[0014] In a second aspect, an embodiment of the present application provides a load identification device, including: An acquisition module is used to obtain electrical parameters of the power equipment to be tested when the load is connected; An initial recognition module, used to input the electrical parameters into an integrated learning model, and perform recognition processing based on the electrical parameters through at least two types of recognition algorithm models included in the integrated learning model to obtain corresponding at least two types of initial recognition results; The voting module is used to perform voting processing based on the at least two types of initial identification results to generate a load identification result for the power equipment to be tested.

[0015] In a third aspect, an embodiment of the present application provides a storage medium on which a computer program is stored, and when the program is executed by a processor, the load identification method as described in the first aspect above is implemented.

[0016] Compared with the related art, the load identification method, device and storage medium provided in the embodiment of the present application obtain the electrical parameters of the power equipment to be tested when the load is connected; input the electrical parameters into the integrated learning model, and perform identification processing based on the electrical parameters through at least two types of identification algorithm models included in the integrated learning model to obtain corresponding at least two types of initial identification results; perform voting processing based on at least two types of initial identification results to generate load identification results for the power equipment to be tested. Based on this, it is possible to combine the output results of multiple different dimensions to determine the type of the measured load, avoiding the absoluteness of the output results of a single algorithm, effectively solving the problem of low accuracy of load identification caused by the phenomenon of misidentification when a single algorithm fails to identify the load in the related art, and also enhancing the robustness and adaptability of the algorithm, thereby realizing a non-intrusive load identification method based on integrated learning.

[0017] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 is a hardware structure block diagram of a terminal of a load identification method according to an embodiment of the present application; Figure 2 is a flow chart of a load identification method according to an embodiment of the present application; Figure 3 is a flow chart of another load identification method according to an embodiment of the present application; Figure 4 is a schematic diagram of a clustering result according to an embodiment of the present application; Figure 5 It is a structural block diagram of a load identification device according to an embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application. In addition, it can also be understood that although the efforts made in this development process may be complex and lengthy, for ordinary technicians in the field related to the contents disclosed in the present application, some changes such as design, manufacturing or production based on the technical contents disclosed in the present application are only conventional technical means, and should not be understood as insufficient contents disclosed in the present application.

[0020] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0021] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantity limitation, and may indicate the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application means greater than or equal to two. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. The terms "first", "second", "third" and the like involved in the present application are merely used to distinguish similar objects and do not represent a specific ordering of the objects.

[0022] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. Taking running on a terminal as an example, Figure 1 is a hardware structure block diagram of a terminal of a load identification method according to an embodiment of the present application. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is for illustration only and does not limit the structure of the above terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0023] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as a computer program corresponding to a load identification method in an embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0024] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0025] This embodiment provides a load identification method. Figure 2 is a flow chart of a load identification method according to an embodiment of the present application. Figure 2 As shown, the process includes the following steps: Step S210, obtaining electrical parameters of the power equipment to be tested when the load is connected.

[0026] The electrical parameters of the power equipment to be tested are collected when the load is stable. These electrical parameters include but are not limited to data such as voltage signals and current signals. The above electrical parameters can reflect the working state of the power equipment under the load.

[0027] Step S220, inputting the electrical parameters into the integrated learning model, and performing recognition processing based on the electrical parameters through at least two types of recognition algorithm models included in the integrated learning model to obtain corresponding at least two types of initial recognition results.

[0028] The electrical parameters collected through the above steps are input into the trained ensemble learning model. The ensemble learning model is a machine learning technology that combines the prediction results of multiple models (i.e., recognition algorithm models) to improve the accuracy of the overall prediction. In this step, the ensemble learning model uses at least two different recognition algorithm models (such as decision trees, clustering models, neural networks, etc.) to identify and process the input electrical parameters. Each recognition algorithm model independently analyzes and processes the input electrical parameters to generate corresponding initial recognition results.

[0029] Step S230: Perform voting based on at least two types of initial identification results to generate a load identification result for the power equipment to be tested.

[0030] Voting is performed on at least two types of initial identification results obtained above. Voting is a simple and effective integrated method, which counts the prediction results of multiple models and selects the prediction result with the most occurrences as the final load identification result, thereby determining whether the power equipment to be tested is a dangerous load equipment. This method can make full use of the advantages of different models and improve the accuracy and robustness of load identification.

[0031] In the above-mentioned load identification method, by collecting the electrical parameters of the power equipment, multiple identification algorithm models in the integrated learning model are used for identification processing, and the final load identification result is generated through voting processing. Therefore, the output results of multiple different dimensions can be combined to judge the type of the measured load, avoiding the absoluteness of the output results of a single algorithm, and effectively solving the problem of low load identification accuracy caused by misidentification when a single algorithm fails to identify the load in the related technology, and also enhancing the robustness and adaptability of the algorithm, thereby realizing a non-invasive load identification method based on integrated learning.

[0032] In some embodiments, the at least two types of recognition algorithm models included in the above-mentioned integrated learning model perform recognition processing based on electrical parameters to obtain corresponding at least two types of initial recognition results, including: Based on the electrical parameters, the power parameters of the power equipment to be tested are obtained; based on the first recognition algorithm model, the power parameters are clustered to obtain a first initial recognition result; the preset dangerous equipment characteristic quantity is obtained; based on the second recognition algorithm model, the electrical characteristic quantity of the power equipment is compared with the dangerous equipment characteristic quantity to obtain a second initial recognition result; based on the electrical parameters, a power parameter trajectory diagram of the power equipment to be tested is generated; the power parameter trajectory diagram is input into the third recognition algorithm model, and the third initial recognition result is output.

[0033] First, extract power parameters such as active power and reactive power from the electrical parameters of the power equipment to be tested. These parameters can reflect the energy conversion of the equipment under load. Use the first recognition algorithm model to cluster the power parameters. Clustering is an unsupervised learning method that can classify similar data points into the same category.

[0034] Through clustering processing, the power parameters of the power equipment to be tested can be matched with the known load type or equipment status, thereby obtaining a first initial recognition result, which can be a label of a certain load type.

[0035] Secondly, obtain the preset dangerous equipment feature quantity; the dangerous equipment feature quantity is based on historical data or expert experience and is used to identify the characteristics of equipment that may have safety hazards. These feature quantities may include specific voltage ranges, current fluctuation data, power factors, etc. For example, since the current waveforms of different electrical equipment vary greatly, a feature library can be established using the preset dangerous equipment current data, and then the dangerous equipment feature quantity can be obtained based on the constructed feature library. Then calculate the similarity between the electrical equipment to be tested and the dangerous equipment, Next, the electrical characteristic quantity of the electrical equipment to be tested is compared with the characteristic quantity of the dangerous equipment using the second identification algorithm model. By comparison, it can be determined whether the electrical equipment to be tested has characteristics similar to those of the dangerous equipment, thereby obtaining a second initial identification result. The result can be an indicator indicating whether the equipment is dangerous.

[0036] Again, based on the electrical parameters of the power equipment to be tested, an electric parameter trajectory diagram is generated. The trajectory diagram can show the changing trend of parameters such as voltage, current, and power of the equipment over a period of time. Based on the trajectory diagram, it is helpful to intuitively understand the operating status and load changes of the equipment. The electric parameter trajectory diagram is then input into the third recognition algorithm model for analysis. The third recognition algorithm model can be a method based on deep learning or machine learning, which can identify specific patterns or anomalies in the trajectory diagram. Through analysis, a third initial recognition result can be obtained, which can be a predictive identification indicating whether the equipment is a dangerous equipment.

[0037] It should be understood that the generation process of the first initial recognition result, the second initial recognition result and the third initial recognition result may be performed in parallel or sequentially.

[0038] Through the above embodiments, a method for determining the category of the measured load through three specific different dimensions, namely clustering recognition, similarity features and convolutional neural networks, is provided. By combining the clustering processing of power parameters, the comparison of electrical appliance characteristic quantities and the analysis of power parameter trajectory diagrams, the load status and potential risks of the tested power equipment can be more comprehensively and accurately identified, thereby helping to improve the accuracy and reliability of load identification and provide strong support for the operation and maintenance management of power equipment.

[0039] In some embodiments, the method of obtaining the power parameters of the power equipment to be tested based on the electrical parameters includes: Based on the electrical parameters, the periodic voltage signal and the periodic current signal of the power equipment within the sampling period are determined; the angle difference between the periodic voltage signal and the periodic current signal is calculated; and the power parameter is calculated according to the angle difference.

[0040] Among them, the electrical parameters of the power equipment are sampled to obtain the periodic voltage signal and the periodic current signal within the sampling period. It should be understood that in this step, it is also possible to check whether the sampled data contains abnormal values ​​(such as noise, spikes, missing values, etc.), and process these abnormal values, such as by filtering, interpolation or deletion. In addition, it is ensured that the sampling data of voltage and current are aligned in time, that is, each voltage sampling point corresponds to a current sampling point.

[0041] Perform Fourier transform on the voltage and current data to obtain the spectrum information of the signal and obtain the voltage signal and current signal. In this step, you can also check whether the reconstructed periodic signal waveform meets expectations, such as whether it presents a sine waveform and whether the phase relationship between voltage and current is reasonable. If the waveform does not meet expectations, you can adjust the filter parameters or re-perform Fourier transform analysis until a satisfactory periodic signal is obtained.

[0042] Specifically, based on the angle difference between the periodic voltage signal and the periodic current signal, various power parameters are calculated as shown in the following formula:

[0043]

[0044]

[0045]

[0046] In the above formula, P P Indicates active power, Represents the angle between the voltage signal and the current signal, P Q Represents reactive power; P S Represents apparent power; P d Indicates distortion power; V i Represents the i-th voltage value in a sampling period; I i represents the i-th current value in a sampling period; M represents the number of sampling points collected in a sampling period.

[0047] Through the above embodiments, the periodic voltage signal and the periodic current signal of the electric power equipment can be extracted from the sampled data, providing a basis for subsequent analysis and processing.

[0048] In some embodiments, the above-mentioned calculation of the angle difference between the periodic voltage signal and the periodic current signal includes: Perform frequency domain transformation processing on the periodic voltage signal to obtain an initial voltage signal phase, and perform unwinding processing on the initial voltage signal phase to obtain a target voltage signal phase; perform frequency domain transformation processing on the periodic current signal to obtain an initial current signal phase, and perform unwinding processing on the initial current signal phase to obtain a target current signal phase; calculate the angle difference based on the phase difference between the target voltage signal phase and the target current signal phase.

[0049] Specifically, for a periodic voltage signal within a sampling period, a fast Fourier transform is used to convert it into the frequency domain space to obtain a set of complex numbers with the same length as the current periodic voltage signal. The real and imaginary parts of the set of complex numbers are taken, and the square root and inverse tangent operations are performed in sequence to obtain the amplitude and phase of the voltage signal. Since a phase jump phenomenon, i.e., phase wrapping, may occur when the inverse tangent operation is used to extract the phase, it is also necessary to perform an unwrapping operation on the acquired initial voltage signal phase to obtain the unwrapped target voltage signal phase.

[0050] Similarly, for a periodic current signal within a sampling period, the fast Fourier transform is used to convert it into the frequency domain space to obtain a set of complex numbers with the same length as the current periodic current signal. Take the real and imaginary parts of this set of complex numbers, perform square and square root operations and inverse tangent operations in sequence to obtain the amplitude and phase of the current signal. Since phase jumps may occur when the inverse tangent operation is used to extract the phase, that is, phase wrapping, it is also necessary to unwrap the initial current signal phase to obtain the unwrap target current signal phase.

[0051] Based on the above operation, the index corresponding to the maximum value of the voltage signal amplitude is obtained, and the voltage signal phase value corresponding to the index is taken; the index corresponding to the maximum value of the current signal amplitude is obtained, and the current signal phase value corresponding to the index is taken. The difference between the two phase values ​​obtained in this step can obtain the angle value between the voltage signal and the current signal.

[0052] Through the above embodiments, the periodic voltage signal and the periodic current signal are subjected to frequency domain transformation processing, phase unwrapping and phase difference calculation, so that the phase information can be accurately extracted, the phase discontinuity problem is solved, and it helps to improve the signal processing efficiency, thereby improving the accuracy and efficiency of load identification.

[0053] In some embodiments, the power parameters include reactive power and distortion power; based on the first recognition algorithm model, clustering is performed on the power parameters to obtain a first initial recognition result, including: Based on reactive power and distortion power, clustering features are determined; based on the first recognition algorithm model, the distance between the clustering features and the initial clustering center is calculated; the initial clustering center is iteratively updated to a new clustering center according to the distance until the distance between the clustering features and the new clustering center converges; based on the new clustering center, a first initial recognition result is obtained.

[0054] Specifically, the active power, reactive power, apparent power and distortion power of the power equipment to be tested can be calculated by using the voltage and current signal data during steady-state operation of the load. Since the cluster intervals between different devices are the farthest when the clustering elements are reactive power and distortion power, in this embodiment, reactive power and distortion power are selected as clustering features so that the power characteristics and power quality in the power grid can be comprehensively considered, providing an effective data basis for subsequent clustering analysis.

[0055] In the cluster analysis process, the distance between the above cluster features and the initial cluster center is calculated; the initial cluster center is an initial value determined according to a certain strategy (such as random selection, based on data distribution, etc.). Using the first recognition algorithm model such as K-means clustering algorithm, density-based spatial clustering of applications with noise (DBSCAN), etc., the distance between each cluster feature (i.e., each data point) and the initial cluster center is calculated. According to the calculated distance, the initial cluster center is iteratively updated to a new cluster center according to a certain rule (such as minimum distance sum, maximum likelihood estimation, etc.). The iterative process continues until the distance between the cluster feature and the new cluster center converges (i.e., the change is less than a certain threshold) or the preset number of iterations is reached. When the cluster center converges, the data can be classified or identified based on the new cluster center to obtain the first initial recognition result; the first initial recognition result represents the cluster category or a certain specific pattern to which the data point belongs.

[0056] Through the above embodiments, an integrated learning model is provided, in which an identification method for classifying power equipment using a clustering algorithm is used, and whether the equipment is a dangerous equipment is determined based on the clustering results, which helps to timely discover and handle potential power equipment failures or abnormalities, and improve the stability and safety of the power system.

[0057] In some embodiments, the second initial recognition result is obtained by comparing the electrical equipment characteristic quantity of the electric equipment with the characteristic quantity of the dangerous equipment based on the second recognition algorithm model, including: Based on the second recognition algorithm model, the similarity characteristics between the electrical equipment characteristic quantity and the dangerous equipment characteristic quantity are calculated; and the second initial recognition result is obtained according to the similarity characteristics.

[0058] The electrical characteristic quantity data of each power device is obtained from the power system. These characteristic quantities may include reactive power, distorted power, voltage, current, temperature, etc. The characteristic quantities of each power device are sorted into feature vectors for similarity calculation. In addition, a database or feature library containing characteristic quantities of known dangerous devices is constructed. These characteristic quantities can be determined based on historical data or prior experience. More specifically, since the voltage waveforms of different power devices are less different and the current waveforms are more different, in this embodiment, the current data of pre-set dangerous devices can be used to establish a feature library, and the current characteristics of the power device to be tested are compared with the current characteristics in the feature library during feature comparison.

[0059] Using the preset second recognition algorithm model, calculate the similarity characteristics between the feature vector of the power equipment to be identified and the feature vector in the dangerous equipment feature library. This can be achieved by calculating the distance (such as Euclidean distance, Manhattan distance, etc.), similarity coefficient (such as cosine similarity, Pearson correlation coefficient, etc.) or the probability value output by the model.

[0060] For example, the Pearson correlation coefficient can be used to calculate the similarity between the measured equipment and the dangerous equipment. The calculation process is shown in the following formula:

[0061] In the above formula, X is used to represent the current signal of the power equipment to be tested; Y represents the current signal in the dangerous equipment feature library. N represents the number of features to be compared. Pearson is the calculated Pearson coefficient, which is the value of feature similarity.

[0062] Finally, based on the calculation results of the similarity characteristics, one or more thresholds are set to determine whether the electrical equipment to be tested is a dangerous device. If the similarity characteristic value of the electrical equipment to be tested is lower than the set threshold (for distance measurement) or higher than the set threshold (for similarity coefficient or probability value), it is considered that the electrical characteristic quantity of the electrical equipment to be tested is related to the characteristic quantity of the dangerous equipment, otherwise, it is considered unrelated. The number of features of the characteristic quantity of the dangerous equipment related to the electrical characteristic quantity of the electrical equipment to be tested is determined in the statistical library. If the number of features is greater than a certain preset number threshold, it is identified as a dangerous device; otherwise, it is identified as a non-dangerous device, and the above-mentioned second initial identification result is output.

[0063] Through the above embodiments, the second identification algorithm model can be used to calculate the similarity characteristics between the electrical characteristic quantities of the power equipment and the characteristic quantities of the dangerous equipment, and obtain the second initial identification result based on this, which helps to timely discover potential power equipment risks and improve the safety and reliability of the power system.

[0064] In some embodiments, the power parameter trajectory diagram is input into the third recognition algorithm model, and the third initial recognition result is output, including: According to the electric power parameter trajectory diagram, a corresponding grayscale image is generated; based on the grayscale image, a pixel value matrix is ​​calculated, and the pixel value matrix is ​​input into a third recognition algorithm model, and a third initial recognition result is output.

[0065] Among them, first, a trajectory diagram of the power parameters in the power equipment to be tested is obtained; for example, it can be a voltage and current trajectory drawn according to the electrical parameters of the power equipment to be tested, that is, a VI trajectory diagram. The power parameter trajectory diagram is converted into a grayscale image. This is because the grayscale image only contains brightness information, which simplifies the subsequent processing steps and reduces the amount of calculation. After that, the grayscale image is converted into a two-dimensional matrix (or pixel value matrix), in which each element represents the pixel value (i.e. brightness) of the corresponding position in the image. The pixel value matrix is ​​input into the third recognition algorithm model; the third recognition algorithm model can use a three-layer convolutional neural network to extract the features of the pixel value matrix, and use the activation function in the convolutional neural network to output the recognition result.

[0066] Through the above steps, the power parameter trajectory diagram can be effectively input into the third identification algorithm model, and a third initial identification result with practical significance can be output. These results can provide important references for monitoring, control and maintenance of the power system.

[0067] In some embodiments, the voting process is performed based on at least two types of initial identification results to generate a load identification result for the power equipment to be tested, including: Voting is performed based on at least two types of initial identification results to obtain a first voting result indicating that the corresponding power equipment under test is a dangerous device, and a second voting result indicating that the power equipment under test is a non-dangerous device; based on the first voting result and the second voting result, a load identification result is generated.

[0068] When dealing with the load identification task of the power equipment under test, if a voting strategy based on at least two types of initial identification results is adopted, we can follow the following steps to generate the final load identification result. This method combines the decisions of multiple classifiers and improves the accuracy and robustness of identification through a voting mechanism.

[0069] After the load of the power equipment to be tested is preliminarily identified by at least two different identification algorithms, voting is performed according to the initial identification results of each algorithm. Two voting categories are defined, namely "dangerous equipment" and "non-dangerous equipment". The initial identification results of each algorithm are counted, and the number of votes indicating that the power equipment to be tested belongs to "dangerous equipment" (the first voting result) and the number of votes indicating that it belongs to "non-dangerous equipment" (the second voting result) are recorded.

[0070] Finally, based on the results of the voting process, the final load identification result can be generated: If the first vote result (dangerous equipment) is greater than the second vote result (non-dangerous equipment), the final identification result is "dangerous equipment". If the second vote result (non-dangerous equipment) is greater than the first vote result (dangerous equipment), the final identification result is "non-dangerous equipment". If the votes for the two categories are equal (i.e., a tie), a variety of strategies can be used to resolve it, such as randomly selecting a category, weighted voting based on the confidence or reliability of the algorithm, or introducing additional decision-making mechanisms and determining the final identification result based on a pre-set tie handling strategy.

[0071] Through the above steps, voting can be performed based on at least two types of initial identification results to generate a load identification result for the power equipment to be tested. This method combines the advantages of multiple classifiers and improves the accuracy and robustness of identification.

[0072] The present application is described below through specific embodiments. Figure 3 is a flow chart of another load identification method according to an embodiment of the present application. Figure 3 As shown, the process includes the following steps: Step S301, obtaining the steady-state voltage and current of the power equipment within a sampling period.

[0073] Step S302: Calculate reactive power and distortion power according to the above steady-state voltage and current.

[0074] Step S303, using reactive power and distortion power as clustering features of the unsupervised K-means clustering algorithm, and training the model to obtain the cluster center, ie, the centroid, of each type of load equipment.

[0075] Step S304: construct a current signature library of dangerous equipment.

[0076] Step S305: Calculate the correlation between the power equipment to be tested and the dangerous equipment based on the Pearson coefficient.

[0077] Step S306, when the Pearson coefficient Pearson calculated by the above step S305 is greater than H, it is related; obtain the number number related to the current signal of the dangerous equipment in the feature library; when number>T, it means that the current power equipment to be tested is a dangerous equipment.

[0078] Step S307, obtaining the voltage and current trajectory of the power equipment.

[0079] Step S308: convert the voltage and current trajectories into grayscale images, and perform feature extraction using a convolutional neural network.

[0080] Step S309: output the currently measured load type based on the activation function.

[0081] Step S310, obtaining recognition results based on a voting mechanism.

[0082] In order to better understand the present application, the following also provides Figure 3 The method shown identifies the instance of the measured load: First, obtain a periodic voltage signal within a sampling period and a periodic current signal within a sampling period. The periodic voltage signal includes 128 data points: [0.13810656; 0.29494221; 0.43069145; 0.55645988; 0.68335739; 0.81311646; 0.94158699; 1.07263031; 1.20988358; 1.35415695; 1.50152039; 1.64806931; 1.79576447; 1.94038696…]; similarly, the periodic current signal includes 128 data points: [0.00017647; 0.00020396; 0.00024591; 0.0002271;0.00022999;0.00020974;0.00021408;0.00018515;0.00022855;0.00015767; 0.00013597;0.00029798;0.00034282;0.00021264……].

[0083] Through the above Figure 3 In step S302 of the load identification method shown, the reactive power and distortion power of the current power equipment to be tested are obtained, and the Euclidean distances between the cluster centers in step S303 are calculated in sequence. The closest ones are classified into the class to which the current cluster center belongs. Figure 4 The black dots indicated by the arrows in the figure are the cluster centers obtained by the three clustering in this embodiment.

[0084] Through the above Figure 3 In step S306 of the load identification method shown, the number of similarities between the current signal of the currently measured load device and the current signal of the dangerous reservoir is calculated. When the similarity value is higher than the set threshold, the currently measured device is determined to be dangerous.

[0085] Through the above Figure 3 In steps S307 to S309 of the load identification method shown, the VI trajectory is drawn and converted into a grayscale image; a convolutional neural network is used to extract grayscale image features and output the identification result.

[0086] Based on the integrated learning model, the above three recognition results are voted and the output results are as follows: compliance_number:28 cluster_label:1 network_label:0 Category:0 compliance_number is used to indicate the number of similar features; in this embodiment, the number of similar features is 28, which does not reach the number threshold, and the current device is judged to be not a dangerous device based on similarity. ‌cluster_label‌ refers to the label used to identify the cluster to which each data point belongs in the clustering algorithm. If the output label of the K-means unsupervised clustering is 1, it means that the current device is a dangerous device. If the output label network_label of the convolutional neural network is 0, it means that the current device is not a dangerous device. Therefore, based on the voting mechanism in the integration idea, there are two methods to determine whether the currently measured device is a dangerous device. One method determines that the currently measured device is a dangerous device, and two votes are greater than one vote. Finally, it is determined that the currently measured device is not a dangerous device, which is consistent with the actual device.

[0087] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0088] This embodiment also provides a load identification device, which is used to implement the above-mentioned embodiments and preferred implementation modes, and will not be repeated here. As used below, the terms "module", "unit", "subunit", etc. can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0089] Figure 5 is a structural block diagram of a load identification device according to an embodiment of the present application, such as Figure 5 As shown, the device includes: an acquisition module 51, an initial recognition module 52 and a voting module 53; wherein: The acquisition module 51 is used to acquire the electrical parameters of the power equipment to be tested when the load is connected.

[0090] The initial recognition module 52 is used to input the electrical parameters into the integrated learning model, and perform recognition processing based on the electrical parameters through at least two types of recognition algorithm models included in the integrated learning model to obtain corresponding at least two types of initial recognition results.

[0091] The voting module 53 is used to perform voting processing based on at least two types of initial identification results to generate a load identification result for the power equipment to be tested.

[0092] In some embodiments, the initial identification module 52 is further used to obtain power parameters of the power equipment to be tested based on the electrical parameters; and cluster the power parameters based on the first identification algorithm model to obtain a first initial identification result.

[0093] The initial identification module 52 is also used to obtain a preset dangerous device characteristic quantity; based on the second identification algorithm model, the electrical device characteristic quantity of the power equipment is compared with the dangerous device characteristic quantity to obtain a second initial identification result.

[0094] The initial identification module 52 is also used to generate a power parameter trajectory diagram of the power equipment to be tested based on the electrical parameters; input the power parameter trajectory diagram into the third identification algorithm model, and output a third initial identification result.

[0095] In some of the embodiments, the initial identification module 52 is further configured to determine a periodic voltage signal and a periodic current signal of the power device within a sampling period based on the electrical parameters.

[0096] The initial identification module 52 is further configured to calculate an angle difference between the periodic voltage signal and the periodic current signal.

[0097] The initial identification module 52 is further configured to calculate a power parameter according to the angle difference.

[0098] In some embodiments, the initial identification module 52 is further configured to perform frequency domain transformation processing on the periodic voltage signal to obtain an initial voltage signal phase, and perform unwrapping processing on the initial voltage signal phase to obtain a target voltage signal phase.

[0099] The initial identification module 52 is further used to perform frequency domain change processing on the periodic current signal to obtain an initial current signal phase, and to perform unwinding processing on the initial current signal phase to obtain a target current signal phase; The initial identification module 52 is further configured to calculate an angle difference value according to a phase difference between a target voltage signal phase and a target current signal phase.

[0100] In some embodiments, the power parameters include reactive power and distortion power; the initial identification module 52 is further configured to determine clustering features based on the reactive power and distortion power.

[0101] The initial recognition module 52 is also used to calculate the distance between the clustering feature and the initial clustering center based on the first recognition algorithm model; and iteratively update the initial clustering center to a new clustering center according to the distance until the distance between the clustering feature and the new clustering center converges.

[0102] The initial recognition module 52 is further configured to obtain a first initial recognition result based on the new cluster center.

[0103] In some embodiments, the initial identification module 52 is further used to calculate similarity characteristics between the electrical equipment characteristic quantity of the electric equipment and the characteristic quantity of the dangerous equipment based on the second identification algorithm model.

[0104] The initial recognition module 52 is further configured to obtain a second initial recognition result according to the similarity feature.

[0105] In some embodiments, the initial identification module 52 is further configured to generate a corresponding grayscale image according to the power parameter trajectory image.

[0106] The initial recognition module 52 is further used to calculate a pixel value matrix based on the grayscale image, and input the pixel value matrix into the third recognition algorithm model, and output a third initial recognition result.

[0107] In some embodiments, the voting module 53 is also used to perform voting based on at least two types of initial identification results to obtain a first voting result indicating that the corresponding power equipment under test is a dangerous device, and a second voting result indicating that the power equipment under test is a non-dangerous device.

[0108] The voting module 53 is further configured to generate a load identification result based on the first voting result and the second voting result.

[0109] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0110] This embodiment further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0111] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0112] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program: S1, obtaining electrical parameters of the power equipment to be tested when the load is connected.

[0113] S2, inputting the electrical parameters into the integrated learning model, and performing recognition processing based on the electrical parameters through at least two types of recognition algorithm models included in the integrated learning model to obtain corresponding at least two types of initial recognition results.

[0114] S3, performing voting processing based on at least two types of initial identification results to generate a load identification result for the power equipment to be tested.

[0115] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0116] In addition, in combination with the load identification method in the above embodiment, the embodiment of the present application can provide a storage medium for implementation. The storage medium stores a computer program; when the computer program is executed by a processor, any one of the load identification methods in the above embodiment is implemented.

[0117] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0118] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0119] Those skilled in the art should understand that the technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0120] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A load identification method, characterized in that: The method comprises: Obtaining electrical parameters of the power equipment under test when the load is connected; Inputting the electrical parameters into an integrated learning model, and performing identification processing based on the electrical parameters through at least two types of identification algorithm models included in the integrated learning model to obtain corresponding at least two types of initial identification results; Voting is performed based on the at least two types of initial identification results to generate a load identification result for the power equipment to be tested.

2. The load identification method according to claim 1, characterized in that: The at least two types of recognition algorithm models included in the integrated learning model are used to perform recognition processing based on the electrical parameters to obtain corresponding at least two types of initial recognition results, including: Based on the electrical parameters, power parameters of the power equipment to be tested are obtained; based on the first recognition algorithm model, the power parameters are clustered to obtain a first initial recognition result; Obtaining a preset dangerous device characteristic quantity; based on a second recognition algorithm model, comparing the electrical device characteristic quantity of the electric equipment with the dangerous device characteristic quantity to obtain a second initial recognition result; Based on the electrical parameters, an electric parameter trajectory diagram of the electric power equipment to be tested is generated; the electric parameter trajectory diagram is input into a third recognition algorithm model, and a third initial recognition result is output.

3. The load identification method according to claim 2, characterized in that: The acquiring the power parameter of the electric power device to be tested based on the electric parameter comprises: Based on the electrical parameters, determining a periodic voltage signal and a periodic current signal of the electrical equipment within a sampling period; Calculating an angle difference between the periodic voltage signal and the periodic current signal; The power parameter is calculated according to the angle difference.

4. The load identification method according to claim 3, characterized in that: The calculating the angle difference between the periodic voltage signal and the periodic current signal comprises: Performing frequency domain transformation processing on the periodic voltage signal to obtain an initial voltage signal phase, and performing unwinding processing on the initial voltage signal phase to obtain a target voltage signal phase; Performing frequency domain change processing on the periodic current signal to obtain an initial current signal phase, and performing unwinding processing on the initial current signal phase to obtain a target current signal phase; The angle difference is calculated according to a phase difference between the target voltage signal phase and the target current signal phase.

5. The load identification method according to claim 2, characterized in that: The power parameters include reactive power and distortion power; the power parameters are clustered based on the first recognition algorithm model to obtain a first initial recognition result, including: Determining clustering features based on the reactive power and the distortion power; Based on the first recognition algorithm model, calculating the distance between the cluster feature and the initial cluster center; iteratively updating the initial cluster center to a new cluster center according to the distance until the distance between the cluster feature and the new cluster center converges; Based on the new cluster center, the first initial recognition result is obtained.

6. The load identification method according to any one of claims 2 to 5, characterized in that: The step of comparing the electrical equipment characteristic quantity of the electric equipment with the characteristic quantity of the dangerous equipment based on the second recognition algorithm model to obtain a second initial recognition result includes: Based on the second identification algorithm model, calculating the similarity characteristics between the electrical equipment characteristic quantity of the electric equipment and the characteristic quantity of the dangerous equipment; The second initial recognition result is obtained according to the similarity feature.

7. The load identification method according to any one of claims 2 to 5, characterized in that: The step of inputting the electric power parameter trajectory diagram into a third recognition algorithm model and outputting a third initial recognition result includes: According to the electric power parameter trajectory diagram, generating a corresponding grayscale diagram; A pixel value matrix is ​​calculated based on the grayscale image, and the pixel value matrix is ​​input into the third recognition algorithm model, and the third initial recognition result is output.

8. The load identification method according to claim 1, characterized in that: The performing voting based on the at least two types of initial identification results to generate a load identification result for the power equipment to be tested includes: Performing voting based on the at least two types of initial identification results to obtain a first vote result indicating that the corresponding power equipment to be tested is a dangerous device, and a second vote result indicating that the power equipment to be tested is a non-dangerous device; The load identification result is generated based on the first voting result and the second voting result.

9. A load identification device, characterized in that: include: An acquisition module is used to obtain electrical parameters of the power equipment to be tested when the load is connected; An initial recognition module, used to input the electrical parameters into an integrated learning model, and perform recognition processing based on the electrical parameters through at least two types of recognition algorithm models included in the integrated learning model to obtain corresponding at least two types of initial recognition results; The voting module is used to perform voting processing based on the at least two types of initial identification results to generate a load identification result for the power equipment to be tested.

10. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the load identification method according to any one of claims 1 to 8 when running.

Images