Method for judging category of electric appliance of user based on low-voltage power waveform

Optimize the BP neural network through particle swarm algorithm, extract feature parameters using low-voltage power waveform data, and build an electrical category judgment model, solving the accuracy and robustness of electrical category judgment in the existing technology, and achieving more efficient energy management and Failure prediction.

CN119066486AInactive Publication Date: 2024-12-03BAODING MINYI POWER PURCHASE & SALES SERVICE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411068606.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-12-03
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Among the existing methods for determining user electrical appliance categories, intrusive methods are difficult to install and are not suitable for promotion. However, the non-intrusive methods have large errors in the identification model through the rated power and cannot accurately identify electrical appliance categories, resulting in low efficiency in energy management and fault prediction.

Method used

The BP neural network is optimized by a particle swarm algorithm based on low-voltage power waveform data, extracting time-domain and frequency-domain feature parameters, and generating frequency change vectors and harmonic change vectors, which are used to build an electrical category judgment model and realize the accurate classification of electrical categories.

Benefits of technology

It improves the accuracy of electrical category judgment and the robustness of the model, can more accurately identify electrical appliance types, supports energy management and fault prediction, and reduces maintenance costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119066486A_ABST
    Figure CN119066486A_ABST
Patent Text Reader

Abstract

The invention relates to a method for judging the category of a user electric appliance based on a low-voltage power waveform, and the method comprises the steps: obtaining the low-voltage power waveform data of the user electric appliance, and extracting the feature parameters of the low-voltage power waveform data, the feature parameters comprising a time domain feature and a frequency domain feature; based on the characteristic parameters, obtaining a frequency change vector and a harmonic change vector, inputting the frequency change vector and the harmonic change vector into an electric appliance category judgment model, and obtaining a judgment result output by the electric appliance category judgment model, the training set comprises historical frequency change vectors and harmonic wave change vectors of the user electric appliances and corresponding label information, the label information comprises labeled categories of the electric appliances, and the electric appliance category judgment model is constructed through a BP neural network based on particle swarm optimization. According to the method, the BP neural network is optimized by introducing the particle swarm optimization, and the electric appliances can be accurately classified based on the characteristics in the low-voltage power waveform data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method for determining the category of a user's electrical appliance based on a low-voltage power waveform. Background Art

[0002] Identifying the type of user appliances is important in many ways, especially in smart grids and energy management systems: (1) By identifying the specific type of appliance used by each user, the system can more accurately monitor and manage energy consumption. This refined management helps optimize energy distribution and use, thereby achieving the goals of improving energy efficiency and saving energy and reducing emissions. (2) Identifying the type of user appliances can provide users with personalized energy usage recommendations and optimization plans. (3) Identifying appliance types can also help detect appliance failures or abnormal behavior early. The system can compare the expected appliance operating mode with the actual power consumption data, thereby promptly identifying possible signs of failure and predicting maintenance needs, reducing the risk of power outages and repair costs caused by equipment failures.

[0003] The current method for determining the category of user electrical appliances is to determine the category of electrical appliances through power load monitoring technology. Power load monitoring technology can be divided into two categories: invasive and non-invasive. The invasive method installs sensors with digital communication functions at the interface between each electrical appliance and the power grid to monitor the operating status and power consumption of each electrical appliance. It is necessary to install circuit facilities on each electrical appliance in the electricity user. The installation work requires entering the interior of the electrical appliance, which is difficult to implement and will affect normal production work. It is not suitable for full promotion. The non-invasive method can only identify electrical appliances through the power consumption data of the total electric meter, but in the current method, the model error of rated power identification is large. Therefore, it is urgent to provide a method for determining the category of user electrical appliances based on low-voltage power waveforms. Summary of the invention

[0004] The purpose of the present invention is to provide a method for determining the category of user electrical appliances based on low-voltage power waveforms. By introducing a particle swarm algorithm to optimize the BP neural network, electrical appliances can be accurately classified based on the features in the low-voltage power waveform data.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] The method for determining the type of a user's electrical appliance based on a low-voltage power waveform includes:

[0007] Acquire low-voltage power waveform data of the user's electrical appliances, and extract characteristic parameters of the low-voltage power waveform data, wherein the characteristic parameters include time domain characteristics and frequency domain characteristics;

[0008] Based on the characteristic parameters, a frequency change vector and a harmonic change vector are obtained, and input into an electrical appliance category judgment model to obtain a judgment result output by the electrical appliance category judgment model, wherein the electrical appliance category judgment model is obtained by training with a training set, and the training set includes historical frequency change vectors and harmonic change vectors of user electrical appliances and corresponding label information, and the label information includes the labeled category of the electrical appliance, and the electrical appliance category judgment model is constructed by a BP neural network based on particle swarm optimization.

[0009] Optionally, based on the characteristic parameter, obtaining the frequency change vector and the harmonic wave change vector includes:

[0010] According to a preset window length and a preset sliding length, the time domain feature is divided into a plurality of time domain window features;

[0011] Converting the multiple time domain window features into corresponding frequency domain window features, and sorting the multiple frequency domain window features according to time;

[0012] Determine the similarity between the current frequency domain window feature and the previous adjacent frequency domain window feature;

[0013] The frequency change vector and the harmonic change vector are acquired according to the similarity.

[0014] Optionally, the time domain features include mean, variance, standard deviation, maximum value, minimum value, and the difference between the maximum value and the minimum value.

[0015] Optionally, acquiring the frequency change vector and the harmonic change vector according to the similarity includes:

[0016] When the similarity is less than or equal to a preset similarity, extracting the frequency domain vector and the harmonic wave vector corresponding to the current frequency domain window feature, and extracting the frequency domain vector and the harmonic wave vector corresponding to the previous adjacent frequency domain window feature;

[0017] A frequency change vector is generated based on the difference between the frequency domain vector corresponding to the current frequency domain window feature and the frequency domain vector corresponding to the previous adjacent frequency domain window feature, and a harmonic change vector is generated based on the difference between the harmonic vector corresponding to the current frequency domain window feature and the harmonic vector corresponding to the previous adjacent frequency domain window feature.

[0018] Optionally, obtaining the training set includes:

[0019] Collect historical low-voltage power waveform data of multiple electrical appliances under test during normal use;

[0020] Setting a sliding window to segment the historical low-voltage power waveform data, checking whether the segmented historical low-voltage power waveform data is valid, if so, recording the corresponding electrical appliance category, otherwise deleting the corresponding historical low-voltage power waveform data, and obtaining the filtered historical low-voltage power waveform data;

[0021] Based on the filtered historical low-voltage power waveform data, the frequency change vector and harmonic wave change vector and corresponding label information are obtained.

[0022] Optionally, constructing the electrical appliance category judgment model based on a BP neural network optimized by a particle swarm includes:

[0023] Determine the initial parameters of the BP neural network;

[0024] Iteratively optimize the initial parameters in the BP neural network based on the particle swarm algorithm to obtain the final parameters;

[0025] The BP neural network is constructed based on the final parameters, thus completing the construction of the electrical appliance category judgment model.

[0026] Optionally, iteratively optimizing the initial parameters in the BP neural network based on a particle swarm algorithm to obtain the final parameters includes:

[0027] S1, optimizing the initial parameters in the BP neural network as particles in the particle swarm algorithm to obtain the speed and position of the particles, wherein the initial parameters include the weight and threshold of the BP neural network;

[0028] S2, taking the judgment result as the fitness value, iteratively updating the speed and position of the particles in the particle swarm algorithm;

[0029] S3: Determine whether the iteration end condition is met. If not, return to S1. If so, obtain the final parameter.

[0030] Optionally, obtaining the low-voltage power waveform data of the user's electrical appliances includes:

[0031] The low-voltage power waveform data is transmitted to a host computer via wireless communication or wired communication to obtain a frequency change vector and a harmonic wave change vector; wherein the wireless communication method includes ZigBee, Bluetooth, WiFi, and 433MHz data transmission; and the wired communication method includes 485 bus, CAN bus, Internet, and power carrier.

[0032] The beneficial effects of the present invention are as follows: the present invention obtains the low-voltage power waveform of the user's electrical appliances, and extracts the frequency domain vectors and harmonic wave vectors of the two frequency domain window features based on the low-voltage power waveform data, and uses their changes to input into the trained electrical appliance category judgment model, which not only improves the classification accuracy, but also improves the robustness and flexibility of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0034] Figure 1 The present invention is a flowchart of a method for determining the type of a user's electrical appliance based on a low-voltage power waveform according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0037] like Figure 1 As shown, the present invention discloses a method for determining the category of user electrical appliances based on low-voltage power waveforms, including: obtaining low-voltage power waveform data of the user electrical appliances, extracting characteristic parameters of the low-voltage power waveform data, wherein the characteristic parameters include time domain characteristics and frequency domain characteristics, and the time domain characteristics include mean, variance, standard deviation, maximum value, minimum value, and the difference between the maximum value and the minimum value.

[0038] Based on the characteristic parameters, the frequency change vector and the harmonic change vector are obtained and input into the electrical appliance category judgment model to obtain the judgment result output by the electrical appliance category judgment model, wherein the electrical appliance category judgment model is obtained by training with a training set, and the training set includes the historical frequency change vector and harmonic change vector of the user's electrical appliances and the corresponding label information, and the label information includes the labeled category of the electrical appliance, and the electrical appliance category judgment model is constructed by a BP neural network based on particle swarm optimization.

[0039] In this embodiment, a suitable power waveform sensor is installed on the power supply line or circuit board of the electrical appliance to monitor and record the voltage waveform of the electrical appliance when it is working in real time.

[0040] Further acquisition of low-voltage power waveform data of user electrical appliances includes:

[0041] The low-voltage power waveform data is transmitted to the host computer via wireless communication or wired communication to obtain the frequency change vector and harmonic wave change vector; the wireless communication methods include ZigBee, Bluetooth, WiFi, and 433MHz data transmission methods; the wired communication methods include 485 bus, CAN bus, Internet, and power carrier methods.

[0042] Further, based on the characteristic parameters, obtaining the frequency change vector and the harmonic change vector includes:

[0043] According to a preset window length and a preset sliding length, the time domain feature is divided into a plurality of time domain window features;

[0044] Convert multiple time domain window features into corresponding frequency domain window features, and sort the multiple frequency domain window features according to time;

[0045] Determine the similarity between the current frequency domain window feature and the previous adjacent frequency domain window feature;

[0046] According to the similarity, the frequency change vector and the harmonic wave change vector are obtained.

[0047] Furthermore, according to the similarity, the frequency change vector and the harmonic change vector are obtained, including:

[0048] When the similarity is less than or equal to the preset similarity, the frequency domain vector and the harmonic wave vector corresponding to the current frequency domain window feature are extracted, and the frequency domain vector and the harmonic wave vector corresponding to the previous adjacent frequency domain window feature are extracted;

[0049] A frequency change vector is generated based on the difference between the frequency domain vector corresponding to the current frequency domain window feature and the frequency domain vector corresponding to the previous adjacent frequency domain window feature, and a harmonic change vector is generated based on the difference between the harmonic vector corresponding to the current frequency domain window feature and the harmonic vector corresponding to the previous adjacent frequency domain window feature.

[0050] Specifically, after discretizing each time domain window feature and generating multiple discrete time domain sequences, the discrete time domain sequence can be subjected to discrete Fourier transform, i.e., DFT algorithm. By calculating Fourier transform, filtering, compression, and feature extraction can be performed to generate multiple frequency domain coefficients. The frequency domain coefficients are analyzed and calculated to obtain modulus values ​​and phase angles. The amplitude spectrum and phase spectrum can be obtained through the modulus values ​​and phase angles. Finally, the frequency domain window features are generated according to the frequency domain coefficients, amplitude spectrum, and phase spectrum.

[0051] The horizontal axis of the time domain spectrum of the time domain feature is time, and the vertical axis is the amplitude of the total frequency. In order to more intuitively represent the frequency of each current in the current data, it is necessary to decompose the total frequency of the vertical axis of the time domain feature into separate frequencies for the next step of analysis and calculation; the horizontal axis of the frequency domain spectrum of the frequency domain feature is frequency, and the vertical axis is amplitude; specifically, the Fourier transform algorithm can be used to convert the time domain features into frequency domain features, and sort them in chronological order to facilitate subsequent calculations.

[0052] Furthermore, obtaining a training set includes:

[0053] Collect historical low-voltage power waveform data of multiple electrical appliances under test during normal use;

[0054] Setting a sliding window to segment the historical low-voltage power waveform data, checking whether the segmented historical low-voltage power waveform data is valid, if so, recording the corresponding electrical appliance category, otherwise deleting the corresponding historical low-voltage power waveform data, and obtaining the filtered historical low-voltage power waveform data;

[0055] Based on the filtered historical low-voltage power waveform data, the frequency change vector, the harmonic change vector and the corresponding label information are obtained.

[0056] In this embodiment, the judgment of data validity is specifically that when an abnormal peak value suddenly appears in a single window, the waveform data is considered to be invalid data.

[0057] Furthermore, the BP neural network based on particle swarm optimization constructs an electrical appliance category judgment model including:

[0058] Determine the initial parameters of the BP neural network;

[0059] Iteratively optimize the initial parameters in the BP neural network based on the particle swarm algorithm to obtain the final parameters;

[0060] The BP neural network is constructed based on the final parameters, thus completing the construction of the electrical appliance category judgment model.

[0061] Furthermore, the initial parameters in the BP neural network are iteratively optimized based on the particle swarm algorithm to obtain the final parameters including:

[0062] S1. Optimize the initial parameters in the BP neural network as particles in the particle swarm algorithm to obtain the speed and position of the particles, wherein the initial parameters include the weight and threshold of the BP neural network;

[0063] S2, using the judgment result as the fitness value, iteratively updating the speed and position of the particles in the particle swarm algorithm;

[0064] S3: Determine whether the iteration end condition is met. If not, return to S1. If so, the final parameters are obtained.

[0065] The principle of BP neural network algorithm is: suppose there are s input neurons in the input layer, l hidden neurons in the second hidden layer, and k output neurons in the last output layer. Then, randomly select an input sample from the input layer neurons for training in sequence, and give an expected output value. The formula for calculating the input and output of each neuron in the hidden layer is:

[0066] H ih =∑w ih *x i -b h n i =1, h = 1, 2, 3 ... s

[0067] Among them, H ih To transform the eigenvalue matrix x i Passed to the hidden layer, through the connection weight w between the input layer and the hidden layer ih , and the threshold b of the hidden layer h Another eigenvalue matrix is ​​obtained by transforming the eigenvalue matrix in the hidden layer. Then, the obtained eigenvalue matrix is ​​substituted into and used as the eigenvalue matrix output by the hidden layer and the input eigenvalue matrix of the output layer.

[0068] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for determining the type of a user's electrical appliance based on a low-voltage power waveform, characterized in that: include: Acquire low-voltage power waveform data of the user's electrical appliances, and extract characteristic parameters of the low-voltage power waveform data, wherein the characteristic parameters include time domain characteristics and frequency domain characteristics; Based on the characteristic parameters, a frequency change vector and a harmonic change vector are obtained, and input into an electrical appliance category judgment model to obtain a judgment result output by the electrical appliance category judgment model, wherein the electrical appliance category judgment model is obtained by training with a training set, and the training set includes historical frequency change vectors and harmonic change vectors of user electrical appliances and corresponding label information, and the label information includes the labeled category of the electrical appliance, and the electrical appliance category judgment model is constructed by a BP neural network based on particle swarm optimization.

2. The method for determining the type of user electrical appliances based on low-voltage power waveform according to claim 1, characterized in that: Based on the characteristic parameters, obtaining the frequency change vector and the harmonic wave change vector includes: According to a preset window length and a preset sliding length, the time domain feature is divided into a plurality of time domain window features; Converting the multiple time domain window features into corresponding frequency domain window features, and sorting the multiple frequency domain window features according to time; Determine the similarity between the current frequency domain window feature and the previous adjacent frequency domain window feature; The frequency change vector and the harmonic change vector are acquired according to the similarity.

3. The method for determining the type of user electrical appliances based on low-voltage power waveform according to claim 2, characterized in that: The time domain features include mean, variance, standard deviation, maximum value, minimum value, and the difference between the maximum value and the minimum value.

4. The method for determining the type of user electrical appliances based on low-voltage power waveform according to claim 2, characterized in that: According to the similarity, acquiring the frequency change vector and the harmonic change vector comprises: When the similarity is less than or equal to a preset similarity, extracting the frequency domain vector and the harmonic wave vector corresponding to the current frequency domain window feature, and extracting the frequency domain vector and the harmonic wave vector corresponding to the previous adjacent frequency domain window feature; A frequency change vector is generated based on the difference between the frequency domain vector corresponding to the current frequency domain window feature and the frequency domain vector corresponding to the previous adjacent frequency domain window feature, and a harmonic change vector is generated based on the difference between the harmonic vector corresponding to the current frequency domain window feature and the harmonic vector corresponding to the previous adjacent frequency domain window feature.

5. The method for determining the type of user electrical appliances based on low-voltage power waveform according to claim 1, characterized in that: Acquiring the training set includes: Collect historical low-voltage power waveform data of multiple electrical appliances under test during normal use; Setting a sliding window to segment the historical low-voltage power waveform data, checking whether the segmented historical low-voltage power waveform data is valid, if so, recording the corresponding electrical appliance category, otherwise deleting the corresponding historical low-voltage power waveform data, and obtaining the filtered historical low-voltage power waveform data; Based on the filtered historical low-voltage power waveform data, the frequency change vector and harmonic wave change vector and corresponding label information are obtained.

6. The method for determining the type of user electrical appliances based on low-voltage power waveform according to claim 1, characterized in that: The electrical appliance category judgment model is constructed based on the BP neural network optimized by particle swarm optimization, including: Determine the initial parameters of the BP neural network; Iteratively optimize the initial parameters in the BP neural network based on the particle swarm algorithm to obtain the final parameters; The BP neural network is constructed based on the final parameters, thus completing the construction of the electrical appliance category judgment model.

7. The method for determining the type of user electrical appliances based on low-voltage power waveform according to claim 6, characterized in that: The initial parameters in the BP neural network are iteratively optimized based on the particle swarm algorithm to obtain the final parameters including: S1, optimizing the initial parameters in the BP neural network as particles in the particle swarm algorithm to obtain the speed and position of the particles, wherein the initial parameters include the weight and threshold of the BP neural network; S2, taking the judgment result as the fitness value, iteratively updating the speed and position of the particles in the particle swarm algorithm; S3: Determine whether the iteration end condition is met. If not, return to S1. If so, obtain the final parameter.

8. The method for determining the type of user electrical appliances based on low-voltage power waveform according to claim 1, characterized in that: After obtaining the low-voltage power waveform data of the user's electrical appliances, it includes: The low-voltage power waveform data is transmitted to a host computer via wireless communication or wired communication to obtain a frequency change vector and a harmonic wave change vector; wherein the wireless communication method includes ZigBee, Bluetooth, WiFi, and 433MHz data transmission; and the wired communication method includes 485 bus, CAN bus, Internet, and power carrier.