CN114239762A - Non-invasive load identification method and system based on structured load characteristic spectrum - Google Patents
Non-invasive load identification method and system based on structured load characteristic spectrum Download PDFInfo
- Publication number
- CN114239762A CN114239762A CN202210183198.6A CN202210183198A CN114239762A CN 114239762 A CN114239762 A CN 114239762A CN 202210183198 A CN202210183198 A CN 202210183198A CN 114239762 A CN114239762 A CN 114239762A
- Authority
- CN
- China
- Prior art keywords
- load
- attribute
- determining
- svm
- classifier
- 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 38
- 238000001228 spectrum Methods 0.000 title abstract description 9
- 230000004927 fusion Effects 0.000 claims abstract description 17
- 230000007246 mechanism Effects 0.000 claims abstract description 8
- 238000010606 normalization Methods 0.000 claims description 18
- 238000005070 sampling Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000005406 washing Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 5
- 230000005611 electricity Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 240000002853 Nelumbo nucifera Species 0.000 description 1
- 235000006508 Nelumbo nucifera Nutrition 0.000 description 1
- 235000006510 Nelumbo pentapetala Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
-
- 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
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2310/00—The network for supplying or distributing electric power characterised by its spatial reach or by the load
- H02J2310/70—Load identification
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Public Health (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Power Engineering (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a non-invasive load identification method and a non-invasive load identification system based on a structured load characteristic spectrum, and relates to the field of non-invasive load identification. The method comprises the steps of performing feature extraction and feature fusion on a running signal of the load to determine an optimized load attribute set; determining the weight of each attribute characteristic according to the optimized load attribute set and the information entropy redundancy between each load attribute characteristic; determining a structured load characteristic map according to the weight of each attribute characteristic; constructing an SVM load classifier according to the optimized load attribute set and the corresponding load category, optimizing parameters of the SVM load classifier according to the structured load feature map, and determining an error set of the SVM load classifier; determining the number of load classifiers according to the SVM load classifier error set, and constructing a combined SVM load classifier; and (3) carrying out load identification by adopting a combined SVM load classifier, and determining the load category by a voting group decision mechanism. The invention can improve the accuracy of load identification of different users.
Description
Technical Field
The invention relates to the field of non-invasive load identification, in particular to a non-invasive load identification method and system based on a structured load characteristic spectrum.
Background
Electric energy is one of the most central energy sources in China, and the demand for the electric energy is continuously increased along with the improvement of the living standard of people in recent years. On the premise of limited electric energy output, saving electricity and reasonably utilizing electric energy are one of the most effective means for improving the utilization rate of electric energy at present. The key to improving the utilization rate of electric energy is the electricity management and energy efficiency optimization of the demand side. Load monitoring is a key technology of power utilization management, and is beneficial to accurately analyzing the power grid energy consumption level, so that the optimization and adjustment of an industrial structure are realized.
The traditional load monitoring is mainly to estimate the energy efficiency level of power grid operation according to the line loss rate of the power grid, although the index can reflect the energy efficiency level to a certain extent, the index can only give a final result but cannot describe process quantities such as power consumption information of a demand side in detail, and invasive detection is required in the data acquisition process, so that the protection of user information is not facilitated. Therefore, it is necessary to provide a method which does not interfere with the power users, has universality, and can accurately describe the switching situation of the demand side load.
Disclosure of Invention
The invention aims to provide a non-invasive load identification method and a non-invasive load identification system based on a structured load characteristic spectrum, which can improve the accuracy of load identification of different users.
In order to achieve the purpose, the invention provides the following scheme:
a non-invasive load identification method based on a structured load characteristic spectrum comprises the following steps:
acquiring an operation signal of a load corresponding to each load type of a residential user; extracting the characteristics of the running signals of the loads corresponding to each load type, and determining a load attribute set corresponding to the load type; the operation signal of the load corresponding to each load category comprises: the current waveform and the voltage waveform measured when an air conditioner, a washing machine, an electric kettle, a refrigerator, a television, a computer, a blower and an electric cooker run independently; the set of load attributes includes: active power, reactive power, power factor, current distortion rate and Pearson's coefficient between current waveform and standard sine wave;
performing feature fusion on the load attribute set to determine an optimized load attribute set;
determining the weight of each attribute characteristic of each load category according to the optimized load attribute set and the information entropy redundancy between the load attribute characteristics; determining a structured load feature map according to the weight of each attribute feature of each load category;
constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, and optimizing parameters of the SVM load classifier according to the structured load feature map so as to determine an SVM load classifier error set;
determining the number of load classifiers corresponding to each load class according to the SVM load classifier error set, and further constructing a combined SVM load classifier corresponding to each load class;
and (3) carrying out load identification by adopting a combined SVM load classifier corresponding to each class load, and determining the class of the load through a voting group decision mechanism.
Optionally, acquiring an operation signal of a load corresponding to each load category of the residential subscriber; and extracting the characteristics of the running signals of the loads corresponding to each load type, and determining a load attribute set corresponding to the load type, wherein the method specifically comprises the following steps:
determining the voltage waveform zero-crossing point position after the load enters a stable operation state according to the operation signal of the load corresponding to each load type;
extracting the monocycle current and voltage waveforms of the loads corresponding to each load type under the steady state according to the zero-crossing positions of the voltage waveforms;
calculating the steady-state operation characteristic attribute of the load according to the single-period current and voltage waveform of the load corresponding to each load type under the steady state;
and determining a load attribute set of the load of the corresponding class according to the steady-state operation characteristic attribute and the entity attribute of the load corresponding to each load class.
Optionally, the determining the optimized load attribute set by performing feature fusion on the load attribute set specifically includes:
clustering the attribute characteristics in the load attribute set according to the dependency and the correlation degree among the attribute characteristics, and determining a plurality of load attribute subsets of the load attribute set;
and selecting each load attribute subset according to a scatter matrix criterion for attribute fusion, and determining an optimized load attribute set.
Optionally, the weight of each attribute feature of each load category is determined according to the optimized load attribute set and the information entropy redundancy between the load attribute features; further, determining a structured load feature map according to the weight of each attribute feature of each load category, specifically comprising:
normalizing the corresponding attribute characteristics according to the numerical range corresponding to the attribute characteristics in the optimized load attribute set;
and giving the weight of the attribute features after the normalization processing according to an entropy weight method, and determining a structured load feature map according to the weight of the attribute features after the normalization processing.
Optionally, the constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, and optimizing parameters of the SVM load classifier according to the structured load feature map, so as to determine an error set of the SVM load classifier specifically includes:
constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, determining an RBF kernel function, and initializing a penalty factor and kernel function parameters;
randomly sampling from the structured load characteristic map to determine a data set; the data set includes: the weight of each attribute feature of the category of a first set number and the weight of each attribute feature of the non-category of a second set number;
inputting the data set into an SVM load classifier to obtain a classification result;
determining the error of the SVM load classifier according to the classification result;
optimizing a penalty factor and a kernel function parameter, and further determining an optimized SVM load classifier error;
and traversing all load categories, and determining an error set of the SVM load classifier.
A structured load signature based non-intrusive load identification system, comprising:
the load attribute set determining module is used for acquiring an operation signal of a load corresponding to each load category of a residential user; extracting the characteristics of the running signals of the loads corresponding to each load type, and determining a load attribute set corresponding to the load type; the operation signal of the load corresponding to each load category comprises: the current waveform and the voltage waveform measured when an air conditioner, a washing machine, an electric kettle, a refrigerator, a television, a computer, a blower and an electric cooker run independently; the set of load attributes includes: active power, reactive power, power factor, current distortion rate and Pearson's coefficient between current waveform and standard sine wave;
the optimized load attribute set determining module is used for performing feature fusion on the load attribute set to determine an optimized load attribute set;
the structured load characteristic map determining module is used for determining the weight of each attribute characteristic of each load category according to the optimized load attribute set and the information entropy redundancy among the load attribute characteristics; determining a structured load feature map according to the weight of each attribute feature of each load category;
the SVM load classifier error set determining module is used for constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, optimizing parameters of the SVM load classifier according to the structured load feature map, and further determining an SVM load classifier error set;
the combined SVM load classifier determining module is used for determining the number of load classifiers corresponding to each load class according to the SVM load classifier error set so as to construct a combined SVM load classifier corresponding to each load class;
and the load type determining module is used for carrying out load identification by adopting the combined SVM load classifier corresponding to each type of load and determining the load type through a voting group decision mechanism.
Optionally, the load attribute set determining module specifically includes:
the zero-crossing point position determining unit is used for determining the voltage waveform zero-crossing point position after the load enters a stable operation state according to the operation signal of the load corresponding to each load type;
the single-cycle current and voltage waveform determining unit is used for extracting the single-cycle current and voltage waveforms of the loads corresponding to each load type under the steady state according to the zero crossing point positions of the voltage waveforms;
the steady-state operation characteristic attribute determining unit is used for calculating the steady-state operation characteristic attribute of the load according to the single-period current and voltage waveform of the load corresponding to each load type under the steady state;
and the load attribute set determining unit is used for determining the load attribute set of the load of the corresponding category according to the steady-state operation characteristic attribute and the entity attribute of the load corresponding to each load category.
Optionally, the optimized load attribute set determining module specifically includes:
the load attribute subset determining unit is used for clustering the attribute characteristics in the load attribute set according to the dependency and the correlation degree among the attribute characteristics to determine a plurality of load attribute subsets of the load attribute set;
and the optimized load attribute set determining unit is used for selecting each load attribute subset according to the scatter matrix criterion to perform attribute fusion and determining the optimized load attribute set.
Optionally, the structured load feature map determining module specifically includes:
the normalization processing unit is used for performing normalization processing on the corresponding attribute characteristics according to the numerical value range corresponding to the attribute characteristics in the optimized load attribute set;
and the structured load characteristic map determining unit is used for giving the weight of the attribute features after the normalization processing according to an entropy weight method and determining the structured load characteristic map according to the weight of the attribute features after the normalization processing.
Optionally, the module for determining an error set of the SVM load classifier specifically includes:
the SVM load classifier building unit is used for building a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, determining an RBF kernel function, and initializing a penalty factor and kernel function parameters;
the data set determining unit is used for randomly sampling from the structured load characteristic map to determine a data set; the data set includes: the weight of each attribute feature of the category of a first set number and the weight of each attribute feature of the non-category of a second set number;
the classification result determining unit is used for inputting the data set into the SVM load classifier to obtain a classification result;
the SVM load classifier error determination unit is used for determining an SVM load classifier error according to the classification result;
the optimized SVM load classifier error determination unit is used for optimizing a penalty factor and a kernel function parameter so as to determine an optimized SVM load classifier error;
and the SVM load classifier error set determining unit is used for traversing all load categories and determining an SVM load classifier error set.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the non-invasive load identification method based on the structured load characteristic spectrum establishes a load characteristic knowledge spectrum with universality and a combined classifier for common electric appliances, so that the loads of different brands and models of the same electric appliances can be applicable. The electricity load of the resident user can be quickly identified.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
Fig. 1 is a schematic flow chart of a non-invasive load identification method based on a structured load signature provided in the present invention;
FIG. 2 is a schematic diagram illustrating a non-intrusive load identification principle based on a structured load signature according to an embodiment of the present invention;
FIG. 3 is a structured load signature constructed in accordance with an embodiment of the present invention;
FIG. 4 is a diagram of a combined SVM load classifier constructed according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a non-intrusive load identification system based on a structured load signature provided in the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a non-invasive load identification method and a non-invasive load identification system based on a structured load characteristic spectrum, which can improve the accuracy of load identification of different users.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a schematic flow chart of a structured load signature-based non-invasive load identification method, and fig. 2 is a schematic diagram of a structured load signature-based non-invasive load identification principle according to an embodiment of the present invention, as shown in fig. 1 and fig. 2, the structured load signature-based non-invasive load identification method according to the present invention includes:
s101, acquiring an operation signal of a load corresponding to each load type of a residential user; and for each load class pairCharacteristic extraction is carried out on the running signals of the corresponding load, and a load attribute set corresponding to the load category is determined
(ii) a The operation signal of the load corresponding to each load category includes but is not limited to: the current waveform and the voltage waveform measured when an air conditioner, a washing machine, an electric kettle, a refrigerator, a television, a computer, a blower and an electric cooker run independently; the set of load attributes includes, but is not limited to, active power
Reactive power
Power factor of the power converter
Current distortion ratio
And the pearson coefficient r between the current waveform and the constructed standard sine wave;
s101 specifically comprises the following steps:
determining the voltage waveform zero-crossing point position after the load enters a stable operation state according to the operation signal of the load corresponding to each load type;
i.e. using the voltage waveform zero crossing formula
Determining the voltage waveform zero crossing point position;
wherein,
，
is as follows
The voltage waveform collected at each of the sampling points,
the total number of sample points.
Extracting the monocycle current and voltage waveforms of the loads corresponding to each load type under the steady state according to the zero-crossing positions of the voltage waveforms;
calculating the steady-state operation characteristic attribute of the load according to the single-period current and voltage waveform of the load corresponding to each load type under the steady state; steady state operating characteristic attributes include, but are not limited to, active power
Reactive power
Power factor of the power converter
Current distortion ratio
And the pearson coefficient r between the current waveform and the constructed standard sine wave; the specific formula is as follows:
wherein k is the kth load, i is the harmonic order,
in order to be the active power,
in order to be the reactive power,
the current-voltage phase angle of the kth load at the ith sampling point,
in order to be the power factor of the power,
in order to be able to measure the rate of current distortion,
is the pearson coefficient between the load current and the constructed standard sine wave.
And determining a load attribute set of the load of the corresponding class according to the steady-state operation characteristic attribute and the entity attribute of the load corresponding to each load class.
The process of determining the entity attribute specifically comprises the following steps:
using formulas
Judging whether the load is a resistive load or not; wherein, when the load power factor
And correlation coefficient
When the above formula is satisfied, the load can be determined to be a resistive load;
using formulas
Judging whether switching action frequently occurs during load operation; wherein,
is composed of
The number of times of detection of load switching within a period of time,
determining a threshold value for the appliance to be continuously operated;
by using maleFormula (II)
And formula
Judging whether the time length of each running of the load is fixed or not; wherein,
is front
The average duration of the sub-homogeneous load,
for the duration of the detected load belonging to the class,
is a duration float factor;
s102, performing feature fusion on the load attribute set to determine an optimized load attribute set;
s102 specifically comprises the following steps:
clustering the attribute characteristics in the load attribute set according to the dependency and the correlation degree between the attribute characteristics, and determining a plurality of load attribute subsets of the load attribute set
(ii) a Wherein,
is a set of attributes that an entity is involved with,
a subset formed by clustering attributes with high similarity;
selecting each load attribute subset according to a scatter matrix criterion for attribute fusion, and determining an optimized load attribute set
。
The method specifically comprises the following formula:
wherein,
in the form of a load category,
is the overall mean vector of the load samples,m i is as followsiThe mean vector of the class samples is then calculated,
is an inter-class scatter matrix that is,
to spread the eigenvalues of the matrix when
The maximum time subset is the subset after attribute fusion
；
S103, determining each load attribute according to the optimized load attribute set and the information entropy redundancy among the load attribute characteristicsWeight of each attribute feature of load class
(ii) a Determining a structured load feature map according to the weight of each attribute feature of each load category;
s103 specifically comprises the following steps:
normalizing the corresponding attribute characteristics according to the numerical range corresponding to the attribute characteristics in the optimized load attribute set;
wherein,
and
maximum and minimum values of the feature, respectively;
and
respectively before and after standardization;
and giving the weight of the attribute features after the normalization processing according to an entropy weight method, and determining a structured load feature map according to the weight of the attribute features after the normalization processing.
The method specifically comprises the following formula:
wherein,
is as follows
A first sample of
The characteristics of the device are as follows,
is as follows
In the feature of
The specific gravity of each sample in the feature,
to load the
The information entropy redundancy of the individual features,
to load the
A weight of the individual feature;
s104, constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, optimizing parameters of the SVM load classifier according to the structured load feature map, and further determining an error set of the SVM load classifier
；
S104 specifically comprises the following steps:
constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, determining an RBF kernel function, and initializing a penalty factor
And kernel function parameters
；
Randomly sampling from a structured load characteristic map to determine a data set
(ii) a The data set includes:
the sum of the weights of the attribute features of the subject class
The weight of each attribute feature of each non-local category of the person;
generally, 200 can be taken;
inputting data set into SVM load classifier
In (1), obtaining classification results
；
That is, the classifier outputs "1" when it determines that the sample data is "this type of load", and outputs "0" when it determines that the sample data is "non-this type of load".
Determining SVM negative according to the classification resultError of the lotus classifier
(ii) a Wherein,
，
representing the number of samples with correct classification results;
to be provided with
Repeating the steps for the target function, optimizing the penalty factor and the kernel function parameter, and further determining the error of the optimized SVM load classifier;
generally, 0.5 can be taken;
traversing all load classes to determine an error set of the SVM load classifier
。
S105, determining the number of load classifiers corresponding to each load class according to the SVM load classifier error set, and further constructing a combined SVM load classifier corresponding to each load class;
the concrete formula is as follows:
wherein,
in order to combine the error rates of the SVM load classifiers,
and
respectively is the lower limit and the upper limit of the number of SVM single classifiers contained in the combined SVM load classifier,
for the classification error of a single classifier of the SVM,
a threshold is determined for the error rate. Number of current SVM single classifiers
Is taken to satisfy
And meanwhile, the classification accuracy of the combined SVM load classifier is considered to meet the requirement.
And S106, carrying out load identification by adopting the combined SVM load classifier corresponding to each class load, and determining the class of the load through a voting group decision mechanism.
S106 specifically comprises:
extracting a set of load attributes of an unknown load waveform to be identified
；
Pre-classifying the load according to the entity characteristics of the unknown load, determining the entity type corresponding to the load and the combined SVM load classifier set to be input by the unknown load
；
Inputting load waveform into combined SVM load classifier
In the method, the output classification result is obtained
The concrete formula is as follows:
wherein,
is a combined SVM load classifier
And (5) classifying the load waveform by the N SVM load classifiers. If the SVM load classifier determines that the waveform is 'yes'
Corresponding load category, output "1", if judge "not"
And outputting '0' according to the corresponding load type,
after the judgment result of each base classifier is obtained, whether the unknown load belongs to the load class judged by the classifier is judged through a voting group decision mechanism, wherein the specific formula is as follows:
repeating the above steps until the unknown load waveform is classified into a classifier set
The classifier in (1) is traversed;
selecting the load category of the classifier with the highest vote number and the number being 'yes' as a category voting result of the unknown load waveform, wherein the specific formula is as follows:
wherein,
voting results for the load categories.
And if the voting results of all the combined SVM load classifiers are 'non' -class loads, indicating that a new load class which does not belong to the class set appears.
The acquisition device is arranged at an entrance of a commonly used electrical load of residents which operates independently, so that voltage waveform and current waveform signal data are acquired, and the sampling frequency is 10 kHz.
According to the method of the invention:
1. and collecting current waveforms and voltage waveforms when typical electricity loads of residential users run independently. And carrying out steady-state waveform extraction on the current and voltage waveforms of the independent operation loads.
2. And calculating a characteristic value according to the extracted steady-state current waveform, performing characteristic fusion and weighting, and forming a structured load characteristic map, as shown in fig. 3.
3. And constructing an SVM load classifier for each type of load.
4. And respectively constructing a combined SVM load classifier for each type of load.
5. The load is pre-classified according to the entity characteristics of the load to be identified, the entity type corresponding to the load is determined, and the unknown load waveform is input into the combined SVM load classifier set.
6. Selecting the load category of the classifier with the highest vote number and the number of yes as a category voting result of the unknown load waveform according to the output result; if the voting results of all the combined SVM load classifiers are 'not' load of the same type, a new load class which does not belong to the class set appears, as shown in FIG. 4.
Fig. 5 is a schematic structural diagram of a structured load signature-based non-invasive load identification system according to the present invention, and as shown in fig. 5, the structured load signature-based non-invasive load identification system according to the present invention includes:
a load attribute set determining module 501, configured to obtain an operation signal of a load corresponding to each load category of a residential user; extracting the characteristics of the running signals of the loads corresponding to each load type, and determining a load attribute set corresponding to the load type; the operation signal of the load corresponding to each load category comprises: the current waveform and the voltage waveform measured when an air conditioner, a washing machine, an electric kettle, a refrigerator, a television, a computer, a blower and an electric cooker run independently; the set of load attributes includes: active power, reactive power, power factor, current distortion rate and Pearson's coefficient between current waveform and standard sine wave;
an optimized load attribute set determining module 502, configured to perform feature fusion on the load attribute set to determine an optimized load attribute set;
a structured load feature map determining module 503, configured to determine a weight of each attribute feature of each load category according to the optimized load attribute set and the information entropy redundancy between the load attribute features; determining a structured load feature map according to the weight of each attribute feature of each load category;
an SVM load classifier error set determining module 504, configured to construct a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, and optimize parameters of the SVM load classifier according to the structured load feature map, so as to determine an SVM load classifier error set;
a combined SVM load classifier determining module 505, configured to determine the number of load classifiers corresponding to each load category according to the error set of the SVM load classifiers, and further construct a combined SVM load classifier corresponding to each load category;
and a load category determining module 506, configured to perform load identification by using the combined SVM load classifier corresponding to each category load, and determine a load category through a voting group decision mechanism.
The load attribute set determining module 501 specifically includes:
the zero-crossing point position determining unit is used for determining the voltage waveform zero-crossing point position after the load enters a stable operation state according to the operation signal of the load corresponding to each load type;
the single-cycle current and voltage waveform determining unit is used for extracting the single-cycle current and voltage waveforms of the loads corresponding to each load type under the steady state according to the zero crossing point positions of the voltage waveforms;
the steady-state operation characteristic attribute determining unit is used for calculating the steady-state operation characteristic attribute of the load according to the single-period current and voltage waveform of the load corresponding to each load type under the steady state;
and the load attribute set determining unit is used for determining the load attribute set of the load of the corresponding category according to the steady-state operation characteristic attribute and the entity attribute of the load corresponding to each load category.
The optimized load attribute set determining module 502 specifically includes:
the load attribute subset determining unit is used for clustering the attribute characteristics in the load attribute set according to the dependency and the correlation degree among the attribute characteristics to determine a plurality of load attribute subsets of the load attribute set;
and the optimized load attribute set determining unit is used for selecting each load attribute subset according to the scatter matrix criterion to perform attribute fusion and determining the optimized load attribute set.
The structural load feature map determining module 503 specifically includes:
the normalization processing unit is used for performing normalization processing on the corresponding attribute characteristics according to the numerical value range corresponding to the attribute characteristics in the optimized load attribute set;
and the structured load characteristic map determining unit is used for giving the weight of the attribute features after the normalization processing according to an entropy weight method and determining the structured load characteristic map according to the weight of the attribute features after the normalization processing.
The SVM load classifier error set determination module 504 specifically includes:
the SVM load classifier building unit is used for building a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, determining an RBF kernel function, and initializing a penalty factor and kernel function parameters;
the data set determining unit is used for randomly sampling from the structured load characteristic map to determine a data set; the data set includes: the weight of each attribute feature of the category of a first set number and the weight of each attribute feature of the non-category of a second set number;
the classification result determining unit is used for inputting the data set into the SVM load classifier to obtain a classification result;
the SVM load classifier error determination unit is used for determining an SVM load classifier error according to the classification result;
the optimized SVM load classifier error determination unit is used for optimizing a penalty factor and a kernel function parameter so as to determine an optimized SVM load classifier error;
and the SVM load classifier error set determining unit is used for traversing all load categories and determining an SVM load classifier error set.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (10)
1. A non-invasive load identification method based on a structured load characteristic map is characterized by comprising the following steps:
acquiring an operation signal of a load corresponding to each load type of a residential user; extracting the characteristics of the running signals of the loads corresponding to each load type, and determining a load attribute set corresponding to the load type; the operation signal of the load corresponding to each load category comprises: the current waveform and the voltage waveform measured when an air conditioner, a washing machine, an electric kettle, a refrigerator, a television, a computer, a blower and an electric cooker run independently; the set of load attributes includes: active power, reactive power, power factor, current distortion rate and Pearson's coefficient between current waveform and standard sine wave;
performing feature fusion on the load attribute set to determine an optimized load attribute set;
determining the weight of each attribute characteristic of each load category according to the optimized load attribute set and the information entropy redundancy between the load attribute characteristics; determining a structured load feature map according to the weight of each attribute feature of each load category;
constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, and optimizing parameters of the SVM load classifier according to the structured load feature map so as to determine an SVM load classifier error set;
determining the number of load classifiers corresponding to each load class according to the SVM load classifier error set, and further constructing a combined SVM load classifier corresponding to each load class;
and (3) carrying out load identification by adopting a combined SVM load classifier corresponding to each class load, and determining the class of the load through a voting group decision mechanism.
2. The method according to claim 1, wherein the method comprises obtaining an operation signal of a load corresponding to each load category of a residential user; and extracting the characteristics of the running signals of the loads corresponding to each load type, and determining a load attribute set corresponding to the load type, wherein the method specifically comprises the following steps:
determining the voltage waveform zero-crossing point position after the load enters a stable operation state according to the operation signal of the load corresponding to each load type;
extracting the monocycle current and voltage waveforms of the loads corresponding to each load type under the steady state according to the zero-crossing positions of the voltage waveforms;
calculating the steady-state operation characteristic attribute of the load according to the single-period current and voltage waveform of the load corresponding to each load type under the steady state;
and determining a load attribute set of the load of the corresponding class according to the steady-state operation characteristic attribute and the entity attribute of the load corresponding to each load class.
3. The method according to claim 1, wherein the determining the optimized load attribute set by performing feature fusion on the load attribute set specifically comprises:
clustering the attribute characteristics in the load attribute set according to the dependency and the correlation degree among the attribute characteristics, and determining a plurality of load attribute subsets of the load attribute set;
and selecting each load attribute subset according to a scatter matrix criterion for attribute fusion, and determining an optimized load attribute set.
4. The method according to claim 1, wherein the weight of each attribute feature of each load class is determined according to the optimized load attribute set and the information entropy redundancy between each load attribute feature; further, determining a structured load feature map according to the weight of each attribute feature of each load category, specifically comprising:
normalizing the corresponding attribute characteristics according to the numerical range corresponding to the attribute characteristics in the optimized load attribute set;
and giving the weight of the attribute features after the normalization processing according to an entropy weight method, and determining a structured load feature map according to the weight of the attribute features after the normalization processing.
5. The method according to claim 1, wherein the method for non-intrusive load identification based on the structured load feature map comprises the steps of constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load class, and optimizing parameters of the SVM load classifier according to the structured load feature map to determine an SVM load classifier error set, and specifically comprises:
constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, determining an RBF kernel function, and initializing a penalty factor and kernel function parameters;
randomly sampling from the structured load characteristic map to determine a data set; the data set includes: the weight of each attribute feature of the category of a first set number and the weight of each attribute feature of the non-category of a second set number;
inputting the data set into an SVM load classifier to obtain a classification result;
determining the error of the SVM load classifier according to the classification result;
optimizing a penalty factor and a kernel function parameter, and further determining an optimized SVM load classifier error;
and traversing all load categories, and determining an error set of the SVM load classifier.
6. A structured load signature based non-intrusive load identification system, comprising:
the load attribute set determining module is used for acquiring an operation signal of a load corresponding to each load category of a residential user; extracting the characteristics of the running signals of the loads corresponding to each load type, and determining a load attribute set corresponding to the load type; the operation signal of the load corresponding to each load category comprises: the current waveform and the voltage waveform measured when an air conditioner, a washing machine, an electric kettle, a refrigerator, a television, a computer, a blower and an electric cooker run independently; the set of load attributes includes: active power, reactive power, power factor, current distortion rate and Pearson's coefficient between current waveform and standard sine wave;
the optimized load attribute set determining module is used for performing feature fusion on the load attribute set to determine an optimized load attribute set;
the structured load characteristic map determining module is used for determining the weight of each attribute characteristic of each load category according to the optimized load attribute set and the information entropy redundancy among the load attribute characteristics; determining a structured load feature map according to the weight of each attribute feature of each load category;
the SVM load classifier error set determining module is used for constructing a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, optimizing parameters of the SVM load classifier according to the structured load feature map, and further determining an SVM load classifier error set;
the combined SVM load classifier determining module is used for determining the number of load classifiers corresponding to each load class according to the SVM load classifier error set so as to construct a combined SVM load classifier corresponding to each load class;
and the load type determining module is used for carrying out load identification by adopting the combined SVM load classifier corresponding to each type of load and determining the load type through a voting group decision mechanism.
7. The system according to claim 6, wherein the load attribute set determination module specifically comprises:
the zero-crossing point position determining unit is used for determining the voltage waveform zero-crossing point position after the load enters a stable operation state according to the operation signal of the load corresponding to each load type;
the single-cycle current and voltage waveform determining unit is used for extracting the single-cycle current and voltage waveforms of the loads corresponding to each load type under the steady state according to the zero crossing point positions of the voltage waveforms;
the steady-state operation characteristic attribute determining unit is used for calculating the steady-state operation characteristic attribute of the load according to the single-period current and voltage waveform of the load corresponding to each load type under the steady state;
and the load attribute set determining unit is used for determining the load attribute set of the load of the corresponding category according to the steady-state operation characteristic attribute and the entity attribute of the load corresponding to each load category.
8. The system according to claim 6, wherein the optimized load attribute set determination module specifically comprises:
the load attribute subset determining unit is used for clustering the attribute characteristics in the load attribute set according to the dependency and the correlation degree among the attribute characteristics to determine a plurality of load attribute subsets of the load attribute set;
and the optimized load attribute set determining unit is used for selecting each load attribute subset according to the scatter matrix criterion to perform attribute fusion and determining the optimized load attribute set.
9. The system according to claim 6, wherein the structured load signature determination module specifically comprises:
the normalization processing unit is used for performing normalization processing on the corresponding attribute characteristics according to the numerical value range corresponding to the attribute characteristics in the optimized load attribute set;
and the structured load characteristic map determining unit is used for giving the weight of the attribute features after the normalization processing according to an entropy weight method and determining the structured load characteristic map according to the weight of the attribute features after the normalization processing.
10. The structured load signature based non-intrusive load recognition system of claim 6, wherein the SVM load classifier error set determination module specifically comprises:
the SVM load classifier building unit is used for building a corresponding SVM load classifier according to the optimized load attribute set and the corresponding load category, determining an RBF kernel function, and initializing a penalty factor and kernel function parameters;
the data set determining unit is used for randomly sampling from the structured load characteristic map to determine a data set; the data set includes: the weight of each attribute feature of the category of a first set number and the weight of each attribute feature of the non-category of a second set number;
the classification result determining unit is used for inputting the data set into the SVM load classifier to obtain a classification result;
the SVM load classifier error determination unit is used for determining an SVM load classifier error according to the classification result;
the optimized SVM load classifier error determination unit is used for optimizing a penalty factor and a kernel function parameter so as to determine an optimized SVM load classifier error;
and the SVM load classifier error set determining unit is used for traversing all load categories and determining an SVM load classifier error set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210183198.6A CN114239762A (en) | 2022-02-28 | 2022-02-28 | Non-invasive load identification method and system based on structured load characteristic spectrum |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210183198.6A CN114239762A (en) | 2022-02-28 | 2022-02-28 | Non-invasive load identification method and system based on structured load characteristic spectrum |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114239762A true CN114239762A (en) | 2022-03-25 |
Family
ID=80748324
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210183198.6A Pending CN114239762A (en) | 2022-02-28 | 2022-02-28 | Non-invasive load identification method and system based on structured load characteristic spectrum |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114239762A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116028853A (en) * | 2023-02-14 | 2023-04-28 | 华北电力大学 | Household appliance identification method and system based on knowledge graph |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109492667A (en) * | 2018-10-08 | 2019-03-19 | 国网天津市电力公司电力科学研究院 | A kind of feature selecting discrimination method for non-intrusive electrical load monitoring |
| CN111209976A (en) * | 2020-01-14 | 2020-05-29 | 华北电力大学 | Non-invasive load monitoring method and system |
-
2022
- 2022-02-28 CN CN202210183198.6A patent/CN114239762A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109492667A (en) * | 2018-10-08 | 2019-03-19 | 国网天津市电力公司电力科学研究院 | A kind of feature selecting discrimination method for non-intrusive electrical load monitoring |
| CN111209976A (en) * | 2020-01-14 | 2020-05-29 | 华北电力大学 | Non-invasive load monitoring method and system |
Non-Patent Citations (1)
| Title |
|---|
| 焦点: ""非侵入式负荷自动在线监测方法研究"", 《中国优秀硕士学位论文全文数据库 (工程科技Ⅱ辑)》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116028853A (en) * | 2023-02-14 | 2023-04-28 | 华北电力大学 | Household appliance identification method and system based on knowledge graph |
| CN116028853B (en) * | 2023-02-14 | 2023-09-12 | 华北电力大学 | Household appliance identification method and system based on knowledge graph |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106786534B (en) | A kind of non-intrusive electrical load transient process discrimination method and system | |
| CN110954744A (en) | Non-invasive load monitoring method based on event detection | |
| CN110514889A (en) | A kind of method and system of non-intrusion type household electricity remained capacity | |
| TWI440862B (en) | Electrical detection method and system based on user feedback information | |
| CN110119545B (en) | Non-invasive power load identification method based on stack type self-encoder | |
| CN109345409A (en) | A kind of resident's comprehensive energy efficiency management method based on bandwidth carrier | |
| CN109359665B (en) | Household appliance load identification method and device based on support vector machine | |
| CN108376262A (en) | A kind of analysis model construction method of wind power output typical characteristics | |
| CN111639586B (en) | Non-invasive load identification model construction method, load identification method and system | |
| CN112580471A (en) | Non-invasive load identification method based on AdaBoost feature extraction and RNN model | |
| CN111242161B (en) | Non-invasive non-resident user load identification method based on intelligent learning | |
| CN108345906B (en) | Non-invasive electrical appliance identification method based on Boost model | |
| CN112732748A (en) | Non-invasive household appliance load identification method based on adaptive feature selection | |
| CN111563827A (en) | Load decomposition method based on electrical appliance physical characteristics and residential electricity consumption behaviors | |
| CN113887912A (en) | Non-invasive load identification method for deeply learning downward embedded equipment | |
| CN112909923A (en) | Non-invasive household load behavior recognition device based on DTW algorithm | |
| Chen et al. | Non-intrusive load monitoring based on feature extraction of change-point and xgboost classifier | |
| CN116796271A (en) | Resident energy abnormality identification method | |
| CN114239762A (en) | Non-invasive load identification method and system based on structured load characteristic spectrum | |
| CN112305441A (en) | Power battery health state assessment method under integrated clustering | |
| CN114942344A (en) | Non-invasive electrical appliance identification method, system, medium, equipment and terminal | |
| CN116610922A (en) | Non-invasive load identification method and system based on multi-strategy learning | |
| Jazizadeh et al. | Unsupervised clustering of residential electricity consumption measurements for facilitated user-centric non-intrusive load monitoring | |
| CN115932435A (en) | Resident non-invasive load monitoring method based on low-frequency acquisition signals | |
| CN115687999A (en) | Non-invasive load sensing method fusing multi-time scale electric appliance characteristics |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220325 |