CN113901973A - Power load identification method and device, storage medium and chip equipment - Google Patents
Power load identification method and device, storage medium and chip equipment Download PDFInfo
- Publication number
- CN113901973A CN113901973A CN202111149243.8A CN202111149243A CN113901973A CN 113901973 A CN113901973 A CN 113901973A CN 202111149243 A CN202111149243 A CN 202111149243A CN 113901973 A CN113901973 A CN 113901973A
- Authority
- CN
- China
- Prior art keywords
- current
- switching event
- vector
- category
- load
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 239000013598 vector Substances 0.000 claims abstract description 178
- 238000013136 deep learning model Methods 0.000 claims abstract description 48
- 238000005070 sampling Methods 0.000 claims description 58
- 238000012549 training Methods 0.000 claims description 36
- 230000008859 change Effects 0.000 claims description 21
- 238000004590 computer program Methods 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000000354 decomposition reaction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000001052 transient effect Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000005406 washing Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Economics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Human Resources & Organizations (AREA)
- Evolutionary Biology (AREA)
- Signal Processing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Bioinformatics & Computational Biology (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
Abstract
The invention discloses a power load identification method, a device, a storage medium and a chip device, wherein the power load identification method comprises the following steps: acquiring a first current signal and a first voltage signal within preset time before and after a switching event of a power network; obtaining a load characteristic vector according to the first current signal and the first voltage signal; and inputting the load characteristic vector into a pre-trained deep learning model to obtain a category characteristic vector, and determining the electric appliance category of the switching event according to the category characteristic vector. According to the power load identification method, load identification is carried out through the load characteristic vector and the pre-trained deep learning model, and the accuracy of load identification can be improved.
Description
Technical Field
The present invention relates to the field of power technologies, and in particular, to a power load identification method and apparatus, a storage medium, and a chip device.
Background
With the rapid development of national economy, the demand of society on electric energy is increasing day by day, and the requirement on electric energy quality is also higher and higher. Therefore, the improvement of the power quality has important significance for safe operation of a power grid and electrical equipment, guarantee of the product quality and normal life of people. In order to improve the quality of electric energy, the electric load needs to be identified so as to know the load composition of the electric power system, grasp the change rule and the development trend of the electric load and scientifically manage the electric load.
At present, load identification methods are mainly classified into invasive methods and non-invasive methods. For a non-intrusive load identification method, a clustering algorithm and the like are proposed in the related art to carry out classification research on the power load. The above techniques, however, have some drawbacks to varying degrees: the clustering algorithm needs to determine the number of categories in advance, and if the number of categories is incorrect, the classification result is prone to be inaccurate.
In addition, with the continuous progress of non-invasive load identification research, some technologies propose to use load transient and steady state information as features for identification analysis. For example: load identification is carried out on disturbance information of voltage when the electric equipment is used for switching, but the method is greatly influenced by voltage fluctuation; for another example, a method for matching the household load by using the closeness of the transient power characteristics when the household load is switched on and off is used, but when the transient characteristics are mixed due to the close turn-on time of a plurality of household appliances, the identification accuracy of the method is greatly influenced.
Disclosure of Invention
The present invention is directed to solving, at least to some extent, one of the technical problems in the related art. Therefore, an object of the present invention is to provide a power load identification method to improve the accuracy of load identification.
A second object of the invention is to propose a computer-readable storage medium.
A third object of the present invention is to provide a chip apparatus.
A fourth object of the present invention is to provide a power load recognition apparatus.
In order to achieve the above object, an embodiment of a first aspect of the present invention provides a power load identification method, where the method includes: acquiring a first current signal and a first voltage signal within preset time before and after a switching event of a power network; obtaining a load characteristic vector according to the first current signal and the first voltage signal; and inputting the load characteristic vector into a pre-trained deep learning model to obtain a category characteristic vector, and determining the electric appliance category of the switching event according to the category characteristic vector.
According to the power load identification method, when a switching event occurs in a power network, load characteristic vectors are obtained according to first current signals and first voltage signals within preset time before and after the switching event, the load characteristic vectors are input into a pre-trained deep learning model, category characteristic vectors are obtained, and the category of an electric appliance of the switching event is determined according to the category characteristic vectors. Therefore, the load identification is carried out through the load characteristic vector and the pre-trained deep learning model, and the accuracy of the load identification can be improved.
In addition, the power load identification method of the embodiment of the invention may further have the following additional technical features:
according to an embodiment of the present invention, the acquiring a first current signal and a first voltage signal within a preset time before and after a switching event of a power network further includes: and acquiring a second current signal at the inlet wire of the power network, and judging whether a switching event occurs in the power network according to the second current signal.
According to an embodiment of the present invention, the determining whether a switching event occurs in the power network according to the second current signal includes: calculating a current change value according to the second current signal; judging whether the current change value is larger than a preset current change threshold value or not; if so, judging that the switching event occurs in the power network.
According to an embodiment of the present invention, after determining the appliance category of the switching event according to the category feature vector, the method further includes: and decomposing the current signal at the inlet wire of the power network according to the electric appliance type.
According to an embodiment of the present invention, the obtaining a load characteristic vector according to the first current signal and the first voltage signal includes: extracting current sampling points within preset time before and after a switching event from the first current signal to obtain a first sampling point sequence and a second sampling point sequence; performing discrete Fourier transform on the first sampling point sequence to obtain a first harmonic component before a switching event occurs, and performing discrete Fourier transform on the second sampling point sequence to obtain a second harmonic component after the switching event occurs; making a difference between the second harmonic component and the first harmonic component to obtain a harmonic component difference vector; calculating the current effective value of each alternating current period within preset time before and after a switching event according to the first current signal to obtain a current effective value vector; calculating the voltage effective value of each alternating current period within preset time before and after a switching event according to the first voltage signal to obtain a voltage effective value vector; and splicing the harmonic component difference vector, the current effective value vector and the voltage effective value vector to obtain the load characteristic vector.
According to one embodiment of the invention, the deep learning model is trained as follows: constructing an initial deep learning model, and acquiring a training set, wherein the training set comprises training samples marked with a plurality of sample labels; predicting the training sample by adopting the initial deep learning model to obtain a class characteristic vector of the training sample; and acquiring a cross entropy loss function corresponding to the sample label, and converging the category characteristic vector of the training sample and the sample label according to the cross entropy loss function to obtain the pre-trained deep learning model.
According to an embodiment of the present invention, the determining the appliance category of the switching event according to the category feature vector includes: acquiring reference vectors of a plurality of electrical appliances; respectively calculating Euclidean distances between the category feature vectors and the reference vectors as confidence degrees; and determining the electric appliance type of the switching event according to the confidence coefficient.
According to an embodiment of the present invention, the obtaining the reference vectors of the plurality of appliances includes: and when the pre-trained deep learning model is used, respectively inputting the load characteristic vectors corresponding to the switching events of the electrical appliances into the pre-trained deep learning model, and outputting the category characteristic vectors of the electrical appliances as the reference vectors of the electrical appliances.
According to an embodiment of the invention, the decomposing the current signal according to the appliance category comprises: obtaining current variation according to the current signal; and superposing the current variation quantity to the current initial waveform of the electric appliance corresponding to the electric appliance type.
In order to achieve the above object, a second embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to implement the above power load identification method.
In order to achieve the above object, a chip device according to a third embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory, where the computer program is executed by the processor to implement the above power load identification method.
In order to achieve the above object, a fourth aspect of the present invention provides a power load identification apparatus, including: the acquisition module is used for acquiring a first current signal and a first voltage signal within preset time before and after a switching event of the power network; the calculation module is used for obtaining a load characteristic vector according to the first current signal and the first voltage signal; and the determining module is used for inputting the load characteristic vector to a pre-trained deep learning model to obtain a category characteristic vector, and determining the electric appliance category of the switching event according to the category characteristic vector.
According to the power load identification device provided by the embodiment of the invention, when the switching event of the power network is judged according to the current signal, the load characteristic vector is obtained according to the first voltage signal of the first current signal within the preset time before and after the switching event, the load characteristic vector is further input into the pre-trained deep learning model, the category characteristic vector is obtained, and the electric appliance category of the switching event is determined according to the category characteristic vector. Therefore, the load identification is carried out through the load characteristic vector and the pre-trained deep learning model, and the accuracy of the load identification can be improved.
In addition, the power load identification device of the embodiment of the invention may further have the following additional technical features:
according to an embodiment of the present invention, the calculation module is specifically configured to: extracting current sampling points within preset time before and after a switching event from the first current signal to obtain a first sampling point sequence and a second sampling point sequence; performing discrete Fourier transform on the first sampling point sequence to obtain a first harmonic component before a switching event occurs, and performing discrete Fourier transform on the second sampling point sequence to obtain a second harmonic component after the switching event occurs; making a difference between the second harmonic component and the first harmonic component to obtain a harmonic component difference vector; calculating the current effective value of each alternating current period within preset time before and after a switching event according to the first current signal to obtain a current effective value vector; calculating the voltage effective value of each alternating current period within preset time before and after a switching event according to the first voltage signal to obtain a voltage effective value vector; and splicing the harmonic component difference vector, the current effective value vector and the voltage effective value vector to obtain the load characteristic vector.
According to an embodiment of the present invention, when determining the category of the electrical appliance of the switching event according to the category eigenvector, the determining module is specifically configured to: acquiring reference vectors of a plurality of electrical appliances; respectively calculating Euclidean distances between the category feature vectors and the reference vectors as confidence degrees; and determining the electric appliance type of the switching event according to the confidence coefficient.
According to an embodiment of the present invention, when the determining module obtains the reference vectors of the plurality of electrical appliances, the determining module is specifically configured to: and when the pre-trained deep learning model is used, respectively inputting the load characteristic vectors corresponding to the switching events of the electrical appliances into the pre-trained deep learning model, and outputting the category characteristic vectors of the electrical appliances as the reference vectors of the electrical appliances.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
FIG. 1 is a flow chart of a power load identification method of one embodiment of the present invention;
FIG. 2 is a flow chart of a power load identification method of another embodiment of the present invention;
fig. 3 is a block diagram showing the structure of a power load recognition apparatus according to an embodiment of the present invention;
fig. 4 is a block diagram of a power load recognition apparatus according to another embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
Fig. 1 is a flowchart of a power load identification method according to an embodiment of the present invention.
As shown in fig. 1, the power load identification method includes the steps of:
s1, acquiring a first current signal and a first voltage signal within a preset time before and after the switching event of the power network.
Wherein the current signal and the voltage signal may be collected at an incoming line of the power network.
In an embodiment of the invention, the current signal and the voltage signal at the inlet wire of the power network can be collected in real time, the current signal acquired before the switching event of the power network occurs is recorded as a second current signal, and whether the switching event occurs in the power network can be judged according to the second current signal.
Specifically, the voltage input at the inlet of the power network may be an alternating current with a fundamental frequency of 50Hz, and the sampling frequency of the current and the voltage may be above 1KHz, for example, 6 KHz. Therefore, the voltage and the current change condition of the inlet wire of the power network can be accurately reflected by the current and the voltage obtained by sampling through high-frequency sampling, and the identification accuracy is further improved to a certain extent.
As a possible implementation manner, the determining whether the switching event occurs in the power network according to the second current signal may include: calculating a current change value according to the second current signal; judging whether the current change value is larger than a preset current change threshold value or not; if yes, whether a switching event occurs in the power network is judged. The preset current change threshold value can be calibrated according to the load condition of the actual power network.
Specifically, the difference between any two adjacent sampling currents can be calculated, and when the difference is greater than a preset current change threshold, a switching event of the power network can be determined. In order to ensure the reliability of the judgment, when a plurality of continuous difference values are all larger than a preset current change threshold value, the switching event of the power network can be judged.
And S2, obtaining a load characteristic vector according to the first current signal and the first voltage signal.
The load characteristic vector is obtained through a first current signal and a first voltage signal within a preset time (the value range can be 0.3-1 second, such as 0.5 second) before and after a switching event occurs in a power network, and the load characteristic vector can be well ensured to be obtained through voltage and current fluctuation caused by the starting of a single electric appliance through the setting of a short preset time; by considering both voltage and current, the identification can be made more accurate than if only voltage is considered.
As a feasible implementation manner, obtaining the load characteristic vector according to the current signal and the voltage signal within the preset time before and after the switching event may include the following steps:
a1, extracting current sampling points in a preset time before and after a switching event from the first current signal to obtain a first sampling point sequence and a second sampling point sequence;
for example, the preset time is 0.5 seconds. After the switching event is determined to occur, the occurrence time of the switching event is recorded as t1, a current signal 0.5 second before the time of t1 and a current signal 0.5 second after the time of t1 are further obtained, the maximum value of the sampling current of each alternating current period (namely 0.02 second) in the range of 0.5 second before the time of t1 is extracted, each maximum value is used as a current sampling point to obtain a first sampling point sequence, meanwhile, the maximum value of the sampling current of each alternating current period (namely 0.02 second) in the range of 0.5 second after the time of t1 is extracted, each maximum value is used as a current sampling point to obtain a second sampling point sequence. The first sampling point sequence and the second sampling point sequence comprise 25 sampling points.
B1, performing discrete Fourier transform on the first sampling point sequence to obtain first harmonic components before and after the switching event, and performing discrete Fourier transform on the second sampling point sequence to obtain second harmonic components after the switching event;
specifically, the sequence of sampling points may be discrete fourier transformed by the following equation:
where x (k) is the data after discrete fourier transform, i.e. the harmonic component, and x (n) is the nth sample point in the sequence of sample points.
C1, making a difference between the second harmonic component and the first harmonic component to obtain a harmonic component difference vector;
d1, calculating current effective values of each alternating current period within preset time before and after a switching event according to the first current signal to obtain a current effective value vector;
e1, calculating voltage effective values of all alternating current periods within preset time before and after a switching event according to the first voltage signal to obtain voltage effective value vectors;
for example, the preset time is 0.5 seconds. After the switching event is determined to occur, the occurrence time of the switching event is recorded as t1, a current signal 0.5 second before the time of t1 and a current signal and a voltage signal 0.5 second after the time of t1 are further obtained, a current effective value and a voltage effective value of each alternating current period (namely 0.02 second) within a range of 0.5 second before the time of t1 are calculated, and a current effective value and a voltage effective value of each alternating current period (namely 0.02 second) within a range of 0.5 second after the time of t1 are calculated at the same time, so that a current effective value vector containing 50 current effective values and a voltage effective value vector containing 50 voltage effective values are obtained.
F1, splicing the harmonic component difference vector, the current effective value vector and the voltage effective value vector to obtain a load characteristic vector.
Specifically, the harmonic component difference vector, the current effective value vector, and the voltage effective value vector may be spliced in a head-to-tail connection manner to obtain a one-dimensional load characteristic vector. The splicing sequence among the harmonic component difference vector, the current effective value vector and the voltage effective value vector can be changed, for example, the harmonic component difference vector, the voltage effective value vector and the current effective value vector can be spliced in a head-to-tail connection mode.
And S3, inputting the load characteristic vector into a pre-trained deep learning model to obtain a category characteristic vector, and determining the electric appliance category of the switching event according to the category characteristic vector.
In the embodiment, the power load identification is that how the electric energy of the power network is utilized and consumed by the electric appliances is monitored, namely, the power consumption monitoring of the power network, so that the electric appliances corresponding to the switching events are obtained through identification, the power load identification can be realized, and further, the power-saving regulation and control measures and the more efficient power consumption mode aiming at the power network are favorably found.
As one example, the training process for the deep learning model may be as follows:
a2, constructing an initial deep learning model, and acquiring a training set, wherein the training set comprises training samples marked with a plurality of sample labels;
alternatively, the initial depth model may be a 5-layer one-dimensional convolution model.
The training samples are obtained in the same way as the load feature vectors. The sample label is an electrical appliance type corresponding to the training sample, for example, the training sample a, and the label is an air conditioner (which may be denoted by reference numeral 1); training sample B, labeled refrigerator (may be denoted by reference numeral 2); training sample C, labeled washing machine (may be referred to by reference numeral 3).
B2, predicting the training samples by adopting an initial deep learning model to obtain class characteristic vectors of the training samples;
and C2, acquiring a cross entropy loss function corresponding to the sample label, and converging the class characteristic vector of the training sample and the sample label according to the cross entropy loss function to obtain a pre-trained deep learning model.
As a possible implementation manner, the determining the appliance category of the switching event according to the category feature vector in step S3 may include: acquiring reference vectors of a plurality of electrical appliances; respectively calculating Euclidean distances between the category feature vectors and the reference vectors as confidence coefficients; and determining the electric appliance type of the switching event according to the confidence coefficient. For example: the confidence of the minimum Euclidean distance can be set to be maximum, and the electric appliance category to which the reference vector corresponding to the maximum confidence belongs is used as the electric appliance category of the switching event.
Wherein, obtaining the reference vectors of the plurality of electrical appliances may include: when the pre-trained deep learning model is used, the load characteristic vectors corresponding to the switching events of the electrical appliances are respectively input into the pre-trained deep learning model, and the category characteristic vectors of the electrical appliances are output to serve as the reference vectors of the electrical appliances.
Specifically, after the deep learning model is trained, the load characteristic vector corresponding to the switching event corresponding to the electrical appliance of which the electrical appliance category is known is input into the deep learning model, so as to obtain the electrical appliance category characteristic vector, which is used as the reference vector of the electrical appliance. And executing the operation on each possible electric appliance to obtain the reference vectors of the plurality of electric appliances.
In an embodiment of the present invention, as shown in fig. 2, after determining the appliance category of the switching event according to the category feature vector, the power load identification method may further include:
and S4, decomposing the current signal at the inlet wire of the power network according to the electric appliance type.
In this embodiment, when the power load is identified in the power supply process of the power network, one or more switching events may occur, so that the current signal at the incoming line of the power network may be continuously obtained, and the current signal to be decomposed is the continuously obtained current signal, which may include a second current signal for determining the switching event. The current signals at the inlet wire of the power network are decomposed according to the type of the electric appliances, the current signals at the inlet wire of the power network which are continuously acquired can be decomposed into the working current of each electric appliance, and then the working condition of each electric appliance can be displayed through the waveform, so that a user can conveniently know the power utilization condition of the power network.
Specifically, decomposing the current signal at the incoming line of the power network according to the appliance category may include: obtaining current variation according to a current signal at an inlet wire of the power network; and superposing the current variation quantity on the current initial waveform of the electric appliance corresponding to the electric appliance type.
Specifically, the current initial value in the current initial waveforms of all the electrical appliances is set to be 0, and the switching event is caused by which electrical appliance, the corresponding current variation is superimposed on the current initial waveform of the corresponding electrical appliance, so that the total electric meter waveform (i.e., the waveform of the current signal) is decomposed into the current waveforms of various electrical appliances for viewing. Wherein, the current variation may be a current effective value variation, and the current waveform may be a current effective value waveform.
In summary, the power load identification method of the embodiment of the invention can improve the accuracy of load identification, realize real-time online identification of the load, and only input the total data of the power consumption parameters into the deep learning model without manually selecting and extracting various required load characteristics, so that the method has higher user friendliness and usability, and can be widely applied to identification of the power load of residents.
Based on the power load identification method of the above embodiment, the present invention provides a computer-readable storage medium.
In this embodiment, a computer-readable storage medium has stored thereon a computer program which, when executed by a processor, implements the power load identification method of the above-described embodiment.
The computer readable storage medium of the embodiment of the invention, when the computer program stored thereon and corresponding to the power load identification method is executed by the processor, can improve the accuracy of load identification, realize real-time online identification of the load, and only need to input the total data of the power consumption parameters into the deep learning model without depending on manual selection and extraction of various required load characteristics, so that the invention has higher user friendliness and usability, and can be widely applied to identification of the power load of residents.
Based on the power load identification method of the embodiment, the invention further provides chip equipment.
In this embodiment, the chip device includes a memory, a processor, and a computer program stored on the memory, and is characterized in that the computer program implements the power load identification method of the above-described embodiment when executed by the processor.
The chip device of the embodiment of the invention can improve the accuracy of load identification and realize the real-time online identification of the load when the computer program corresponding to the power load identification method stored on the memory of the chip device is executed by the processor, and only the total data of the power consumption parameters need to be input into the deep learning model without manually selecting and extracting various required load characteristics, so that the invention has higher user friendliness and usability, and can be widely applied to the identification of the power load of residents.
Fig. 3 is a block diagram of a power load recognition apparatus according to an embodiment of the present invention.
As shown in fig. 3, the power load recognition apparatus 100 includes: an acquisition module 110, a calculation module 130, and a determination module 140.
In this embodiment, the obtaining module 110 is configured to obtain a first current signal and a first voltage signal within a preset time before and after a switching event occurs in the power network.
In an embodiment of the present invention, the obtaining module 110 may be configured to obtain a current signal and a voltage signal at an incoming line of the power network, and record the current signal obtained before the switching event as the second current signal. The obtaining module 110 may determine whether a switching event occurs in the power network according to the second current signal.
Specifically, the voltage input at the inlet of the power network may be an alternating current with a fundamental frequency of 50Hz, and the sampling frequency of the current and the voltage may be above 1KHz, for example, 6 KHz. Therefore, the voltage and the current change condition of the inlet wire of the power network can be accurately reflected by the current and the voltage obtained by sampling through high-frequency sampling, and the identification accuracy is further improved to a certain extent.
Further, the obtaining module 110 may be configured to: calculating a current change value according to the second current signal; judging whether the current change value is larger than a preset current change threshold value or not; if so, judging that the switching event occurs in the power network. The preset current change threshold value can be calibrated according to the load condition of the actual power network.
Specifically, the difference between any two adjacent sampling currents can be calculated, and when the difference is greater than a preset current change threshold, a switching event of the power network can be determined. In order to ensure the reliability of the judgment, when a plurality of continuous difference values are all larger than a preset current change threshold value, the switching event of the power network can be judged.
The calculating module 130 is configured to obtain a load feature vector according to the first current signal and the first voltage signal.
The preset time can be 0.3-1 second, such as 0.5 second.
As a possible implementation, the calculation module 130 may specifically perform the following steps:
a1, extracting current sampling points in a preset time before and after a switching event from the first current signal to obtain a first sampling point sequence and a second sampling point sequence;
for example, the preset time is 0.5 seconds. After the switching event is determined to occur, the occurrence time of the switching event is recorded as t1, a current signal 0.5 second before the time of t1 and a current signal 0.5 second after the time of t1 are further obtained, the maximum value of the sampling current of each alternating current period (namely 0.02 second) in the range of 0.5 second before the time of t1 is extracted, each maximum value is used as a current sampling point to obtain a first sampling point sequence, meanwhile, the maximum value of the sampling current of each alternating current period (namely 0.02 second) in the range of 0.5 second after the time of t1 is extracted, each maximum value is used as a current sampling point to obtain a second sampling point sequence. The first sampling point sequence and the second sampling point sequence comprise 25 sampling points.
B1, performing discrete Fourier transform on the first sampling point sequence to obtain first harmonic components before and after the switching event, and performing discrete Fourier transform on the second sampling point sequence to obtain second harmonic components after the switching event;
specifically, the sequence of sampling points may be discrete fourier transformed by the following equation:
where x (k) is the data after discrete fourier transform, i.e. the harmonic component, and x (n) is the nth sample point in the sequence of sample points.
C1, making a difference between the second harmonic component and the first harmonic component to obtain a harmonic component difference vector;
d1, calculating current effective values of each alternating current period within preset time before and after a switching event according to the first current signal to obtain a current effective value vector;
e1, calculating voltage effective values of all alternating current periods within preset time before and after a switching event according to the first voltage signal to obtain voltage effective value vectors;
for example, the preset time is 0.5 seconds. After the switching event is determined to occur, the occurrence time of the switching event is recorded as t1, a current signal 0.5 second before the time of t1 and a current signal and a voltage signal 0.5 second after the time of t1 are further obtained, a current effective value and a voltage effective value of each alternating current period (namely 0.02 second) within a range of 0.5 second before the time of t1 are calculated, and a current effective value and a voltage effective value of each alternating current period (namely 0.02 second) within a range of 0.5 second after the time of t1 are calculated at the same time, so that a current effective value vector containing 50 current effective values and a voltage effective value vector containing 50 voltage effective values are obtained.
F1, splicing the harmonic component difference vector, the current effective value vector and the voltage effective value vector to obtain a load characteristic vector.
Specifically, the harmonic component difference vector, the current effective value vector, and the voltage effective value vector may be spliced in a head-to-tail connection manner to obtain a one-dimensional load characteristic vector. The splicing sequence among the harmonic component difference vector, the current effective value vector and the voltage effective value vector can be changed, for example, the harmonic component difference vector, the voltage effective value vector and the current effective value vector can be spliced in a head-to-tail connection mode.
The determining module 140 is configured to input the load feature vector to a pre-trained deep learning model to obtain a category feature vector, and determine an appliance category of the switching event according to the category feature vector.
As one example, the training process for the deep learning model may be as follows:
a2, constructing an initial deep learning model, and acquiring a training set, wherein the training set comprises training samples marked with a plurality of sample labels;
alternatively, the initial depth model may be a 5-layer one-dimensional convolution model.
The training samples are obtained in the same way as the load feature vectors. The sample label is an electrical appliance type corresponding to the training sample, for example, the training sample a, and the label is an air conditioner (which may be denoted by reference numeral 1); training sample B, labeled refrigerator (may be denoted by reference numeral 2); training sample C, labeled washing machine (may be referred to by reference numeral 3).
B2, predicting the training samples by adopting an initial deep learning model to obtain class characteristic vectors of the training samples;
and C2, acquiring a cross entropy loss function corresponding to the sample label, and converging the class characteristic vector of the training sample and the sample label according to the cross entropy loss function to obtain a pre-trained deep learning model.
As a possible implementation manner, when determining the appliance category of the switching event according to the category feature vector, the determining module 140 may be specifically configured to: acquiring reference vectors of a plurality of electrical appliances; respectively calculating Euclidean distances between the category feature vectors and the reference vectors as confidence coefficients; and determining the electric appliance type of the switching event according to the confidence coefficient. For example: the confidence of the minimum Euclidean distance can be set to be maximum, and the electric appliance category to which the reference vector corresponding to the maximum confidence belongs is used as the electric appliance category of the switching event.
Wherein, obtaining the reference vectors of the plurality of electrical appliances may include: when the pre-trained deep learning model is used, the load characteristic vectors corresponding to the switching events of the electrical appliances are respectively input into the pre-trained deep learning model, and the category characteristic vectors of the electrical appliances are output to serve as the reference vectors of the electrical appliances.
Specifically, after the deep learning model is trained, the load characteristic vector corresponding to the switching event corresponding to the electrical appliance of which the electrical appliance category is known is input into the deep learning model, so as to obtain the electrical appliance category characteristic vector, which is used as the reference vector of the electrical appliance. And executing the operation on each possible electric appliance to obtain the reference vectors of the plurality of electric appliances.
In an embodiment of the present invention, as shown in fig. 4, the power load recognition apparatus 100 may further include: and the decomposition module 150, wherein the decomposition module 150 is used for decomposing the current signal according to the appliance type.
In this embodiment, the decomposition module 150 may be specifically configured to: obtaining current variation according to the current signal; and superposing the current variation quantity on the current initial waveform of the electric appliance corresponding to the electric appliance type.
Specifically, the current initial value in the current initial waveforms of all the electrical appliances is set to be 0, and the switching event is caused by which electrical appliance, the corresponding current variation is superimposed on the current initial waveform of the corresponding electrical appliance, so that the total electric meter waveform (i.e., the waveform of the current signal) is decomposed into the current waveforms of various electrical appliances for viewing. Wherein, the current variation may be a current effective value variation, and the current waveform may be a current effective value waveform.
In summary, the power load recognition device of the embodiment of the invention can improve the accuracy of load recognition, realize real-time online recognition of the load, and only input the total data of the power consumption parameters into the deep learning model without manually selecting and extracting various required load characteristics, so that the invention has higher user friendliness and usability, and can be widely applied to identification of the power load of residents.
It should be noted that the logic and/or steps represented in the flowcharts or otherwise described herein, such as an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
It should be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
In the description of the present invention, it is to be understood that the terms "central," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," "circumferential," and the like are used in the orientations and positional relationships indicated in the drawings for convenience in describing the invention and to simplify the description, and are not intended to indicate or imply that the referenced devices or elements must have a particular orientation, be constructed and operated in a particular orientation, and are therefore not to be considered limiting of the invention.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
In the present invention, unless otherwise expressly stated or limited, the terms "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; they may be directly connected or indirectly connected through intervening media, or they may be connected internally or in any other suitable relationship, unless expressly stated otherwise. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
In the present invention, unless otherwise expressly stated or limited, the first feature "on" or "under" the second feature may be directly contacting the first and second features or indirectly contacting the first and second features through an intermediate. Also, a first feature "on," "over," and "above" a second feature may be directly or diagonally above the second feature, or may simply indicate that the first feature is at a higher level than the second feature. A first feature being "under," "below," and "beneath" a second feature may be directly under or obliquely under the first feature, or may simply mean that the first feature is at a lesser elevation than the second feature.
Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
Claims (15)
1. A power load identification method, characterized in that the method comprises the steps of:
acquiring a first current signal and a first voltage signal within preset time before and after a switching event of a power network;
obtaining a load characteristic vector according to the first current signal and the first voltage signal;
and inputting the load characteristic vector into a pre-trained deep learning model to obtain a category characteristic vector, and determining the electric appliance category of the switching event according to the category characteristic vector.
2. The power load identification method according to claim 1, wherein the acquiring the first current signal and the first voltage signal within a preset time before and after the switching event of the power network further comprises:
and acquiring a second current signal at the inlet wire of the power network, and judging whether a switching event occurs in the power network according to the second current signal.
3. The power load identification method according to claim 2, wherein the determining whether a switching event occurs in the power network according to the second current signal comprises:
calculating a current change value according to the second current signal;
judging whether the current change value is larger than a preset current change threshold value or not;
if so, judging that the switching event occurs in the power network.
4. The power load identification method according to claim 1, wherein after determining the appliance category of the switching event from the category eigenvector, the method further comprises:
and decomposing the current signal at the inlet wire of the power network according to the electric appliance type.
5. The electrical load identification method of claim 1, wherein said deriving a load signature vector from said first current signal and said first voltage signal comprises:
extracting current sampling points within preset time before and after a switching event from the first current signal to obtain a first sampling point sequence and a second sampling point sequence;
performing discrete Fourier transform on the first sampling point sequence to obtain a first harmonic component before a switching event occurs, and performing discrete Fourier transform on the second sampling point sequence to obtain a second harmonic component after the switching event occurs;
making a difference between the second harmonic component and the first harmonic component to obtain a harmonic component difference vector;
calculating the current effective value of each alternating current period within preset time before and after a switching event according to the first current signal to obtain a current effective value vector;
calculating the voltage effective value of each alternating current period within preset time before and after a switching event according to the first voltage signal to obtain a voltage effective value vector;
and splicing the harmonic component difference vector, the current effective value vector and the voltage effective value vector to obtain the load characteristic vector.
6. The power load identification method according to claim 1, wherein the deep learning model is trained as follows:
constructing an initial deep learning model, and acquiring a training set, wherein the training set comprises training samples marked with a plurality of sample labels;
predicting the training sample by adopting the initial deep learning model to obtain a class characteristic vector of the training sample;
and acquiring a cross entropy loss function corresponding to the sample label, and converging the category characteristic vector of the training sample and the sample label according to the cross entropy loss function to obtain the pre-trained deep learning model.
7. The power load identification method according to claim 1, wherein the determining the appliance category of the switching event according to the category feature vector comprises:
acquiring reference vectors of a plurality of electrical appliances;
respectively calculating Euclidean distances between the category feature vectors and the reference vectors as confidence degrees;
and determining the electric appliance type of the switching event according to the confidence coefficient.
8. The power load identification method according to claim 7, wherein the acquiring reference vectors of a plurality of electrical appliances comprises:
and when the pre-trained deep learning model is used, respectively inputting the load characteristic vectors corresponding to the switching events of the electrical appliances into the pre-trained deep learning model, and outputting the category characteristic vectors of the electrical appliances as the reference vectors of the electrical appliances.
9. The electrical load identification method of claim 4, wherein said decomposing the current signal at the incoming line of the electrical power network according to the appliance category comprises:
obtaining current variation according to a current signal at an inlet wire of the power network;
and superposing the current variation quantity to the current initial waveform of the electric appliance corresponding to the electric appliance type.
10. A computer-readable storage medium, on which a computer program is stored, which, when executed by a processor, implements a power load identification method according to any one of claims 1 to 9.
11. A chip device comprising a memory, a processor and a computer program stored on the memory, wherein the computer program, when executed by the processor, implements a power load identification method as claimed in any one of claims 1 to 9.
12. An electrical load recognition apparatus, the apparatus comprising:
the acquisition module is used for acquiring a first current signal and a first voltage signal within preset time before and after a switching event of the power network;
the calculation module is used for obtaining a load characteristic vector according to the first current signal and the first voltage signal;
and the determining module is used for inputting the load characteristic vector to a pre-trained deep learning model to obtain a category characteristic vector, and determining the electric appliance category of the switching event according to the category characteristic vector.
13. The electrical load recognition device of claim 12, wherein the computing module is specifically configured to:
extracting current sampling points within preset time before and after a switching event from the first current signal to obtain a first sampling point sequence and a second sampling point sequence;
performing discrete Fourier transform on the first sampling point sequence to obtain a first harmonic component before a switching event occurs, and performing discrete Fourier transform on the second sampling point sequence to obtain a second harmonic component after the switching event occurs;
making a difference between the second harmonic component and the first harmonic component to obtain a harmonic component difference vector;
calculating the current effective value of each alternating current period within preset time before and after a switching event according to the first current signal to obtain a current effective value vector;
calculating the voltage effective value of each alternating current period within preset time before and after a switching event according to the first voltage signal to obtain a voltage effective value vector;
and splicing the harmonic component difference vector, the current effective value vector and the voltage effective value vector to obtain the load characteristic vector.
14. The electrical load recognition device according to claim 12, wherein the determination module, when determining the appliance category of the switching event according to the category eigenvector, is specifically configured to:
acquiring reference vectors of a plurality of electrical appliances;
respectively calculating Euclidean distances between the category feature vectors and the reference vectors as confidence degrees;
and determining the electric appliance type of the switching event according to the confidence coefficient.
15. The electrical load recognition device of claim 14, wherein the determination module, when obtaining the reference vectors for the plurality of electrical appliances, is specifically configured to:
and when the pre-trained deep learning model is used, respectively inputting the load characteristic vectors corresponding to the switching events of the electrical appliances into the pre-trained deep learning model, and outputting the category characteristic vectors of the electrical appliances as the reference vectors of the electrical appliances.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111149243.8A CN113901973A (en) | 2021-09-29 | 2021-09-29 | Power load identification method and device, storage medium and chip equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111149243.8A CN113901973A (en) | 2021-09-29 | 2021-09-29 | Power load identification method and device, storage medium and chip equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113901973A true CN113901973A (en) | 2022-01-07 |
Family
ID=79189313
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111149243.8A Pending CN113901973A (en) | 2021-09-29 | 2021-09-29 | Power load identification method and device, storage medium and chip equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113901973A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114330461A (en) * | 2022-03-16 | 2022-04-12 | 北京智芯微电子科技有限公司 | V-I track generation method and device for non-invasive load identification and neural network |
| CN115018011A (en) * | 2022-07-19 | 2022-09-06 | 深圳江行联加智能科技有限公司 | Power load type identification method, device, equipment and storage medium |
-
2021
- 2021-09-29 CN CN202111149243.8A patent/CN113901973A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114330461A (en) * | 2022-03-16 | 2022-04-12 | 北京智芯微电子科技有限公司 | V-I track generation method and device for non-invasive load identification and neural network |
| CN115018011A (en) * | 2022-07-19 | 2022-09-06 | 深圳江行联加智能科技有限公司 | Power load type identification method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108021736B (en) | Load switching action monitoring method based on sliding window residual error model | |
| CN105067966B (en) | The low-voltage alternating-current fault arc detection method of feature based modal components energy spectrometer | |
| CN107025365B (en) | A kind of non-intruding load discrimination method for user side | |
| CN111985824A (en) | Non-invasive load monitoring method and monitoring equipment for intelligent ammeter box | |
| CN113901973A (en) | Power load identification method and device, storage medium and chip equipment | |
| Lee | Electric load information system based on non-intrusive power monitoring | |
| CN106872824A (en) | A kind of network load appliance type identification and the method and its device of different electrical equipment electricity statistics | |
| CN109387712A (en) | Non-intrusion type cutting load testing and decomposition method based on state matrix decision tree | |
| JP5729162B2 (en) | Power management equipment | |
| CN110416995B (en) | Non-invasive load decomposition method and device | |
| CN109447473B (en) | Power load monitoring method, device, equipment and readable storage medium | |
| Ghosh et al. | An improved load feature extraction technique for smart Homes using fuzzy-based NILM | |
| JP5328858B2 (en) | Operating status determination device, operating status determination program, operating status determination method, waveform pattern learning device, waveform pattern learning program, and waveform pattern learning method | |
| JP2004056969A (en) | Method for estimating operating state of electric device with frequently changing power consumption and monitor system of the same device | |
| CN104036111A (en) | Methods and systems for evaluating and diagnosing energy efficiency of energy consuming equipment | |
| EP4174505A1 (en) | Battery state determination method, and battery state determination apparatus | |
| Girmay et al. | Simple event detection and disaggregation approach for residential energy estimation | |
| CN111178393A (en) | Electric appliance power consumption classification metering method and device based on intelligent electric meter | |
| US10942205B2 (en) | Method and system for analyzing electricity consumption | |
| CN111639586A (en) | Non-invasive load identification model construction method, load identification method and system | |
| US20160109492A1 (en) | Method and system for analysing electricity consumption | |
| CN116861316B (en) | Electrical appliance monitoring method and device | |
| CN109377052B (en) | Power load matching method, device, equipment and readable storage medium | |
| Welikala et al. | A real-time non-intrusive load monitoring system | |
| de Souza et al. | An effective CPT-based nonintrusive load monitoring for cognitive meters |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |