Power network adaptive harmonic discrimination method and system based on cloud side-end cooperation

Abstract

The invention provides a power network adaptive harmonic discrimination method and system based on cloud side-end cooperation, and is applied to edge equipment, the edge equipment establishes communication with a cloud server and terminal equipment in a power network, and the method comprises the steps: obtaining a to-be-discriminated electric signal, inputting the electric signal into a preset model, the Gaussian noise parameter and the analysis signal parameter are determined by the preset model based on the electric signal; performing empirical mode decomposition on the to-be-decomposed signal to obtain an IMF dominated by a common signal; and removing the IMF from the electric signal to obtain a harmonic signal. The Gaussian noise parameter is adaptively set and the signal parameter is analyzed based on the parameter of the electric signal, so that the condition of inaccurate harmonic separation caused by inaccurate manual parameter setting is avoided, and the harmonic discrimination accuracy is improved. The IMF corresponding to the fixed frequency can be separated by adding the analysis signal, and the effect of eliminating the noise in the electric signal is improved.

Description

Technical Field

[0001] The present invention relates to the field of power network technology, and in particular to a power network adaptive harmonic identification method and system based on cloud-edge-end collaboration. Background Art

[0002] High-quality and reliable power quality is an important foundation for ensuring the rapid development of the social economy and the electricity demand for a better life for the people. The continued aggravation of the harmonic problem in the power system will threaten the stability of the power grid and the safety of electricity use. Therefore, it is imperative to accurately trace the source of harmonic disturbances, reasonably evaluate the impact of power supply and power users on power quality, and then strengthen the sense of responsibility of power users to ensure power quality.

[0003] At present, harmonic identification is mainly based on measured data, and the dominant source is found by refining data correlation and other methods. Under the new form of power electronics, disturbances from multiple sources are intertwined and superimposed. Not only are harmonic disturbances global, but the parameters of the power supply and power consumption parties are also changing dynamically, and the difficulty of identification is increasing day by day. It will be difficult to cope with the heavy harmonic identification work in the new power system by relying solely on the physical model of the power grid as the theoretical basis, combined with the traditional analysis model of on-site inspection and node-by-node analysis by operation and maintenance personnel. Entering the digital age with the "Internet of Everything" as its prominent feature, the digital transformation of the new power system will strongly promote the path tracking and source analysis of power grid harmonic disturbances.

[0004] The digital background provides data support for harmonic identification, but also brings new technical challenges: on the one hand, massive power quality monitoring data is collected in the cloud, occupying a large amount of communication storage resources, and multi-source data belongs to different information management areas, and the full-domain interoperability of databases has not yet been fully realized; on the other hand, with the application of various semiconductor materials in power systems, a large number of asymmetric loads have emerged, resulting in the performance of traditional harmonic detection methods becoming increasingly poor, and the harmonics in the power system are uncertain, and the optimal decomposition results cannot be obtained by manually configuring parameters. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method and system for adaptive harmonic identification of power networks based on cloud-edge-end collaboration to improve the accuracy and efficiency of harmonic identification of power networks.

[0006] According to one aspect of the present invention, a method for adaptive harmonic identification of a power network based on cloud-edge-end collaboration is provided, which is applied to an edge device, wherein the edge device establishes communication with a cloud server and a terminal device in the power network; the method comprises:

[0007] Obtaining a harmonic identification request, wherein the harmonic identification request includes an electrical signal to be identified, and the electrical signal to be identified is an original signal in the power network;

[0008] Inputting the electrical signal to be identified into a preset model so that the preset model adds Gaussian noise and analytical signal to the electrical signal to obtain a signal to be decomposed, wherein the Gaussian noise parameters include the amplitude peak and frequency range of the Gaussian noise, and the Gaussian noise parameters are determined by the preset model based on the electrical signal; the analytical signal parameters include the amplitude and frequency of the analytical signal, and the analytical signal parameters are determined by the preset model based on the electrical signal;

[0009] Performing empirical mode decomposition on the signal to be decomposed to obtain multiple groups of IMFs corresponding to the signal to be decomposed, wherein each group of IMFs corresponds to a frequency contained in the electrical signal;

[0010] Determine the average energy of the multiple groups of IMFs and obtain an IMF energy spectrum diagram;

[0011] Based on the IMF energy spectrum diagram, determining a first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and determining an IMF whose frequency is less than the frequency of the first minimum point as an IMF dominated by a common signal;

[0012] The IMF dominated by the ordinary signal is removed from the electrical signal to obtain a harmonic signal.

[0013] In a possible embodiment, the common signal includes a fundamental wave signal and a noise signal;

[0014] The method of determining, based on the IMF energy spectrum diagram, a first minimum point appearing with increasing frequency in the IMF energy spectrum diagram, and determining that an IMF having a frequency less than the frequency of the first minimum point is an IMF dominated by a common signal, comprises:

[0015] Based on the IMF energy spectrum diagram, determining a first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and determining an IMF whose frequency is less than the frequency of the first minimum point as a noise-dominated IMF;

[0016] The method further comprises:

[0017] Eliminating the noise-dominated IMF from the multiple groups of IMFs, and reconstructing the remaining IMFs after eliminating the noise-dominated IMFs to obtain a reconstructed signal;

[0018] Performing empirical mode decomposition on the reconstructed signal to obtain multiple groups of second IMFs corresponding to the reconstructed signal, and determining average energy of the multiple groups of second IMFs to obtain a second IMF energy spectrum diagram;

[0019] Based on the second IMF energy spectrum diagram, determining a first minimum point that appears in the second IMF energy spectrum diagram as the frequency increases, and determining that a second IMF having a frequency less than the frequency of the first minimum point is a fundamental-dominated IMF;

[0020] The step of removing the IMF dominated by the ordinary signal from the electrical signal to obtain a harmonic signal includes:

[0021] The noise-dominated IMF and the fundamental-dominated IMF are removed from the electrical signal to obtain a harmonic signal.

[0022] In a possible embodiment, the preset model includes a particle swarm algorithm, and the preset model is obtained by the following steps:

[0023] Acquire data to be trained, wherein the data to be trained is original electrical signal data, and the original electrical signal data includes noise, fundamental wave and harmonics; the label of the training data is a Gaussian noise amplitude vertex range and a Gaussian noise frequency range set for the training data;

[0024] Set the size of the particle swarm, the initial velocity and initial position of each particle, and the preset number of iterations;

[0025] Based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle group and the random amount, the speed and position of each particle are updated, wherein the inertia weight is used to evaluate the influence of the current speed of the particle on the updated particle speed, the first learning factor is used to evaluate the influence of the difference between the optimal position of the particle and the current position of the particle on the updated particle speed, the second learning factor is used to evaluate the influence of the difference between the optimal position of the particle group and the current position of the particle on the updated particle speed, and the random amount is the product of a preset coefficient and a random number in [0,1];

[0026] When the preset number of iterations is reached, determine whether the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data; if the optimal result does not fall within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data, adjust the inertia weight and the preset coefficient, and return to the step of updating the speed and position of each particle based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle swarm, and the random amount, until the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data;

[0027] The inertia weight at this time is determined as the target inertia weight, and the preset coefficient at this time is determined as the target coefficient.

[0028] In a possible embodiment, the velocity of the particle is updated according to the following formula:

[0029] v p (i+1)=wv p (i)+c1r1(i)[y p,best (i)-y p (i)]+c2r2(i)[y gbest (i)-y p (i)]+δr3(i)

[0030] Among them, v p (i+1) is the updated velocity of particle p, w is the inertia weight, v p (i) is the current speed of particle p, c1 is the first learning factor, c2 is the second learning factor, r1(i), r2(i) and r3(i) are all random numbers in [0,1], y p,best (i) is the optimal position of particle p, y p (i) is the current position of particle p, y gbest (i) is the optimal position of the particle swarm, δ is a preset coefficient; the inertia weight is positively correlated with the particle quality parameter, the particle quality parameter is positively correlated with the difference between the current position of the particle and the optimal position of the particle swarm, and negatively correlated with the difference between the optimal position of the particle swarm and the worst position of the particle swarm.

[0031] In a possible embodiment, the first learning factor and the second learning factor are obtained by the following formula:

[0032]

[0033] Among them, c1 is the first learning factor, c2 is the second learning factor, and c max is the maximum value of the first learning factor and the second learning factor of each particle in the particle swarm, c min is the minimum value of the first learning factor and the second learning factor of each particle in the particle swarm, I is the preset number of iterations, and i is the current number of iterations.

[0034] In a possible embodiment, the method further includes:

[0035] In the case that the optimal position of the particle is not updated for an iteration number exceeding a preset iteration number threshold, determining a low-speed particle whose current speed is lower than a preset speed threshold;

[0036] The speed and current position of the low-speed particle are updated so that the current speed of the low-speed particle is not less than the preset speed threshold.

[0037] In a possible embodiment, the Gaussian noise parameters are obtained by the following steps:

[0038] The Gaussian noise amplitude peak range and the frequency range are set based on the electrical signal to be identified; wherein the Gaussian noise amplitude peak is smaller than the electrical signal extreme value, and the frequency range minimum value is larger than 1 / 2 of the frequency range contained in the electrical signal;

[0039] Determine the optimal result of the Gaussian noise parameter based on the target inertia weight and the target coefficient by using the particle swarm algorithm;

[0040] When the optimal result falls within the Gaussian noise extreme value range and the frequency range, the optimal result is used as the target Gaussian noise parameter.

[0041] In a possible embodiment, the harmonic identification request further includes noise parameters, and the noise parameters include noise amplitude and frequency; the noise parameters are obtained through a filter set at the terminal device;

[0042] The analytical signal is a sine function or a cosine function, and the amplitude and frequency of the analytical signal are the same as the amplitude and frequency of the noise.

[0043] According to another aspect of the present invention, there is provided a power network adaptive harmonic identification system based on cloud-edge-end collaboration, the system comprising: a cloud server, an edge device and a terminal device, the terminal device being any type of terminal device in the power network;

[0044] The terminal device is used to send a harmonic identification request, wherein the harmonic identification request includes an electrical signal to be identified and noise parameters of noise included in the electrical signal, wherein the noise parameters include noise amplitude and frequency, and the electrical signal to be identified is an original signal in the power network;

[0045] The edge device is used to input the electrical signal to be identified into a preset model so that the preset model adds Gaussian noise and analytical signals to the electrical signal to obtain a signal to be decomposed, wherein the Gaussian noise parameters include the amplitude peak and frequency range of the Gaussian noise, and the Gaussian noise parameters are determined by the preset model based on the electrical signal; the analytical signal parameters include the amplitude and frequency of the analytical signal, and the analytical signal parameters are determined by the preset model based on the electrical signal; perform empirical mode decomposition on the signal to be decomposed to obtain multiple groups of IMFs corresponding to the signal to be decomposed, wherein, Each group of IMFs corresponds to a frequency contained in the electrical signal; the average energy of the multiple groups of IMFs is determined to obtain an IMF energy spectrum diagram; based on the IMF energy spectrum diagram, the first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and the IMF with a frequency less than the first minimum point is determined as the IMF dominated by the common signal; the IMF dominated by the common signal is removed from the electrical signal to obtain a harmonic signal; the electrical signal, Gaussian noise parameters, analytical signal parameters, the IMF dominated by the common signal, and the harmonic signal are sent to the cloud server;

[0046] The cloud server is used to store the electrical signal, Gaussian noise parameters, analytical signal parameters, IMF dominated by ordinary signals and the harmonic signal, and to update the Gaussian parameter range corresponding to the electrical signal based on the Gaussian noise parameters.

[0047] According to another aspect of the present invention, there is provided a power network adaptive harmonic identification device based on cloud-edge-end collaboration, which is applied to an edge device, and the edge device establishes communication with a cloud server and a terminal device in the power network; the device comprises:

[0048] An acquisition module, used for acquiring a harmonic identification request, wherein the harmonic identification request includes an electrical signal to be identified, and the electrical signal to be identified is an original signal in the power network;

[0049] An input module, used for inputting the electrical signal to be identified into a preset model, so that the preset model adds Gaussian noise and analytical signal to the electrical signal to obtain a signal to be decomposed, wherein the Gaussian noise parameters include the amplitude peak and frequency range of the Gaussian noise, and the Gaussian noise parameters are determined by the preset model based on the electrical signal; the analytical signal parameters include the amplitude and frequency of the analytical signal, and the analytical signal parameters are determined by the preset model based on the electrical signal;

[0050] A decomposition module, used for performing empirical mode decomposition on the signal to be decomposed to obtain multiple groups of IMFs corresponding to the signal to be decomposed, wherein each group of IMFs corresponds to a frequency contained in the electrical signal;

[0051] A spectrum module, used to determine the average energy of the multiple groups of IMFs and obtain an IMF energy spectrum diagram;

[0052] A determination module, configured to determine, based on the IMF energy spectrum diagram, a first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and determine that an IMF having a frequency less than the frequency of the first minimum point is an IMF dominated by a common signal;

[0053] The elimination module is used to eliminate the IMF dominated by the ordinary signal from the electrical signal to obtain a harmonic signal.

[0054] In a possible embodiment, the common signal includes a fundamental wave signal and a noise signal;

[0055] The method of determining, based on the IMF energy spectrum diagram, a first minimum point appearing with increasing frequency in the IMF energy spectrum diagram, and determining that an IMF having a frequency less than the frequency of the first minimum point is an IMF dominated by a common signal, comprises:

[0056] Based on the IMF energy spectrum diagram, determining a first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and determining an IMF whose frequency is less than the frequency of the first minimum point as a noise-dominated IMF;

[0057] The device also includes: a reconstruction module, which is used to remove the noise-dominated IMF from the multiple groups of IMFs, and reconstruct the remaining IMFs after removing the noise-dominated IMFs to obtain a reconstructed signal; perform empirical mode decomposition on the reconstructed signal to obtain multiple groups of second IMFs corresponding to the reconstructed signal, and determine the average energy of the multiple groups of second IMFs to obtain a second IMF energy spectrum diagram; based on the second IMF energy spectrum diagram, determine the first minimum point that appears in the second IMF energy spectrum diagram as the frequency increases, and determine that the second IMF with a frequency less than the frequency of the first minimum point is the fundamental-dominated IMF;

[0058] The step of removing the IMF dominated by the ordinary signal from the electrical signal to obtain a harmonic signal includes:

[0059] The noise-dominated IMF and the fundamental-dominated IMF are removed from the electrical signal to obtain a harmonic signal.

[0060] In a possible embodiment, the preset model includes a particle swarm algorithm, and the device further includes: a training module, wherein the training module is used to obtain the preset model through the following steps:

[0061] Acquire data to be trained, wherein the data to be trained is original electrical signal data, and the original electrical signal data includes noise, fundamental wave and harmonics; the label of the training data is a Gaussian noise amplitude vertex range and a Gaussian noise frequency range set for the training data;

[0062] Set the size of the particle swarm, the initial velocity and initial position of each particle, and the preset number of iterations;

[0063] Based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle group and the random amount, the speed and position of each particle are updated, wherein the inertia weight is used to evaluate the influence of the current speed of the particle on the updated particle speed, the first learning factor is used to evaluate the influence of the difference between the optimal position of the particle and the current position of the particle on the updated particle speed, the second learning factor is used to evaluate the influence of the difference between the optimal position of the particle group and the current position of the particle on the updated particle speed, and the random amount is the product of a preset coefficient and a random number in [0,1];

[0064] When the preset number of iterations is reached, determine whether the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data; if the optimal result does not fall within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data, adjust the inertia weight and the preset coefficient, and return to the step of updating the speed and position of each particle based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle swarm, and the random amount, until the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data;

[0065] The inertia weight at this time is determined as the target inertia weight, and the preset coefficient at this time is determined as the target coefficient.

[0066] In a possible embodiment, the velocity of the particle is updated according to the following formula:

[0067] v p (i+1)=wv p (i)+c1r1(i)[y p,best (i)-y p (i)]+c2r2(i)[y gbest (i)-y p (i)]+δr3(i)

[0068] Among them, v p(i+1) is the updated velocity of particle p, w is the inertia weight, v p (i) is the current speed of particle p, c1 is the first learning factor, c2 is the second learning factor, r1(i), r2(i) and r3(i) are all random numbers in [0,1], y p,best (i) is the optimal position of particle p, y p (i) is the current position of particle p, y gbest (i) is the optimal position of the particle swarm, δ is a preset coefficient; the inertia weight is positively correlated with the particle quality parameter, the particle quality parameter is positively correlated with the difference between the current position of the particle and the optimal position of the particle swarm, and negatively correlated with the difference between the optimal position of the particle swarm and the worst position of the particle swarm.

[0069] In a possible embodiment, the first learning factor and the second learning factor are obtained by the following formula:

[0070]

[0071] Among them, c1 is the first learning factor, c2 is the second learning factor, and c max is the maximum value of the first learning factor and the second learning factor of each particle in the particle swarm, cmin is the minimum value of the first learning factor and the second learning factor of each particle in the particle swarm, I is the preset number of iterations, and i is the current number of iterations.

[0072] In a possible embodiment, it is characterized in that the training module is used to determine a low-speed particle whose current speed is lower than a preset speed threshold when the optimal position of the particle is not updated for an iteration number exceeding a preset iteration number threshold;

[0073] The speed and current position of the low-speed particle are updated so that the current speed of the low-speed particle is not less than the preset speed threshold.

[0074] In a possible embodiment, the Gaussian noise parameters are obtained by the following steps:

[0075] The Gaussian noise amplitude peak range and the frequency range are set based on the electrical signal to be identified; wherein the Gaussian noise amplitude peak is smaller than the electrical signal extreme value, and the frequency range minimum value is larger than 1 / 2 of the frequency range contained in the electrical signal;

[0076] Determine the optimal result of the Gaussian noise parameter based on the target inertia weight and the target coefficient by using the particle swarm algorithm;

[0077] When the optimal result falls within the Gaussian noise extreme value range and the frequency range, the optimal result is used as the target Gaussian noise parameter.

[0078] In a possible embodiment, the harmonic identification request also includes noise parameters, which include noise amplitude and frequency; the noise parameters are obtained through a filter set at the terminal device; the analysis signal is a sine function or a cosine function, and the amplitude and frequency of the analysis signal are the same as the noise amplitude and frequency.

[0079] According to another aspect of the present invention, there is provided an electronic device, comprising:

[0080] Processor; and

[0081] Memory for storing programs,

[0082] In which, the program includes instructions, which, when executed by the processor, enable the processor to execute any of the above-mentioned methods for adaptive harmonic identification of power networks based on cloud-edge-end collaboration.

[0083] According to another aspect of the present invention, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable a computer to execute any of the above-mentioned methods for adaptive harmonic identification of power networks based on cloud-edge-end collaboration.

[0084] One or more technical solutions provided in the embodiments of the present invention, by adaptively setting the Gaussian noise parameters and analytical signal parameters added to the electrical signal based on the parameters of the electrical signal, avoid the situation where the harmonic separation is inaccurate due to inaccurate parameter setting caused by manually setting the parameters, improve the empirical mode decomposition effect, and improve the accuracy and efficiency of harmonic identification. In addition, by adding Gaussian noise to the electrical signal, the extreme values ​​caused by transient pulses in the electrical signal can be neutralized, thereby suppressing the modal aliasing phenomenon in the electrical signal, avoiding the influence of the modal aliasing phenomenon on the empirical mode decomposition, and further improving the accuracy and convenience of the empirical mode decomposition. In addition, by adding an analytical signal to the electrical signal, and the frequency of the analytical signal is fixed, the IMF corresponding to the fixed frequency can be separated, thereby improving the effect of removing the noise in the electrical signal, thereby improving the accuracy of the empirical mode decomposition, and improving the accuracy of subsequent harmonic identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Further details, features and advantages of the invention are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which:

[0086] Figure 1 A schematic diagram of a flow chart of a method for adaptive harmonic identification of a power network based on cloud-edge-end collaboration provided by an embodiment of the present invention;

[0087] Figure 2A schematic diagram of the structure of a power network adaptive harmonic identification system based on cloud-edge-end collaboration provided by an embodiment of the present invention;

[0088] Figure 3 Another structural schematic diagram of a power network adaptive harmonic identification system based on cloud-edge-end collaboration provided by an embodiment of the present invention;

[0089] Figure 4 A schematic diagram of the structure of a power network adaptive harmonic identification device based on cloud-edge-end collaboration provided by an embodiment of the present invention;

[0090] Figure 5 A block diagram of an exemplary electronic device that can be used to implement an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0091] Embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not intended to limit the scope of protection of the present invention.

[0092] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0093] The term "including" and its variations used in this document are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments". Relevant definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc. mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0094] It should be noted that the modifications of "one" and "plurality" mentioned in the present invention are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0095] The names of the messages or information exchanged between multiple devices in the embodiments of the present invention are only used for illustrative purposes, and are not used to limit the scope of these messages or information.

[0096] In order to improve the accuracy of harmonic identification in the power network, the embodiments of the present invention provide a power network adaptive harmonic identification method, system, electronic device and storage medium based on cloud-edge-end collaboration. The power network adaptive harmonic identification method based on cloud-edge collaboration provided by the embodiments of the present invention can be applied to any electronic device with harmonic identification function, which can be a server, a computer or a mobile terminal, etc.

[0097] In a possible embodiment, the power network adaptive harmonic identification method based on cloud collaboration provided by the embodiment of the present invention can be applied to edge devices, which can be any electronic device with harmonic identification function, specifically, a device that can provide a network entry point of the power network cloud server to the terminal device, and the edge device can communicate with the cloud and the power network terminal. Among them, the power network terminal can be any device with an electrical signal collection function in the power network, such as a transformer, an electric meter, etc. The above-mentioned cloud is a background server for storing data sent by the edge device. The following describes the scheme of the present invention with reference to the accompanying drawings:

[0098] like Figure 1 As shown, Figure 1 A flow chart of a method for adaptive harmonic identification of a power network based on cloud-edge-end collaboration provided by an embodiment of the present invention may include the following steps:

[0099] S101, obtaining a harmonic identification request, wherein the harmonic identification request includes an electrical signal to be identified, and the electrical signal to be identified is an original signal in the power network;

[0100] S102, inputting the electrical signal to be identified into a preset model, so that the preset model adds Gaussian noise and analytical signal to the electrical signal to obtain a signal to be decomposed, wherein the Gaussian noise parameters include the amplitude peak and frequency range of the Gaussian noise, and the Gaussian noise parameters are determined by the preset model based on the electrical signal; the analytical signal parameters include the amplitude and frequency of the analytical signal, and the analytical signal parameters are determined by the preset model based on the electrical signal;

[0101] S103, performing empirical mode decomposition on the signal to be decomposed to obtain multiple groups of IMFs corresponding to the signal to be decomposed, wherein each group of IMFs corresponds to a frequency contained in the electrical signal;

[0102] S104, determining the average energy of the multiple groups of IMFs, and obtaining an IMF energy spectrum diagram;

[0103] S105, based on the IMF energy spectrum diagram, determining a first minimum point that appears with increasing frequency in the IMF energy spectrum diagram, and determining an IMF with a frequency less than the frequency of the first minimum point as an IMF dominated by a common signal;

[0104] S106: Eliminate the IMF dominated by the ordinary signal from the electrical signal to obtain a harmonic signal.

[0105] By using the embodiment of the present invention, the Gaussian noise parameters and analytical signal parameters added to the electrical signal are adaptively set based on the parameters of the electrical signal, so as to avoid the situation that the harmonic separation is inaccurate due to the inaccurate parameter setting caused by the manual setting of the parameters, thereby improving the effect of the empirical mode decomposition, and improving the accuracy and efficiency of harmonic identification. Furthermore, by adding Gaussian noise to the electrical signal, the extreme values ​​caused by transient pulses in the electrical signal can be neutralized, thereby suppressing the modal aliasing phenomenon in the electrical signal, avoiding the influence of the modal aliasing phenomenon on the empirical mode decomposition, and further improving the accuracy and convenience of the empirical mode decomposition. In addition, by adding the analytical signal to the electrical signal, and the frequency of the analytical signal is fixed, the IMF corresponding to the fixed frequency can be separated, thereby improving the effect of removing the noise in the electrical signal, thereby improving the accuracy of the empirical mode decomposition, and improving the accuracy of the subsequent harmonic identification.

[0106] The above S101-S106 are exemplarily described below:

[0107] The harmonic identification request received by the edge device in S101 is issued by the terminal device, and there may be multiple terminal devices. In a possible embodiment, the terminal device may send a harmonic identification request to the edge device when it detects that the electrical signal meets the preset harmonic identification condition. The harmonic identification condition may be pre-set according to the actual application scenario. As a possible implementation, the harmonic identification condition may be that the voltage distortion rate exceeds the standard for a proportion of time exceeding a specified threshold within a preset time period. Voltage distortion refers to the voltage waveform no longer maintaining an ideal sinusoidal change, but due to the presence of power supply and use equipment with nonlinear impedance characteristics in the power system. These devices inject harmonic currents into the public power grid or generate harmonic voltages in the public power grid, causing the actual voltage waveform to deviate from the sine wave. The voltage distortion rate is usually expressed as a percentage of the ratio of the root mean square value of each harmonic voltage to the effective value of the fundamental voltage, which is used to measure the degree of voltage waveform distortion. The above-mentioned specified threshold can be flexibly set according to the actual application scenario. As a possible implementation, the specified threshold can be calculated by the following formula:

[0108]

[0109] Where V is the voltage level of the monitoring point; P 95％k is the 95% probability value of the total harmonic voltage distortion rate at the monitoring point; l It is the total harmonic phase voltage distortion rate limit of the voltage level kV specified by the national standard. Among them, the voltage level is usually divided into 1. safety voltage (usually below 36V); 2. low voltage (also divided into 220V and 380V); 3. high voltage (10kV-220kV); 4. ultra-high voltage 330kV-750kV; 5. ultra-high voltage 1000kV AC, ±800kV DC and above. In this embodiment, different values ​​can be preset for different voltage levels, and V is the preset value corresponding to the corresponding voltage level.

[0110] The above terminal devices can be flexibly selected by the user according to actual needs. For example, all terminal devices in the power network can be selected as devices that need to be identified by harmonics, or some terminal devices in the power network can be selected as devices that need to be identified by harmonics. The terminal device that needs to be identified by harmonics can send a harmonic identification request to the edge device when it detects that the electrical signal it receives meets the preset harmonic identification conditions.

[0111] The harmonic identification request sent by the terminal device may include an electrical signal within a preset time period that meets the preset harmonic identification conditions. The electrical signal may include a voltage signal, a current signal, and corresponding timestamp information. The timestamp information is used to indicate the time when the voltage signal or current signal is collected, and may be generated in any timing method. The preset time period may be set according to actual business needs, such as 1 hour, 2 hours, etc.

[0112] In actual application scenarios, the electrical signal may include noise signals, fundamental signals and harmonic signals. Therefore, in a possible embodiment, the above-mentioned harmonic identification request may also include the frequency and amplitude of the noise. As a possible implementation method, the above-mentioned terminal device may filter the electrical signal through a filter before sending it. The filter may be a high-pass filter or a band-pass filter. Specifically, the fundamental frequency and amplitude are usually known, while the harmonic frequency and amplitude are usually higher than the fundamental wave. Therefore, the fundamental wave and harmonics can be filtered based on the fundamental frequency to obtain the noise frequency and amplitude. In order to distinguish them from the harmonic signals that ultimately need to be identified, the above-mentioned noise signal and fundamental wave signal are referred to as ordinary signals in the embodiment of the present invention.

[0113] After receiving the harmonic identification request, the edge device can perform harmonic identification on the electrical signal contained in the harmonic identification request. In the embodiment of the present invention, the electrical signal can be harmonically identified by performing EEMD (Ensemble Empirical Mode Decomposition) on the electrical signal. EEMD is an improvement on the traditional EMD (Empirical Mode Decomposition). EMD decomposition is to decompose the signal into several intrinsic mode functions (IMF), each of which represents an essential feature or motion mode in the signal. The traditional EMD process first needs to determine the local extreme points of the electrical signal, and then use cubic splines to connect all local maxima and minima to form upper and lower envelopes, and then obtain the mean curve from the upper and lower envelopes. In the process of obtaining the envelope, when there is an abnormal event in the signal, it will affect the selection of the extreme point, making the extreme point unevenly distributed, resulting in the obtained envelope being a combination of the local envelope of the abnormal event and the real signal envelope. The mean calculated by the envelope and the IMF components screened out contain the inherent modes and abnormal events of the signal or the inherent modes of adjacent characteristic time scales, thereby producing a mode aliasing phenomenon. The above-mentioned abnormalities usually refer to the mixing of harmonic signals and / or noise signals into the electrical signal.

[0114] EEMD is a noise-assisted data analysis method proposed to address the shortcomings of the EMD method. Specifically, in the EEMD method, noise signals can be added to the electrical signal multiple times, and the noise signal is usually Gaussian noise.

[0115] In an embodiment of the present invention, the parameters of the Gaussian noise to be added to the electrical signal can be determined by a pre-trained model, and the Gaussian noise parameters can include the amplitude peak and frequency range of the Gaussian noise. The discrimination model includes a particle swarm algorithm, that is, the Gaussian noise parameters added to the electrical signal can be determined by the particle swarm algorithm. In a possible embodiment, the discrimination model can be trained by the following steps:

[0116] S11. Obtain data to be trained.

[0117] The training data is the original electrical signal in the power network, and the electrical signal may be a signal containing only fundamental wave and noise, or a signal containing fundamental wave, noise and harmonics, which is not specifically limited in the present invention. In a possible embodiment, the power network may store the generated electrical signal data in the cloud, and the training data may be the historical electrical signal data in the power network obtained from the cloud. The label of the training data is the Gaussian noise amplitude vertex range and the Gaussian noise frequency range set for the training data, which may be generated by manual labeling.

[0118] S12, setting the size of the particle group, the initial velocity and initial position of each particle, and the preset number of iterations.

[0119] Particle Swarm Optimization (PSO) originated from the study of bird predation behavior. The basic principle is to seek the optimal solution through information sharing and collaborative screening of individuals in the population. Specifically, the PSO uses multiple particles to simulate bird behavior and seeks the optimal solution by continuously updating particle speed and position.

[0120] The particle group size, the initial velocity of each particle, the initial position of each particle and the preset number of iterations can be flexibly set according to actual needs, and the present invention does not make any specific limitations on this.

[0121] S13. Update the speed and position of each particle based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle swarm and the random amount, wherein the inertia weight is used to evaluate the influence of the current speed of the particle on the updated particle speed, the first learning factor is used to evaluate the influence of the difference between the optimal position of the particle and the current position of the particle on the updated particle speed, the second learning factor is used to evaluate the influence of the difference between the optimal position of the particle swarm and the current position of the particle on the updated particle speed, and the random amount is the product of a preset coefficient and a random number in [0,1].

[0122] The preset period can be flexibly set according to actual needs. In this step, the speed of each particle can be updated based on the current speed, current position, optimal position and optimal position of the particle in the particle group. Among them, the optimal position can be determined according to the evaluation function, and the evaluation function can be set according to the actual application scenario. For example, the evaluation function can be a Griewank function, a Rastrigin function, etc. As a possible implementation, the optimal position of the particle and the optimal position of the particle group can be iterated. For example, the optimal position of the particle can be the function with the largest evaluation function value in all iterations.

[0123] As a possible implementation method, because the PSO algorithm sometimes falls into a local optimal solution, which may cause the optimization process to end prematurely. In order to avoid the algorithm from converging to a suboptimal solution too early, the PSO algorithm can be improved. Specifically, a random amount can be added to the speed of each particle within a certain range during each iteration of the algorithm to ensure the randomness of the algorithm during the search process, thereby reducing the probability of low-speed particles appearing and improving the algorithm's ability to find the global optimum.

[0124] Therefore, as a possible implementation, the speed of each particle can be updated by the following formula:

[0125] v p (i+1)=wv p (i)+c1r1(i)[y p,best (i)-y p (i)]+c2r2(i)[y gbest (i)-y p (i)]+δr3(i)

[0126] Wherein, w is the inertia weight, which can be flexibly set according to the actual application scenario. For example, an inertia weight of about 0.5 can produce a better optimization effect. The specific value of the inertia weight can be obtained through training. p (i) is the current velocity of the particle, c1 and c2 are learning factors, and their specific values ​​can be flexibly set according to the actual application scenario. r1(i), r2(i) and r3(i) are all randomly generated values ​​within [0,1]. p,best (i) is the optimal position of the current particle, y p (i) is the current position of the particle, specifically, y p (i) = y p (i-1)+v p (i-1), y best (i) is the optimal position in the particle swarm, δ is a preset coefficient, and the specific value of the preset coefficient can be obtained through training.

[0127] S14, when the preset number of iterations is reached, determine whether the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data; if the optimal result does not fall within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data, adjust the inertia weight and the preset coefficient, and return to the step of updating the speed and position of each particle based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle swarm and the random amount, until the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data; determine the inertia weight at this time as the target inertia weight, and determine the preset coefficient at this time as the target coefficient.

[0128] As a possible implementation, the quality of the particle position can be considered when setting the inertia weight. For example, the inertia weight is positively correlated with the particle quality parameter, which is positively correlated with the difference between the current position of the particle and the best position of the particle group, and negatively correlated with the difference between the best position of the particle group and the worst position of the particle group. The closer the particle is to the global optimum, the smaller the parameter. In this way, if the particle position is close to the global optimum, its inertia weight value will become smaller, so that the particle can better search near the global optimum. For example, the inertia weight can be set according to the following formula:

[0129]

[0130] Among them, F gbest represents the optimal position of the particle swarm, F gworst represents the worst position of the particle swarm, F p is the current position of the particle, I is the preset number of iterations, and i is the current number of iterations. The above w is the value obtained through training. The above formula is only a training result and does not specifically limit the inertia weight in the embodiment of the present invention.

[0131] In a possible embodiment, the learning factor value of the traditional discrete particle swarm algorithm can be adjusted by adding the influence of the number of iterations to the original formula. For example, the learning factor can be obtained by the following formula:

[0132]

[0133] Among them, c1 is the first learning factor, c2 is the second learning factor, and c max is the maximum value of the first learning factor and the second learning factor of each particle in the particle swarm, c min is the minimum value of the first learning factor and the second learning factor of each particle in the particle swarm, I is the preset number of iterations, and i is the current number of iterations.

[0134] In the early stage of particle swarm iteration, each particle has not found the optimal solution. At this time, it is necessary to focus on self-learning to independently find the optimal solution. At the end of the iteration, some particles have found the optimal solution. At this time, the optimal solution is found based on the social information of the particle swarm. Therefore, the individual learning factor (i.e., the first learning factor) should be gradually reduced according to the above formula, while the social learning factor (i.e., the second learning factor) should be gradually increased according to the above formula.

[0135] The reason why the particle swarm algorithm converges prematurely is that a large number of low-speed particles gather and fall into the local optimal solution. Affected by the low speed, these gathered particles cannot jump out of the local optimal solution where they are, so they take up a large amount of calculation and cannot participate in the search for the optimal solution. These particles are usually called "lazy particles". In order to avoid the influence of lazy particles on the optimization, a replacement mechanism can be set. Whenever "lazy particles" appear, they are replaced with some active particles. In this way, the algorithm can improve the search speed without falling into the local optimal solution.

[0136] In a possible embodiment, the method may further include:

[0137] In the case that the optimal position of the particle is not updated for an iteration number exceeding a preset iteration number threshold, determining a low-speed particle whose current speed is lower than a preset speed threshold;

[0138] The speed and current position of the low-speed particle are updated so that the current speed of the low-speed particle is not less than the preset speed threshold.

[0139] As a possible implementation, a speed threshold v can be set th , iterative threshold i th and counting variable i′, when F gbest When there is no update, i′ counts, when F gbest After the update, i′ is reset to 0. If the global optimal objective function value F gbest After multiple iterations, it is still not updated (i.e., i′>i th When the algorithm converges, we can determine that the algorithm has a convergence trend. At this time, we can select the low-speed particles (i.e., the speed v o (i) Less than the speed threshold v th The particles are replaced by randomizing the positions and velocities of these particles. The replacement formula is as follows:

[0140] v p (i+1)=rand(v min ,v max )

[0141] y p (i+1)=rand(v min ,v max )

[0142] sti′>i th ,v p (i) <v th

[0143] Among them, vmin is the minimum velocity of the particle group, vmax is the maximum velocity of the particle group, sti′>ith ,v p (i) <v th The application condition for the above substitution formula is F gbest The unupdated count exceeds the iteration threshold and the particle speed is less than the speed threshold.

[0144] In a possible embodiment, different Gaussian noise parameter ranges may be set for different electrical signals. When the optimization result of PSO for the electrical signal falls within the Gaussian parameter range pre-set for the electrical signal, it can be determined that PSO obtains the optimal result.

[0145] The Gaussian noise parameter range can be determined according to the amplitude and frequency of the electrical signal. As a possible implementation, the extreme value in the electrical signal can be used as the vertex of the Gaussian noise, and the frequency range included in the electrical signal can be used as the frequency range of the Gaussian noise. Or the Gaussian noise amplitude vertex is set to be smaller than the extreme value of the electrical signal, and the minimum value of the frequency range is greater than 1 / 2 of the frequency range included in the electrical signal.

[0146] In an embodiment of the present invention, in addition to adding Gaussian white noise with optimal parameters to the input signal, an analytical signal can also be added, and the analytical signal is used to separate the noise signal in the electrical signal. As described above, in an embodiment of the present invention, harmonic discrimination is achieved by performing EMD decomposition on the electrical signal, and the fundamental wave and harmonics included in the electrical signal are usually linear superpositions of sine and cosine signals, that is, the IMF obtained after performing EMD decomposition on the electrical signal is usually a sine function or a cosine function. In order to eliminate the electrical signal corresponding to the frequency containing noise, the IMF obtained by performing EMD decomposition on the frequency containing noise can be made to be the added analytical signal, so the analytical signal can be set to a sine signal or a cosine signal.

[0147] Furthermore, the noise signal is usually a low intermediate frequency signal, so the lower frequency may cause the IMF dominated by the analytical signal to be seriously affected by the intermediate frequency noise on the analytical component IMF. On the other hand, if the frequency of the analytical signal is too high, it is equivalent to a high-frequency noise signal and cannot effectively concentrate the intermediate frequency signal. Therefore, the analytical signal can be an intermediate frequency signal, so that when performing EMD decomposition, the noise signal can be decomposed through the intermediate frequency signal.

[0148] In a possible embodiment, the parameters of the analytical signal may be the same as the noise parameters included in the harmonic request, specifically, may be the same as the amplitude and frequency of the noise.

[0149] When the edge device receives a harmonic identification request, it can generate Gaussian noise and analytical signals added to the electrical signal contained in the harmonic identification request through the above model to obtain a signal to be decomposed. The signal to be decomposed can be obtained by the following formula:

[0150] X i (t) = x(t) + x i (t)+n i (t)i=1,2,……,N

[0151] Among them, X i (t) is the signal to be decomposed at time t, x(t) is the electrical signal at time t, x i (t) is the analytical signal added to the electrical signal at time t, n i (t) is the i-th Gaussian noise added to the electrical signal at time t, i represents the frequency number contained in the electrical signal at time t, and N is the number of frequencies contained in the electrical signal at time t.

[0152] Afterwards, EMD decomposition may be performed on the signal to be decomposed at each moment. Specifically, EMD decomposition may be performed on the signal to be decomposed at each of the above-mentioned moments.

[0153] By X i (t) Perform EMD decomposition to obtain N groups of IMFs, each group of IMFs corresponds to different frequencies, and there can be multiple IMFs in each group of IMFs. Then, the overall mean and energy of IMFs can be calculated for each group of IMFs. Specifically, the overall energy mean of IMFs can be calculated according to the following formula:

[0154]

[0155] Among them, imc y is the overall mean of the IMF group corresponding to frequency y, N is the number of IMFs contained in this group of IMFs, and imc x,y (t) is the energy of the eigenmode function corresponding to frequency y obtained by performing the x-th decomposition of Xi(t).

[0156] In a possible embodiment, the energy calculation formula of the eigenmode component is as follows:

[0157]

[0158] Among them, E a represents the energy of the ath eigenmode component, and M is the length of the eigenmode component sequence, that is, the number of IMFs in the IMF group.

[0159] Based on the energy of each group of IMFs, the energy distribution of Xi(t) can be obtained. The energy distribution can be specifically a spectrum energy distribution, where the horizontal axis is the frequencies included in Xi(t) and the vertical axis is the overall energy mean of the IMF group corresponding to each frequency.

[0160] The fundamental wave is usually the main component in an electrical signal, and its energy is usually higher. Harmonics are usually generated by the distortion of the fundamental wave current caused by applying a sinusoidal voltage to a nonlinear load. Therefore, their energy is usually higher. Noise signals usually have lower energy.

[0161] In the spectrum diagram, the energy of the IMF component usually increases with the increase of frequency. If the energy of the IMF component reaches a minimum value as the frequency increases, it can be determined that the IMF corresponding to the minimum value is the IMF corresponding to the new type of wave, and the frequency of the noise or fundamental wave is generally low, that is, the frequency of the ordinary signal is generally low. Therefore, in an embodiment of the present invention, based on the above spectrum diagram, the first energy minimum point that appears as the frequency increases can be determined, and the IMF on the left side of the minimum point is the IMF corresponding to the ordinary signal, and the IMF on the left side of the minimum point can be removed by a filter. Specifically, a high-pass filter or a band-stop filter can be used, and the filter frequency of the filter can be set so that the IMF higher than the filter frequency is retained.

[0162] In a possible embodiment, since the energy and frequency of the noise may be smaller than the energy and frequency of the fundamental wave, the IMF dominated by the common signal obtained through the above steps may be the IMF dominated by noise.

[0163] Through the above technical solution, Gaussian white noise and analytical signals are concentrated into high-frequency IMF components, so that the noise-dominated components are mainly concentrated on low-order IMFs, and the noise-dominated IMF components are filtered. The turning point of the IMF energy curve is used to construct an EMD-based low-pass filter method, so that the noise-dominated IMF can be filtered out.

[0164] In a possible embodiment, the above method may further include:

[0165] Eliminating the noise-dominated IMF from multiple groups of IMFs, and reconstructing the remaining IMFs after eliminating the noise-dominated IMFs to obtain a reconstructed signal;

[0166] Performing empirical mode decomposition on the reconstructed signal to obtain multiple groups of second IMFs corresponding to the reconstructed signal, and determining average energy of the multiple groups of second IMFs to obtain a second IMF energy spectrum diagram;

[0167] Based on the second IMF energy spectrum diagram, determining a first minimum point that appears in the second IMF energy spectrum diagram as the frequency increases, and determining that a second IMF having a frequency less than the frequency of the first minimum point is a fundamental-dominated IMF;

[0168] The noise-dominated IMF and the fundamental-dominated IMF are removed from the electrical signal to obtain a harmonic signal.

[0169] For the convenience of description, in the embodiment of the present invention, the IMF dominated by the removed noise may be referred to as a noise IMF component, and the IMF after the noise IMF component is removed may be referred to as a residual IMF.

[0170] After that, the remaining IMFs can be reconstructed. Specifically, each IMF and its envelope can be used to reconstruct the signal, each IMF and its envelope are multiplied, these combinations are summed to obtain the reconstructed signal, and the reconstructed signal is subjected to EMD decomposition to obtain the spectrum graphs corresponding to the reconstructed signal. The process of obtaining the spectrum graph corresponding to the reconstructed signal and the process of obtaining the spectrum graph corresponding to the original electrical signal can be referred to the above description, and will not be repeated here.

[0171] By determining the minimum point of each spectrum graph corresponding to the reconstructed signal, the IMF on the left side of the minimum point is obtained as the fundamental component IMF. Subtracting the noise IMF and the fundamental IMF from the original electrical signal will obtain the harmonic signal in the electrical signal.

[0172] By applying the above technical solution, there is no mode mixing in the fundamental wave obtained by reconstructing the EMD of the signal, the total harmonic distortion of the obtained fundamental wave component IMF is very small, the decomposed fundamental wave is consistent with the ideal fundamental wave, and the amplitude is almost not attenuated.

[0173] In a possible embodiment, the edge device may send the electrical signal, Gaussian noise parameters, analytical signal parameters, noise-dominated IMF, fundamental-dominated IMF, and the harmonic signal to the cloud server. The cloud server displays the electrical signal, Gaussian noise parameters, analytical signal parameters, noise-dominated IMF, fundamental-dominated IMF, and the harmonic signal.

[0174] By using edge-side devices to process the original electrical signal data and uploading the processing results to the cloud, the pressure on cloud data communication and storage is reduced. At the same time, the cloud can obtain the processing result data uploaded by multiple edge-side devices, realizing the intercommunication of data from different sources.

[0175] In a possible embodiment, a digital platform can also be set up, and the user can use the data stored in the cloud to perform simulation deduction of the identification and analysis results on the digital platform, and can analyze the mapping relationship between multiple disturbance sources and propagation paths in combination with the grid topology, harmonic propagation range and multi-source harmonic emission level, and prove the correctness of the identification results through the joint deduction of multi-source harmonic propagation paths.

[0176] In a possible embodiment, the correspondence between the electrical signal and the Gaussian noise parameter range can also be updated according to the data such as the electrical signal and Gaussian noise parameters stored in the cloud, such as updating the conversion coefficient between the Gaussian noise parameter range and the electrical signal peak value and frequency.

[0177] By applying the embodiment of the present invention, a cloud-edge-end collaborative harmonic identification architecture is constructed. After the terminal device collects data, it is first processed by the edge computing device and then uploaded to the cloud computing platform for large-scale processing and analysis. Adaptive situational awareness is used at the harmonic identification level of the cloud-edge-end architecture to adaptively adjust the collection and recording mode for harmonic scenes with different disturbance frequency bands and time-varying characteristics, obtain more harmonic over-standard recording data, and realize environmental differentiated edge states. In addition, based on the empirical mode decomposition (EEMD) method, parameters are adaptively set according to the characteristics of various harmonic contents in the signal to be separated, thereby allowing more effective separation of harmonics and avoiding the defects of manually preset parameters.

[0178] In the processing of complex signals, white noise and analytical signals with certain parameters are injected into the target signal to suppress mode mixing and fundamental wave attenuation. The parameters of the analytical signal are selected according to the grid conditions and signal characteristics, and then a signal reconstruction method based on the IMF energy extreme point distribution and the law of distinguishing between signal and noise is used to reduce the influence of the dominant component of noise.

[0179] In the process of selecting experimental data, data are sampled from different types of power grid environments to ensure that the trained model has the most universal applicability possible, and a reasonable deep neural network structure is selected based on multiple experiments to ensure the practicality and effectiveness of the trained model.

[0180] According to another aspect of the present invention, a power network adaptive harmonic identification system based on cloud-edge-end collaboration is provided. Figure 2 As shown, the system may include: a terminal device 201, an edge device 202, and a cloud server 203, wherein the terminal device is any type of terminal device in the power network;

[0181] The terminal device 201 is used to send a harmonic identification request, wherein the harmonic identification request includes an electrical signal to be identified and noise parameters of noise included in the electrical signal, wherein the noise parameters include noise amplitude and frequency, and the electrical signal to be identified is an original signal in the power network;

[0182] The edge device 202 is used to input the electrical signal to be identified into a preset model so that the preset model adds Gaussian noise and analytical signals to the electrical signal to obtain a signal to be decomposed, wherein the Gaussian noise parameters include the amplitude peak and frequency range of the Gaussian noise, and the Gaussian noise parameters are determined by the preset model based on the electrical signal; the analytical signal parameters include the amplitude and frequency of the analytical signal, and the analytical signal parameters are determined by the preset model based on the electrical signal; perform empirical mode decomposition on the signal to be decomposed to obtain multiple groups of IMFs corresponding to the signal to be decomposed, wherein , each group of IMFs corresponds to a frequency contained in the electrical signal; determine the average energy of the multiple groups of IMFs to obtain an IMF energy spectrum diagram; based on the IMF energy spectrum diagram, determine the first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and determine that the IMF with a frequency less than the frequency of the first minimum point is the IMF dominated by the common signal; remove the IMF dominated by the common signal from the electrical signal to obtain a harmonic signal; send the electrical signal, Gaussian noise parameters, analytical signal parameters, the IMF dominated by the common signal, and the harmonic signal to the cloud server;

[0183] The cloud server 203 is used to store the electrical signal, Gaussian noise parameters, analytical signal parameters, IMF dominated by ordinary signals, and the harmonic signal, and to update the Gaussian parameter range corresponding to the electrical signal based on the Gaussian noise parameters.

[0184] like Figure 3 As shown, Figure 3 An architecture diagram of a power network adaptive harmonic identification system based on cloud-edge-end collaboration provided by an embodiment of the present invention, specifically, the terminal device initiates a service request and uploads electrical signal data, the edge side device provides data access integration function, information preprocessing function, and the information preprocessing function is implemented by the above-mentioned power network adaptive harmonic identification method based on cloud-edge-end collaboration. The edge side device can send the identification results to the cloud server for centralized processing and fusion analysis.

[0185] Based on the same inventive concept, an embodiment of the present invention also provides a power network adaptive harmonic identification device based on cloud-edge-end collaboration, which is applied to an edge device, and the edge device establishes communication with a cloud server and a terminal device in the power network; Figure 4 As shown, the apparatus 400 may include:

[0186] An acquisition module 401 is used to acquire a harmonic identification request, wherein the harmonic identification request includes an electrical signal to be identified, and the electrical signal to be identified is an original signal in the power network;

[0187] An input module 402 is used to input the electrical signal to be identified into a preset model, so that the preset model adds Gaussian noise and analytical signal to the electrical signal to obtain a signal to be decomposed, wherein the Gaussian noise parameters include the amplitude peak and frequency range of the Gaussian noise, and the Gaussian noise parameters are determined by the preset model based on the electrical signal; the analytical signal parameters include the amplitude and frequency of the analytical signal, and the analytical signal parameters are determined by the preset model based on the electrical signal;

[0188] A decomposition module 403 is used to perform empirical mode decomposition on the signal to be decomposed to obtain multiple groups of IMFs corresponding to the signal to be decomposed, wherein each group of IMFs corresponds to a frequency contained in the electrical signal;

[0189] The spectrum module 404 is used to determine the average energy of the multiple groups of IMFs and obtain an IMF energy spectrum diagram;

[0190] A determination module 405 is used to determine, based on the IMF energy spectrum diagram, a first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and determine that an IMF having a frequency less than the frequency of the first minimum point is an IMF dominated by a common signal;

[0191] The elimination module 406 is used to eliminate the IMF dominated by the common signal from the electrical signal to obtain a harmonic signal.

[0192] In a possible embodiment, the common signal includes a fundamental wave signal and a noise signal;

[0193] The method of determining, based on the IMF energy spectrum diagram, a first minimum point appearing with increasing frequency in the IMF energy spectrum diagram, and determining that an IMF having a frequency less than the frequency of the first minimum point is an IMF dominated by a common signal, comprises:

[0194] Based on the IMF energy spectrum diagram, determining a first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and determining an IMF whose frequency is less than the frequency of the first minimum point as a noise-dominated IMF;

[0195] The device also includes: a reconstruction module, which is used to remove the noise-dominated IMF from the multiple groups of IMFs, and reconstruct the remaining IMFs after removing the noise-dominated IMFs to obtain a reconstructed signal; perform empirical mode decomposition on the reconstructed signal to obtain multiple groups of second IMFs corresponding to the reconstructed signal, and determine the average energy of the multiple groups of second IMFs to obtain a second IMF energy spectrum diagram; based on the second IMF energy spectrum diagram, determine the first minimum point that appears in the second IMF energy spectrum diagram as the frequency increases, and determine that the second IMF with a frequency less than the frequency of the first minimum point is the fundamental-dominated IMF;

[0196] The step of removing the IMF dominated by the ordinary signal from the electrical signal to obtain a harmonic signal includes:

[0197] The noise-dominated IMF and the fundamental-dominated IMF are removed from the electrical signal to obtain a harmonic signal.

[0198] In a possible embodiment, the preset model includes a particle swarm algorithm, and the device further includes: a training module, wherein the training module is used to obtain the preset model through the following steps:

[0199] Acquire data to be trained, wherein the data to be trained is original electrical signal data, and the original electrical signal data includes noise, fundamental wave and harmonics; the label of the training data is a Gaussian noise amplitude vertex range and a Gaussian noise frequency range set for the training data;

[0200] Set the size of the particle swarm, the initial velocity and initial position of each particle, and the preset number of iterations;

[0201] Based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle group and the random amount, the speed and position of each particle are updated, wherein the inertia weight is used to evaluate the influence of the current speed of the particle on the updated particle speed, the first learning factor is used to evaluate the influence of the difference between the optimal position of the particle and the current position of the particle on the updated particle speed, the second learning factor is used to evaluate the influence of the difference between the optimal position of the particle group and the current position of the particle on the updated particle speed, and the random amount is the product of a preset coefficient and a random number in [0,1];

[0202] When the preset number of iterations is reached, determine whether the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data; if the optimal result does not fall within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data, adjust the inertia weight and the preset coefficient, and return to the step of updating the speed and position of each particle based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle swarm, and the random amount, until the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data;

[0203] The inertia weight at this time is determined as the target inertia weight, and the preset coefficient at this time is determined as the target coefficient.

[0204] In a possible embodiment, the velocity of the particle is updated according to the following formula:

[0205] v p (i+1)=wv p (i)+c1r1(i)[y p,best (i)-y p (i)]+c2r2(i)[y gbest (i)-y p (i)]+δr3(i)

[0206] Among them, v p (i+1) is the updated velocity of particle p, w is the inertia weight, v p (i) is the current speed of particle p, c1 is the first learning factor, c2 is the second learning factor, r1(i), r2(i) and r3(i) are all random numbers in [0,1], y p,best (i) is the optimal position of particle p, y p (i) is the current position of particle p, y gbest (i) is the optimal position of the particle swarm, δ is a preset coefficient; the inertia weight is positively correlated with the particle quality parameter, the particle quality parameter is positively correlated with the difference between the current position of the particle and the optimal position of the particle swarm, and negatively correlated with the difference between the optimal position of the particle swarm and the worst position of the particle swarm.

[0207] In a possible embodiment, the first learning factor and the second learning factor are obtained by the following formula:

[0208]

[0209] Among them, c1 is the first learning factor, c2 is the second learning factor, and c maxis the maximum value of the first learning factor and the second learning factor of each particle in the particle swarm, cmin is the minimum value of the first learning factor and the second learning factor of each particle in the particle swarm, I is the preset number of iterations, and i is the current number of iterations.

[0210] In a possible embodiment, it is characterized in that the training module is used to determine a low-speed particle whose current speed is lower than a preset speed threshold when the optimal position of the particle is not updated for an iteration number exceeding a preset iteration number threshold;

[0211] The speed and current position of the low-speed particle are updated so that the current speed of the low-speed particle is not less than the preset speed threshold.

[0212] In a possible embodiment, the Gaussian noise parameters are obtained by the following steps:

[0213] The Gaussian noise amplitude peak range and the frequency range are set based on the electrical signal to be identified; wherein the Gaussian noise amplitude peak is smaller than the electrical signal extreme value, and the frequency range minimum value is larger than 1 / 2 of the frequency range contained in the electrical signal;

[0214] Determine the optimal result of the Gaussian noise parameter based on the target inertia weight and the target coefficient by using the particle swarm algorithm;

[0215] When the optimal result falls within the Gaussian noise extreme value range and the frequency range, the optimal result is used as the target Gaussian noise parameter.

[0216] In a possible embodiment, the harmonic identification request also includes noise parameters, which include noise amplitude and frequency; the noise parameters are obtained through a filter set at the terminal device; the analysis signal is a sine function or a cosine function, and the amplitude and frequency of the analysis signal are the same as the noise amplitude and frequency.

[0217] Among them, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in the present invention are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0218] The exemplary embodiment of the present invention further provides an electronic device, comprising: at least one processor; and a memory connected to the at least one processor in communication. The memory stores a computer program that can be executed by the at least one processor, and the computer program is used to enable the electronic device to perform a method according to an embodiment of the present invention when executed by the at least one processor.

[0219] Exemplary embodiments of the present invention also provide a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to perform a method according to an embodiment of the present invention.

[0220] An exemplary embodiment of the present invention further provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor of a computer, the computer is used to enable the computer to perform a method according to an embodiment of the present invention.

[0221] refer to Figure 5 , a block diagram of an electronic device 500 that can be used as a server or client of the present invention will now be described, which is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0222] like Figure 5 As shown, the electronic device 500 includes a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 can also be stored. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0223] A plurality of components in the electronic device 500 are connected to the I / O interface 505, including: an input unit 506, an output unit 507, a storage unit 508, and a communication unit 509. The input unit 506 may be any type of device capable of inputting information to the electronic device 500, and the input unit 506 may receive input digital or character information, and generate key signal inputs related to user settings and / or function control of the electronic device. The output unit 507 may be any type of device capable of presenting information, and may include, but is not limited to, a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 508 may include, but is not limited to, a disk, an optical disk. The communication unit 509 allows the electronic device 500 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and may include, but is not limited to, a modem, a network card, an infrared communication device, a wireless communication transceiver, and / or a chipset, such as a Bluetooth™ device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0224] The computing unit 501 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 501 performs the various methods and processes described above. For example, in some embodiments, the above-mentioned power network adaptive harmonic identification method based on cloud-edge collaboration may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 500 via the ROM 502 and / or the communication unit 509. In some embodiments, the computing unit 501 may be configured to perform the above-mentioned power network adaptive harmonic identification method based on cloud-edge collaboration in any other appropriate manner (e.g., by means of firmware).

[0225] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0226] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0227] As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and / or device (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0228] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0229] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0230] A computer system may include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship to each other.

Claims

1. A method for adaptive harmonic identification of power network based on cloud-edge-end collaboration, characterized in that: Applied to edge devices, the edge devices establish communication with cloud servers and terminal devices in the power network; the method includes: Obtaining a harmonic identification request, wherein the harmonic identification request includes an electrical signal to be identified, and the electrical signal to be identified is an original signal in the power network; Inputting the electrical signal to be identified into a preset model so that the preset model adds Gaussian noise and analytical signal to the electrical signal to obtain a signal to be decomposed, wherein Gaussian noise parameters include the amplitude peak and frequency range of Gaussian noise, and the Gaussian noise parameters are determined by the preset model based on the electrical signal; analytical signal parameters include the amplitude and frequency of the analytical signal, and the analytical signal parameters are determined by the preset model based on the electrical signal; Performing empirical mode decomposition on the signal to be decomposed to obtain multiple groups of IMFs corresponding to the signal to be decomposed, wherein each group of IMFs corresponds to a frequency contained in the electrical signal; Determine the average energy of the multiple groups of IMFs and obtain an IMF energy spectrum diagram; Based on the IMF energy spectrum diagram, determining a first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and determining an IMF whose frequency is less than the frequency of the first minimum point as an IMF dominated by a common signal; The IMF dominated by the ordinary signal is removed from the electrical signal to obtain a harmonic signal.

2. The method according to claim 1, characterized in that The common signal includes a fundamental wave signal and a noise signal; The method of determining, based on the IMF energy spectrum diagram, a first minimum point appearing with increasing frequency in the IMF energy spectrum diagram, and determining that an IMF having a frequency less than the frequency of the first minimum point is an IMF dominated by a common signal includes: Based on the IMF energy spectrum diagram, determining a first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and determining an IMF whose frequency is less than the frequency of the first minimum point as a noise-dominated IMF; The method further comprises: Eliminating the noise-dominated IMF from the multiple groups of IMFs, and reconstructing the remaining IMFs after eliminating the noise-dominated IMFs to obtain a reconstructed signal; Performing empirical mode decomposition on the reconstructed signal to obtain multiple groups of second IMFs corresponding to the reconstructed signal, and determining average energy of the multiple groups of second IMFs to obtain a second IMF energy spectrum diagram; Based on the second IMF energy spectrum diagram, determining a first minimum point that appears with increasing frequency in the second IMF energy spectrum diagram, and determining that a second IMF having a frequency less than the frequency of the first minimum point is a fundamental-dominated IMF; The step of removing the IMF dominated by the ordinary signal from the electrical signal to obtain a harmonic signal includes: The noise-dominated IMF and the fundamental-dominated IMF are removed from the electrical signal to obtain a harmonic signal.

3. The method according to claim 1, characterized in that The preset model includes a particle swarm algorithm, and the preset model is obtained by the following steps: Acquire data to be trained, wherein the data to be trained is original electrical signal data, and the original electrical signal data includes noise, fundamental wave and harmonics; the label of the training data is a Gaussian noise amplitude vertex range and a Gaussian noise frequency range set for the training data; Set the size of the particle swarm, the initial velocity and initial position of each particle, and the preset number of iterations; Based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle group and the random amount, the speed and position of each particle are updated, wherein the inertia weight is used to evaluate the influence of the current speed of the particle on the updated particle speed, the first learning factor is used to evaluate the influence of the difference between the optimal position of the particle and the current position of the particle on the updated particle speed, the second learning factor is used to evaluate the influence of the difference between the optimal position of the particle group and the current position of the particle on the updated particle speed, and the random amount is the product of a preset coefficient and a random number in [0,1]; When the preset number of iterations is reached, determine whether the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data; if the optimal result does not fall within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data, adjust the inertia weight and the preset coefficient, and return to the step of updating the speed and position of each particle based on the inertia weight, the first learning factor, the second learning factor, the current speed of the particle, the current position of the particle, the optimal position of the particle, the optimal position of the particle swarm, and the random amount, until the optimal result found by the particle swarm falls within the Gaussian noise amplitude vertex range and the Gaussian noise frequency range corresponding to the corresponding training data; The inertia weight at this time is determined as the target inertia weight, and the preset coefficient at this time is determined as the target coefficient.

4. The method according to claim 3, characterized in that Update the particle's velocity according to the following formula: v p (i+1)=wv p (i)+c1r1(i)[y p,best (i)-y p (i)]+c2r2(i)[y gbest (i)-y p (i)]+δr3(i) Among them, v p (i+1) is the updated velocity of particle p, w is the inertia weight, v p (i) is the current speed of particle p, c1 is the first learning factor, c2 is the second learning factor, r1(i), r2(i) and r3(i) are all random numbers in [0,1], y p,best (i) is the optimal position of particle p, y p (i) is the current position of particle p, y gbest (i) is the optimal position of the particle swarm, δ is a preset coefficient; the inertia weight is positively correlated with the particle quality parameter, the particle quality parameter is positively correlated with the difference between the current position of the particle and the optimal position of the particle swarm, and negatively correlated with the difference between the optimal position of the particle swarm and the worst position of the particle swarm; The method further comprises: In the case that the optimal position of the particle is not updated for an iteration number exceeding a preset iteration number threshold, determining a low-speed particle whose current speed is lower than a preset speed threshold; The speed and current position of the low-speed particle are updated so that the current speed of the low-speed particle is not less than the preset speed threshold.

5. The method according to claim 3, characterized in that: The first learning factor and the second learning factor are obtained by the following formula: Among them, c1 is the first learning factor, c2 is the second learning factor, and c max is the maximum value of the first learning factor and the second learning factor of each particle in the particle swarm, cmin is the minimum value of the first learning factor and the second learning factor of each particle in the particle swarm, I is the preset number of iterations, and i is the current number of iterations.

6. The method according to claim 3, characterized in that The Gaussian noise parameters are obtained by the following steps: The Gaussian noise amplitude peak range and the frequency range are set based on the electrical signal to be identified; wherein the Gaussian noise amplitude peak is smaller than the electrical signal extreme value, and the frequency range minimum value is larger than 1 / 2 of the frequency range contained in the electrical signal; Determine the optimal result of the Gaussian noise parameter based on the target inertia weight and the target coefficient by using the particle swarm algorithm; When the optimal result falls within the Gaussian noise extreme value range and the frequency range, the optimal result is used as the target Gaussian noise parameter.

7. The method according to claim 1, characterized in that The harmonic identification request also includes noise parameters, which include noise amplitude and frequency; the noise parameters are obtained through a filter set at the terminal device; The analytical signal is a sine function or a cosine function, and the amplitude and frequency of the analytical signal are the same as the amplitude and frequency of the noise.

8. A power network adaptive harmonic identification system based on cloud-edge-end collaboration, characterized in that: The system comprises: a cloud server, an edge device and a terminal device, wherein the terminal device is any type of terminal device in the power network; The terminal device is used to send a harmonic identification request, wherein the harmonic identification request includes an electrical signal to be identified and noise parameters of noise included in the electrical signal, wherein the noise parameters include noise amplitude and frequency, and the electrical signal to be identified is an original signal in the power network; The edge device is used to input the electrical signal to be identified into a preset model so that the preset model adds Gaussian noise and analytical signals to the electrical signal to obtain a signal to be decomposed, wherein Gaussian noise parameters include the amplitude peak and frequency range of Gaussian noise, and the Gaussian noise parameters are determined by the preset model based on the electrical signal; analytical signal parameters include the amplitude and frequency of the analytical signal, and the analytical signal parameters are determined by the preset model based on the electrical signal; perform empirical mode decomposition on the signal to be decomposed to obtain multiple groups of IMFs corresponding to the signal to be decomposed, wherein each group The IMF corresponds to a frequency contained in the electrical signal; the average energy of the multiple groups of IMFs is determined to obtain an IMF energy spectrum diagram; based on the IMF energy spectrum diagram, the first minimum point that appears in the IMF energy spectrum diagram as the frequency increases, and the IMF with a frequency less than the first minimum point is determined as the IMF dominated by the common signal; the IMF dominated by the common signal is removed from the electrical signal to obtain a harmonic signal; the electrical signal, Gaussian noise parameters, analytical signal parameters, the IMF dominated by the common signal, and the harmonic signal are sent to the cloud server; The cloud server is used to store the electrical signal, Gaussian noise parameters, analytical signal parameters, IMF dominated by ordinary signals and the harmonic signal, and to update the Gaussian parameter range corresponding to the electrical signal based on the Gaussian noise parameters.

9. An electronic device, comprising: processor; as well as Memory for storing programs, The program includes instructions, which, when executed by the processor, cause the processor to perform the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to make a computer execute the method according to any one of claims 1-7.

Images