CN108181533A - A kind of electric load intelligent measurement of non-intrusion type and hierarchical classification method - Google Patents
A kind of electric load intelligent measurement of non-intrusion type and hierarchical classification method Download PDFInfo
- Publication number
- CN108181533A CN108181533A CN201810048734.5A CN201810048734A CN108181533A CN 108181533 A CN108181533 A CN 108181533A CN 201810048734 A CN201810048734 A CN 201810048734A CN 108181533 A CN108181533 A CN 108181533A
- Authority
- CN
- China
- Prior art keywords
- load
- event
- frequency domain
- classification
- temporal signatures
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000005259 measurement Methods 0.000 title claims abstract description 11
- 230000002123 temporal effect Effects 0.000 claims abstract description 30
- 238000012360 testing method Methods 0.000 claims abstract description 11
- 230000001052 transient effect Effects 0.000 claims description 26
- 238000001914 filtration Methods 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000001514 detection method Methods 0.000 abstract description 10
- 238000005516 engineering process Methods 0.000 abstract description 3
- 238000004458 analytical method Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005406 washing Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004378 air conditioning Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Alarm Systems (AREA)
Abstract
The present invention relates to intelligent testing technology fields, a kind of electric load intelligent measurement of non-intrusion type and hierarchical classification method are provided, identify that two parts form by event detection and hierarchical classification, it is compared and judges out with the threshold value set by the difference of the current strength in former and later two periods of calculated current signal, pass event, classified in load identification using temporal signatures and frequency domain character using the SVM algorithm of layering, with the detection of existing electric load, identification, the method of classification is compared, the present invention solves to be sentenced often occur less, sentence more, the situation of erroneous judgement, improve the accuracy of stress event detection and the accuracy of load equipment classification.
Description
Technical field
The present invention relates to intelligent testing technology fields, and in particular to a kind of electric load intelligent measurement of non-intrusion type and point
Layer sorting technique.
Background technology
Non-intrusive electrical load detection method is mainly to utilize the letters such as voltage, electric current and the power of electric load inlet
Breath is analyzed and is studied to its ingredient, point including the analysis method based on load steady state characteristic and based on load transient characteristic
Two kinds of analysis method.Mainly have currently based on the analysis method of load transient characteristic and the transient state of voltage, electric current is sampled, extract
The characteristic value of electric current, such as the root mean square of current peak, current average and electric current, after the characteristic value of these electric currents has been obtained
Classified again using various methods to electric appliance.Sorting technique mainly has k nearest neighbor rule and back-propagation artificial neural network.K
Neighbour's rule classification method training time is shorter, the automatic classification for the class field for being suitble to sample size bigger, but calculation amount ratio
It is larger, to the predictablity rate of classification than relatively low;The self study of back-propagation artificial neural network method and adaptive ability and
Fault-tolerant ability is stronger, but convergence rate is slow, other classifications are easily judged to the classification of classification.Generally speaking, currently based on
The problem of analysis method of load transient characteristic is primarily present is as follows:
(1) there is the phenomenon that sentencing more, sentencing less and judge by accident in the detection of event;
(2) exist to the event detection of low power electric appliance and challenge, there is the phenomenon that can't detect;
(3) electrical equipment of similar characteristic has similar feature, and more difficult in classification, Classification and Identification rate is relatively low.
Invention content
For the detection of existing electric load, identification, classification method there are the problem of, when the present invention proposes a kind of
The hierarchical classification method that domain, frequency domain are combined improves the accuracy of stress event detection and the accuracy of load equipment classification of type.
To achieve the above object, the specific technical solution of the present invention is as follows:A kind of electric load of non-intrusion type is intelligently examined
Survey and hierarchical classification method, include the following steps:
1) system initialization, wherein, initialization content includes:Open threshold value, close threshold value, frequency domain character sample database, when
Characteristic of field sample database 1, temporal signatures sample database 2;
2) acquisition, conversion and the filtering of load signal carries out A/D to collected initial load signal and is converted to number
Then signal carries out Chebyshev's low-pass filtering;It goes to step 3);
3) digital signal is detected, event, the generation of pass event is judged whether there is out, when such as event occurs, goes to step 4),
Otherwise it goes to step 3);
4) transient process that interception event occurs, and the temporal signatures and frequency domain character of event transient process are extracted, turn step
It is rapid 5);
5) classified using frequency domain grader to load, i.e., using the frequency domain character of event transient process as test set,
Using frequency domain character sample database as sample set, using SVM algorithm to load classification, if load matching degree is more than 90%, classify
9) success, goes to step, otherwise, goes to step 6);
6) classified using time domain device 1 to load, i.e., using the temporal signatures of event transient process as test set,
Using temporal signatures sample database 1 as sample set, using SVM algorithm to load classification, if load matching degree is more than 90%, classify
9) success, goes to step, otherwise, goes to step 7);
7) classified using time domain device 2 to load, i.e., using the temporal signatures of event transient process as test set,
Using temporal signatures sample database 2 as sample set, using SVM algorithm to load classification, if load matching degree is more than 90%, classify
9) success, goes to step, otherwise, goes to step 8);
8) classify unsuccessful, report error message, go to step 3);
9) load classification success.
Further, above-mentioned steps 1) in, the frequency domain character of frequency domain character sample database include frequency domain energy, harmonic wave vector,
Harmonic distortion;The temporal signatures of characteristic of field sample database 1 include:Current maxima, current average, electric current root mean square, transient state week
Phase, form factor, transient state energy;The temporal signatures of temporal signatures sample database 2 include crest factor, peak-to-peak value, peak factor,
Instantaneous power, low level ratio, high level ratio.
Further, above-mentioned steps 3) in, judge whether that the method that event occurs is:
3.1) difference of the current strength in former and later two periods of calculated current signal;
3.2) event is detected out if difference is more than and opens threshold value;
3.3) pass event is detected if difference is less than and closes threshold value.
Further, above-mentioned we intercept the transient process of event generation when detecting that event occurs, extract on frequency domain
Characteristic parameter, to load carry out first step classification, if load matching degree be higher than 90%, be considered as and identify successfully output load class
Type, if load matching degree be less than 90%, be considered as it is unfiled go out load, by it is unfiled go out load extraction time domain on one
Characteristic parameter is divided to carry out second step classification, if load matching degree is higher than 90%, is considered as and identifies successfully output load type, if load
Matching degree is less than 90%, then be considered as it is unfiled go out load, by it is unfiled go out load be then transferred to next step time domain device and exist
Other characteristic parameters extracted in time domain carry out final step classification.
The beneficial effects of the invention are as follows:With the detection of existing electric load, identification, classification method compared with, it is of the invention
A kind of electric load intelligent measurement of non-intrusion type and hierarchical classification method, solve often occur lack the feelings sentencing, sentence more, judging by accident
Condition improves the accuracy of stress event detection and the accuracy of load equipment classification.
Description of the drawings
Fig. 1 is non-intrusive electrical load intelligent measurement and hierarchical classification method flow diagram.
Specific embodiment
The invention will be further described with embodiment below in conjunction with the accompanying drawings.
It is as shown in Figure 1 the non-intrusive electrical load intelligent measurement of the present invention and hierarchical classification method flow diagram, it is a kind of
The intelligent measurement and hierarchical classification method of non-intrusion type include the following steps:
1) system initialization, initialization content include:
1.1) initialization opens threshold value, closes threshold value;
1.2) frequency domain character for extracting each electric appliances establishes frequency domain character sample database, wherein, frequency domain character includes frequency domain energy
Amount, harmonic wave vector, harmonic distortion;
1.3) temporal signatures for extracting each electric appliances establish temporal signatures sample database 1, wherein, temporal signatures sample database 1 is wrapped
The temporal signatures contained include:Current maxima, current average, electric current root mean square, transient period, form factor, transient state energy;
1.4) temporal signatures for extracting each electric appliances establish temporal signatures sample database 2, wherein, temporal signatures sample database 2 is wrapped
The temporal signatures contained include crest factor, peak-to-peak value, peak factor, instantaneous power, low level ratio, high level ratio;
2) acquisition, conversion and the filtering of load signal carries out collected initial load signal A/D conversions, conversion
For digital signal, Chebyshev's low-pass filtering is then carried out;Turn in next step;
3) digital signal is detected, event, the generation of pass event is judged whether there is out, when such as event occurs, goes to step 4),
Otherwise it goes to step 3);Wherein, judge whether the method that event occurs be former and later two periods of calculated current signal electric current it is strong
The difference of degree, is compared with threshold value, then detects out event if more than threshold value is opened, pass is detected if less than threshold value is closed
Event;
4) transient process that interception event occurs, and the temporal signatures of event transient process and frequency desire feature are extracted, turn step
It is rapid 5);
5) classified using frequency domain grader to load, i.e., using the frequency domain character of event transient process as test set,
Using frequency domain character sample database as sample set, using SVM algorithm to load classification, if load matching degree is higher than 90%, classify
9) success, goes to step, otherwise, goes to step 6);
6) classified using time domain device 1 to load, i.e., using the temporal signatures of event transient process as test set,
Using temporal signatures sample database 1 as sample set, using SVM algorithm to load classification, if load matching degree is higher than 90%, classify
9) success, goes to step, otherwise, goes to step 7);
7) classified using time domain device 2 to load, i.e., using the temporal signatures of event transient process as test set,
Using temporal signatures sample database 2 as sample set, using SVM algorithm to load classification, if load matching degree is higher than 90%, classify
9) success, goes to step, otherwise, goes to step 8);
8) classify unsuccessful, provide error message, go to step 3);
9) load classification success.
It in the specific embodiment of the invention, opens threshold value and is initialized as 0.15A, close threshold value and be initialized as -0.28A;Load
Including electric light, fan, insulating pot, micro-wave oven, display, host desktop, electric heater, washing machine, refrigerator, water dispenser, TV, sky
12 kinds of tune etc..Electric light, fan, insulating pot, electric heater, water dispenser, TV can accurately be sorted out by step 4), pass through step
5) refrigerator, micro-wave oven, air-conditioning can be sorted out, remaining electric appliance display, washing machine, host desktop can all be divided by step 6)
Class.
Claims (3)
1. a kind of electric load intelligent measurement of non-intrusion type and hierarchical classification method, which is characterized in that include the following steps:
1) system initialization, wherein, initialization content includes:It opens threshold value, close threshold value, frequency domain character sample database, time domain spy
Levy sample database 1, temporal signatures sample database 2;
2) acquisition, conversion and the filtering of load signal carries out A/D to collected initial load signal and is converted to digital letter
Number, then carry out Chebyshev's low-pass filtering;It goes to step 3);
3) digital signal is detected, event, the generation of pass event is judged whether there is out, when such as event occurs, goes to step 4), otherwise turn
Step 3);
4) transient process that interception event occurs, and the temporal signatures and frequency domain character of event transient process are extracted, it goes to step 5);
5) classified using frequency domain grader to load, i.e., using the frequency domain character of event transient process as test set, with frequency
Characteristic of field sample database is as sample set, using SVM algorithm to load classification, if load matching degree is more than 90%, and success of classifying,
It goes to step 9), otherwise, goes to step 6);
6) classified using time domain device 1 to load, i.e., using the temporal signatures of event transient process as test set, with when
Characteristic of field sample database 1 is used as sample set, using SVM algorithm to load classification, if load matching degree is more than 90%, is categorized into
9) work(is gone to step, otherwise, go to step 7);
7) classified using time domain device 2 to load, i.e., using the temporal signatures of event transient process as test set, with when
Characteristic of field sample database 2 is used as sample set, using SVM algorithm to load classification, if load matching degree is more than 90%, is categorized into
9) work(is gone to step, otherwise, go to step 8);
8) classify unsuccessful, report error message, go to step 3);
9) load classification success.
2. the electric load intelligent measurement of non-intrusion type as described in claim 1 and hierarchical classification method, which is characterized in that institute
It states in step 1), the frequency domain character of frequency domain character sample database includes frequency domain energy, harmonic wave vector, harmonic distortion;Characteristic of field sample
The temporal signatures in library 1 include:Current maxima, current average, electric current root mean square, transient period, form factor, transient state energy
Amount;The temporal signatures of temporal signatures sample database 2 include crest factor, peak-to-peak value, peak factor, instantaneous power, low level ratio
Rate, high level ratio.
3. the electric load intelligent measurement of non-intrusion type as described in claim 1 and hierarchical classification method, which is characterized in that institute
It states in step 3), judges whether that the method that event occurs is:
3.1) difference of the current strength in former and later two periods of calculated current signal;
3.2) event is detected out if difference is more than and opens threshold value;
3.3) pass event is detected if difference is less than and closes threshold value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810048734.5A CN108181533B (en) | 2018-01-18 | 2018-01-18 | Non-invasive power load intelligent detection and hierarchical classification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810048734.5A CN108181533B (en) | 2018-01-18 | 2018-01-18 | Non-invasive power load intelligent detection and hierarchical classification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108181533A true CN108181533A (en) | 2018-06-19 |
CN108181533B CN108181533B (en) | 2020-02-04 |
Family
ID=62550885
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810048734.5A Expired - Fee Related CN108181533B (en) | 2018-01-18 | 2018-01-18 | Non-invasive power load intelligent detection and hierarchical classification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108181533B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109596912A (en) * | 2018-11-21 | 2019-04-09 | 河海大学 | A kind of decomposition method of non-intrusion type power load |
CN112034281A (en) * | 2020-07-30 | 2020-12-04 | 河海大学 | Non-invasive load identification method in bedroom electricity environment |
CN114184870A (en) * | 2021-12-14 | 2022-03-15 | 河北科技大学 | Non-invasive load identification method and equipment |
CN115825634A (en) * | 2023-02-16 | 2023-03-21 | 上海红檀智能科技有限公司 | Malignant load identification method based on load complex impedance characteristics |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106646026A (en) * | 2016-11-11 | 2017-05-10 | 华北电力大学 | Non-intrusive household appliance load identification method |
CN106802379A (en) * | 2017-03-06 | 2017-06-06 | 中国海洋大学 | A kind of load switch detection of adaptive threshold and recognition methods and system |
-
2018
- 2018-01-18 CN CN201810048734.5A patent/CN108181533B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106646026A (en) * | 2016-11-11 | 2017-05-10 | 华北电力大学 | Non-intrusive household appliance load identification method |
CN106802379A (en) * | 2017-03-06 | 2017-06-06 | 中国海洋大学 | A kind of load switch detection of adaptive threshold and recognition methods and system |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109596912A (en) * | 2018-11-21 | 2019-04-09 | 河海大学 | A kind of decomposition method of non-intrusion type power load |
CN112034281A (en) * | 2020-07-30 | 2020-12-04 | 河海大学 | Non-invasive load identification method in bedroom electricity environment |
CN114184870A (en) * | 2021-12-14 | 2022-03-15 | 河北科技大学 | Non-invasive load identification method and equipment |
CN115825634A (en) * | 2023-02-16 | 2023-03-21 | 上海红檀智能科技有限公司 | Malignant load identification method based on load complex impedance characteristics |
Also Published As
Publication number | Publication date |
---|---|
CN108181533B (en) | 2020-02-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108181533A (en) | A kind of electric load intelligent measurement of non-intrusion type and hierarchical classification method | |
CN106802379B (en) | The load switch detection and recognition methods of a kind of adaptive threshold and system | |
Wang et al. | ArcNet: Series AC arc fault detection based on raw current and convolutional neural network | |
CN106682303B (en) | A kind of three-level inverter method for diagnosing faults based on empirical mode decomposition and decision tree RVM | |
Abdelsalam et al. | Characterization of power quality disturbances using hybrid technique of linear Kalman filter and fuzzy-expert system | |
Etemadi et al. | High-impedance fault detection using multi-resolution signal decomposition and adaptive neural fuzzy inference system | |
CN110514889A (en) | A kind of method and system of non-intrusion type household electricity remained capacity | |
Devlin et al. | Non-intrusive load monitoring using electricity smart meter data: A deep learning approach | |
CN108343599B (en) | A kind of water pump assembly intelligent failure diagnosis method based on more granularities cascade forest | |
CN106908671A (en) | A kind of non-intrusion type household loads intelligent detecting method and system | |
CN107273920A (en) | A kind of non-intrusion type household electrical appliance recognition methods based on random forest | |
CN111027408A (en) | Load identification method based on support vector machine and V-I curve characteristics | |
CN106483370A (en) | Non-intrusion type household loads real-time identification method based on multi-feature fusion and device | |
CN109165604A (en) | The recognition methods of non-intrusion type load and its test macro based on coorinated training | |
CN110516788A (en) | Bi-LSTM algorithm-based non-invasive load detection method and system | |
CN110504679A (en) | KM matching algorithm-based non-invasive load identification method | |
CN112580471A (en) | Non-invasive load identification method based on AdaBoost feature extraction and RNN model | |
CN113567794B (en) | Electric bicycle indoor charging identification method and system based on dynamic time warping | |
CN110852441B (en) | Fire disaster early warning method based on improved naive Bayes algorithm | |
Fatama et al. | A multi feature based islanding classification technique for distributed generation systems | |
Le et al. | Ensemble machine learning based adaptive arc fault detection for DC distribution systems | |
parimita Mishra et al. | Islanding detection based on variational mode decomposition for inverter based distributed generation systems | |
CN114709926A (en) | Non-invasive load identification method fused with transient and steady state characteristics | |
CN115112989B (en) | Non-invasive load monitoring method based on low-frequency data | |
CN107578016A (en) | A kind of residual current waveform automatic identifying method based on rarefaction representation |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200204 |
|
CF01 | Termination of patent right due to non-payment of annual fee |