CN111242161B - Non-invasive non-resident user load identification method based on intelligent learning - Google Patents
Non-invasive non-resident user load identification method based on intelligent learning Download PDFInfo
- Publication number
- CN111242161B CN111242161B CN201911375093.5A CN201911375093A CN111242161B CN 111242161 B CN111242161 B CN 111242161B CN 201911375093 A CN201911375093 A CN 201911375093A CN 111242161 B CN111242161 B CN 111242161B
- Authority
- CN
- China
- Prior art keywords
- load
- data
- switching
- test data
- data set
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000012360 testing method Methods 0.000 claims description 59
- 238000012549 training Methods 0.000 claims description 49
- 230000006870 function Effects 0.000 claims description 13
- 238000007635 classification algorithm Methods 0.000 claims description 12
- 238000003062 neural network model Methods 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 3
- 238000004422 calculation algorithm Methods 0.000 abstract description 11
- 238000013528 artificial neural network Methods 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 6
- 238000012544 monitoring process Methods 0.000 description 5
- 238000007726 management method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000005312 nonlinear dynamic Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
Classifications
-
- 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
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The application provides a non-invasive non-resident user load identification method based on intelligent learning. Firstly, carrying out load identification on the collected data of the non-resident user load by using a trained K nearest neighbor algorithm model to obtain switching time data corresponding to different loads. And then, carrying out relevant statistical arrangement on the obtained switching time data of each load. And inputting the sorted load switching time data into a trained BP neural network, and fitting the switching time probability distribution curve of each load. And finally, judging whether the identified load type is correct or not according to the fitted switching probability distribution curve of each load. The application can ensure the accuracy of non-resident user load identification, and fits the switching probability distribution curve of the non-resident load through an intelligent learning method to know the using habit of the non-resident user on different loads, thereby helping the load identification to further judge.
Description
Technical Field
The application belongs to the technical field of intelligent electricity utilization and non-invasive load identification, and particularly relates to a non-invasive non-resident user load identification method based on intelligent learning.
Background
Along with the rapid development of smart grid construction, the management of the power demand side is used as a management method for controlling loads and excavating electric energy by clients, so that the management method can be developed continuously, the power resources can be configured more reasonably and effectively in China, and the electric energy use efficiency of the terminal is improved effectively. Meanwhile, non-invasive load monitoring has a huge research space and is widely focused at home and abroad. But is currently applied to monitoring various electrical appliances of residential users, and related researches on non-residential user loads of important components of the power grid load are not much. Non-invasive load monitoring is generally divided into three main steps, event detection, feature extraction and load identification. However, the currently used load identification methods have the problem that partial load feature overlapping is difficult to identify, which also results in insufficient load identification accuracy in the current non-invasive load monitoring system.
Disclosure of Invention
The application provides a non-invasive non-resident user load identification method for solving the problem that partial load characteristics are overlapped and are difficult to identify in the prior art.
The application particularly provides a non-invasive non-resident user load identification method based on intelligent learning, which is characterized by comprising the following steps of:
step S1: collecting load characteristic sample data of each electric appliance in non-resident user data when switching events occur, and preprocessing the load characteristic sample data to be used as a training data set;
step S2: randomly selecting load characteristic sample data of a plurality of unknown electrical appliances when switching events occur from non-resident user data, and preprocessing the load characteristic sample data to be used as a test data set;
step S3: identifying the types of the unknown appliances through a K nearest neighbor classification algorithm model;
step S4: inputting the training data set into a BP neural network model until the training data set is exhausted or an error function meets the requirement;
step S5: inputting the test data set into a BP neural network model to obtain the relation between the switching probability and switching time of a plurality of electric appliances;
step S6: and analyzing whether the relation between the switching probability and the switching time of the electric appliances accords with the use habit of the corresponding electric appliances, and verifying whether the type of the electric appliances obtained by identification is correct.
Further, in step S1, the training data set performs fourier series expansion on the current data according to the steady-state working current data when switching events occur in each electric appliance in the collected non-resident user data, and takes the current harmonic amplitude with relatively large amplitude obtained after expansion as the training data set.
In step S2, the test data set performs fourier series expansion on the current data according to the steady-state working current data when switching events occur in several electric appliances in the collected non-resident user data, and takes the current harmonic amplitude with relatively large amplitude obtained after expansion as the test data set.
Further, in step S3, the method further includes the following steps:
step S31: inputting the training data set and the test data set into the K nearest neighbor classification algorithm model and setting the value of K;
step S32: calculating distance measurement by adopting weighted Euclidean distance, and obtaining K training data similar to the test data;
step S33: and judging the category of the test data as the category of training data which is closer to the test data, and obtaining a classification result.
Further, in step S32, the weighted euclidean distance is expressed as:
wherein ,wk Is the weight of each sample; x is x i Is the ith test data vector, x j Is the j-th training data vector, x i and xj Each having n characteristic components; x is x ik I.e. the kth characteristic component, x of the ith test data jk I.e. the kth feature component of the jth training data.
Further, in step S33, the classification result is expressed as:
wherein ,is the prediction classification result of the test data, x test Is test data, x' j Is and test data x test Training data of the j 'th nearest, f (x' j ) Is x' j Category labels of (c); y is i Is a class i load; δ is a 0,1 function, δ (a, b) =1 when a=b only, otherwise δ (a, b) =0; y is a vector of i×1, i is the total amount of load types; w (w) j Is the weight of the vote.
Further, in step S33, the weight of the vote is expressed as:
further, in step S5, the relationship between the switching probability and the switching time of the plurality of electric appliances is obtained by the test data set, the switching frequency distribution curves of the plurality of electric appliances are integrated to obtain the area under each curve, and the switching probability distribution diagram of the plurality of electric appliances is obtained by dividing the switching frequency distribution curves of the plurality of electric appliances by the area.
The beneficial effects of the application are as follows:
the application has remarkable advantages for solving the problems of various loads of non-resident users, larger load identification error and the like. The non-invasive load monitoring system is used for collecting electricity data of non-resident users so as to monitor and charge energy consumption of various non-resident loads in a time-sharing manner, and various characteristics of the loads are extracted from the data.
According to the application, the K nearest neighbor algorithm model is used for primarily classifying various non-resident user loads, the K0 nearest neighbor algorithm model classifier does not need to be trained by using a training set in advance, the training time complexity is 0, and the primary classification of various non-resident user loads is more convenient and faster.
According to the application, the load identification is carried out by utilizing the characteristics, so that a generalized intelligent learning model is established, a switching probability distribution curve of the non-resident load is fitted, so that the use habit of people on different loads is known, and further the load identification is helped to carry out further judgment. The method has higher accuracy, obviously improves the load identification accuracy, has simple principle and is convenient to implement.
Drawings
FIG. 1 is a schematic diagram of steps of a non-invasive non-resident user load identification method based on intelligent learning according to an embodiment of the present application;
fig. 2 is a current waveform diagram of a classroom lamp according to an embodiment of the present application;
FIG. 3 is a current waveform diagram of a projector according to an embodiment of the present application;
FIG. 4 is a graph showing the convergence trend of the loss function of the electric lamp in the classroom according to the embodiment of the present application;
FIG. 5 is a graph of the convergence trend of the loss function of the projector according to the embodiment of the present application;
fig. 6 is a probability distribution diagram of a classroom electric lamp according to an embodiment of the present application;
fig. 7 is a probability distribution diagram of a projector according to an embodiment of the present application.
Detailed Description
The technical scheme of the application is further specifically described below by means of examples and with reference to fig. 1-7.
As shown in fig. 1, an embodiment of the present application provides a non-invasive non-resident user load identification method based on intelligent learning, comprising the following steps:
step S1: collecting load characteristic sample data of each electric appliance in non-resident user data when switching events occur, and preprocessing the load characteristic sample data to be used as a training data set;
step S2: randomly selecting load characteristic sample data of a plurality of unknown electrical appliances when switching events occur from non-resident user data, and preprocessing the load characteristic sample data to be used as a test data set;
step S3: identifying the types of a plurality of unknown electrical appliances through a K nearest neighbor classification algorithm model;
step S4: inputting the training data set into a BP neural network model until the training data set is exhausted or an error function meets the requirement;
step S5: inputting the test data set into a BP neural network model to obtain the relation between the switching probability and switching time of a plurality of electric appliances;
step S6: and analyzing whether the relation between the switching probability and the switching time of the electric appliances accords with the use habit of the corresponding electric appliances, and verifying whether the type of the electric appliances obtained by identification is correct.
Specifically, in step S1, steady-state working current data of each electric appliance in the collected non-resident user data when a switching event occurs is extracted, fourier series expansion is performed on the current data, and a current harmonic amplitude with a relatively large amplitude obtained after expansion is used as a load characteristic sample to be used as a training data set.
In step S2, steady-state working current data of a plurality of unknown electrical appliances when switching events occur are randomly selected from the collected non-resident user data, fourier series expansion is carried out on the current data, and the current harmonic amplitude with relatively large amplitude obtained after expansion is used as a load characteristic sample to be used as a test data set.
In step S3, non-invasive load identification is performed using a K nearest neighbor classification algorithm, which is a distance-based intelligent learning method.
Since the model built with the K-nearest neighbor classification algorithm has no very obvious training process, it will not start to calculate yet after the training dataset is input, it will actually start to calculate only when the test dataset is input for classification. For an input test data set, the K nearest neighbor classification algorithm model firstly finds K training data closest to the test data, and finally judges the category of the test data by a voting method according to the category marks of the K training data.
The non-invasive load identification process using the K-nearest neighbor algorithm model further comprises the steps of:
step S31: inputting the training data set and the test data set into the K nearest neighbor classification algorithm model and setting the value of K.
Step S32: calculating distance measurement by adopting weighted Euclidean distance, and obtaining K training data similar to the test data; wherein the weighted euclidean distance is expressed as:
wherein ,wk Is the weight of each sample; x is x i Is the ith test data vector, x j Is the j-th training data vector, x i and xj Each having n characteristic components; x is x ik I.e. the kth characteristic component, x of the ith test data jk I.e. the kth feature component of the jth training data.
The K nearest neighbor classification algorithm model is weighted according to the distance between the test data and the adjacent training data, and the training data with the closer test data distance is weighted more heavily, and the optimal weight value is mainly selected by a traversing method. After the traversal is finished, K training data similar to the test data are obtained according to the weight.
Step S33: the test data is determined as the category of training data nearer to the test data, and the result of classification can be expressed as:
wherein ,is the prediction classification result of the test data, x test Is test data, x' j Is and test data x test Training data of the j 'th nearest, f (x' j ) Is x' j Category labels of (c); y is i Is a class i load; δ is a 0,1 function, δ (a, b) =1 when a=b only, otherwise δ (a, b) =0; y is a vector of i×1, i is the total amount of load types; w (w) j Is the weight of the vote:
the type of a plurality of unknown electrical appliances can be identified through the K nearest neighbor classification algorithm model. Since the K value has been determined in advance, K training data nearest to the test data are selected. The training data are classified and marked, y i The i-th class of load is represented only when the index f (x' j ) And y is i When the same is true, delta (y i ,f(x′ j ) 1) can be taken, otherwise 0; thus, y can be changed i I value in (2), respectively calculating y i Corresponding to the sum of the values of i. By passing throughThe function is that y can be selected i The largest one after summation in (a) if y 1 Then the test data can be discriminated as a first type of load.
In step S4, since most of the currently used load identification methods compare the extracted features with features in the feature library and then determine the closest to the extracted features as the load, there is a possibility that misjudgment occurs if the extracted features of several loads have similar features. The probability distribution of the load switching time data obtained by non-invasive load identification is studied, so that the use habit of a user on different electric appliances can be known, further judgment is further carried out on the result of load identification, and the accuracy of load identification is ensured.
BP neural network is used as a widely applied neural network algorithm, and has strong learning capacity and storage capacity for a large number of input-output mapping relations. In addition, since the BP neural network is an adaptive nonlinear dynamic system mainly composed of a large number of neurons, it has a very good effect on fitting a nonlinear function curve. Therefore, the application adopts BP neural network algorithm to fit the load switching probability distribution curve.
And inputting the training data set into the BP neural network model, performing 10000 times of iterative training on the BP neural network model, and continuously adjusting until the training data is an exhaustive set or the error function meets the requirement, wherein the actual output is closest to the expected result.
In step S5, the test data set is input into the trained BP neural network model to perform 10000 times of iterative operation, and the weights among the neural network units are continuously adjusted by a gradient descent method to select the optimal weight, so that the output result is as close to the expected value as possible, the problem of error correction in the BP neural network model is solved, and the switching frequency distribution curve of several electrical appliances can be obtained. And normalizing the switching frequency distribution curve graphs of the electric appliances, namely integrating the switching frequency distribution curve graphs of the electric appliances to obtain the area under each curve, dividing the obtained area by the switching frequency distribution curve of each electric appliance, and obtaining the switching probability distribution curve graphs of the electric appliances through normalization.
In step S6, it is analyzed whether the switching probability distribution curve graphs of several electric appliances conform to the usage habits of the corresponding electric appliances. In this way, the obtained switching probability distribution curve graph of several electric appliances is analyzed, wherein the abscissa of the graph is 24 hours a day, and the ordinate of the graph is probability density, so that the probability of the electric appliance being used in each time period is the area under the curve corresponding to the time period. The result identified by the K nearest neighbor algorithm is judged again through the obtained probability distribution curve graph, so that the misjudgment condition of the K nearest neighbor algorithm is corrected, and the accuracy of load identification is ensured.
2-3, in one embodiment, collecting steady-state working current data of each electric appliance in a school when switching events occur, expanding through Fourier series, and selecting current harmonic amplitude with larger amplitude as training data to be input into an in-school load identification model established by a K nearest neighbor algorithm; and the steady-state working current data of two electrical appliances in the school when switching events occur are randomly collected, and the current harmonic amplitude with larger amplitude is expanded through Fourier series and selected as test data to be input into a load identification model in the school built by a K nearest neighbor algorithm. Two kinds of unknown electrical appliances are identified through a K nearest neighbor classification algorithm model: classroom electric lights and projectors.
As shown in fig. 4-5, training data is input into a BP neural network model for training, and the obtained loss function loss convergence trend chart of each electric appliance is iterated 10000 times, wherein the loss function loss tends to 0, and at the moment, the optimal effect can be obtained by inputting test data.
As shown in fig. 6-7, the test data is input into a trained BP neural network model to perform iterative operation to obtain the switching frequency distribution curves of the two electric appliances, and the switching frequency distribution curves of the two electric appliances are normalized to obtain the switching probability distribution curves of the two electric appliances.
The classroom electric lamp is analyzed firstly, the time period that the electric lamp is basically used in one day can be found to be in an irregular law, the probability that the electric lamp is used after six points at night is relatively high in normal practice, and the electric lamp is rarely used or even not used in other time periods, so that the electric appliance corresponding to the switching probability distribution curve graph is not the classroom electric lamp, misjudgment can be judged when the K nearest neighbor algorithm is utilized for load identification, the electric appliance is classified into a refrigerator of a second-most-class electric appliance-school supermarket found when the load identification is carried out, and the fact that the time period that the electric refrigerator of the school supermarket is used is in a condition identical with the situation represented by the switching probability distribution curve can be found. And by the same method, the time of the projector being used accords with teaching habits, and the projector is frequently used in the morning, afternoon and evening. Therefore, the application can be obtained to improve the accuracy of load identification according to the use habit of non-resident users on the electric appliance.
While the application has been disclosed in terms of preferred embodiments, the embodiments are not intended to limit the application. Any equivalent changes or modifications can be made without departing from the spirit and scope of the present application, and are intended to be within the scope of the present application. The scope of the application should therefore be determined by the following claims.
Claims (2)
1. The non-invasive non-resident user load identification method based on intelligent learning is characterized by comprising the following steps of:
step S1: collecting load characteristic sample data of each electric appliance in non-resident user data when switching events occur, and preprocessing the load characteristic sample data to be used as a training data set;
step S2: randomly selecting load characteristic sample data of a plurality of unknown electrical appliances when switching events occur from non-resident user data, and preprocessing the load characteristic sample data to be used as a test data set;
step S3: identifying the types of the unknown appliances through a K nearest neighbor classification algorithm model;
step S4: inputting the training data set into a BP neural network model until the training data set is exhausted or an error function meets the requirement;
step S5: inputting the test data set into a BP neural network model to obtain the relation between the switching probability and switching time of a plurality of electric appliances;
step S6: analyzing whether the relation between the switching probability and the switching time of a plurality of electric appliances accords with the use habit of the corresponding electric appliances, and verifying whether the type of the electric appliances obtained by identification is correct;
in step S2, the test data set performs fourier series expansion on the current data according to the collected steady-state working current data when switching events occur in a plurality of electric appliances in the non-resident user data, and takes the current harmonic amplitude with relatively large amplitude obtained after expansion as the test data set;
in step S3, the method further includes the following steps:
step S31: inputting the training data set and the test data set into the K nearest neighbor classification algorithm model and setting the value of K;
step S32: calculating distance measurement by adopting weighted Euclidean distance, and obtaining K training data similar to the test data;
step S33: judging the category of the test data as the category of training data which is closer to the test data, and obtaining a classification result;
in step S32, the weighted euclidean distance is expressed as:
wherein ,wk Is the weight of each sample; x is x i Is the ith test data vector, x j Is the j-th training data vector, x i and xj Each having n characteristic components; x is x ik I.e. the kth characteristic component, x of the ith test data jk I.e., the kth feature component of the jth training data;
in step S33, the classification result is expressed as:
wherein ,is the prediction classification result of the test data, x test Is test data, x' j Is and test data x test Training data of the j 'th nearest, f (x' j ) Is x' j Category labels of (c); y is i Is a class i load; δ is a 0,1 function, δ (a, b) =1 when a=b only, otherwise δ (a, b) =0; y is a vector of i×1, i is the total amount of load types; w (w) j Is the weight of the vote;
in step S33, the weight of the vote is expressed as:
in step S5, the relationship between the switching probability and the switching time of the electrical appliances is obtained by the test data set, the switching frequency distribution graph of the electrical appliances is obtained, the switching frequency distribution graph of the electrical appliances is integrated to obtain the area under each curve, and the switching probability distribution graph of the electrical appliances is obtained by dividing the switching frequency distribution graph of the electrical appliances by the area.
2. The non-invasive non-resident user load identification method based on intelligent learning according to claim 1, wherein in step S1, the training data set performs fourier series expansion on the current data through steady-state working current data when switching events occur in each electric appliance in the collected non-resident user data, and the current harmonic amplitude with relatively large amplitude obtained after expansion is used as the training data set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911375093.5A CN111242161B (en) | 2019-12-27 | 2019-12-27 | Non-invasive non-resident user load identification method based on intelligent learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911375093.5A CN111242161B (en) | 2019-12-27 | 2019-12-27 | Non-invasive non-resident user load identification method based on intelligent learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111242161A CN111242161A (en) | 2020-06-05 |
CN111242161B true CN111242161B (en) | 2023-08-25 |
Family
ID=70865575
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911375093.5A Active CN111242161B (en) | 2019-12-27 | 2019-12-27 | Non-invasive non-resident user load identification method based on intelligent learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111242161B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112446416A (en) * | 2020-10-09 | 2021-03-05 | 国网山东省电力公司日照供电公司 | Non-invasive load monitoring and identifying method and system |
CN112381264A (en) * | 2020-10-20 | 2021-02-19 | 昆明理工大学 | Electricity load identification method based on improved graph convolution neural network |
CN113033775B (en) * | 2021-03-10 | 2023-08-18 | 南方电网数字电网研究院有限公司 | Non-invasive load identification network architecture based on supervised learning |
CN112989131B (en) * | 2021-03-29 | 2023-04-07 | 广州水沐青华科技有限公司 | Graph representation learning electric appliance equipment power fingerprint decomposition method, storage medium and system |
CN114977176B (en) * | 2022-07-19 | 2022-11-18 | 深圳江行联加智能科技有限公司 | Power load decomposition method, device, equipment and storage medium |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105972761A (en) * | 2016-05-25 | 2016-09-28 | 华北电力大学(保定) | Non-invasive air conditioner load monitoring method |
CN106908671A (en) * | 2017-01-13 | 2017-06-30 | 中国海洋大学 | A kind of non-intrusion type household loads intelligent detecting method and system |
CN108054749A (en) * | 2017-11-27 | 2018-05-18 | 华北电力大学 | A kind of non-intrusive electrical load decomposition method and device |
WO2018099418A1 (en) * | 2016-11-30 | 2018-06-07 | The Chinese University Of Hong Kong | Analysis of cell-free dna in urine and other samples |
CN109145949A (en) * | 2018-07-19 | 2019-01-04 | 山东师范大学 | Non-intrusive electrical load monitoring and decomposition method and system based on integrated study |
WO2019025938A1 (en) * | 2017-08-04 | 2019-02-07 | 3M Innovative Properties Company | Confined space configuration and operations management system |
CN109752613A (en) * | 2018-12-31 | 2019-05-14 | 天津求实智源科技有限公司 | Default electricity use detection system and method based on non-intrusion type load monitoring |
CN109767054A (en) * | 2018-11-22 | 2019-05-17 | 福建网能科技开发有限责任公司 | Efficiency cloud appraisal procedure and edge efficiency gateway based on deep neural network algorithm |
CN110376457A (en) * | 2019-06-28 | 2019-10-25 | 同济大学 | Non-intrusion type load monitoring method and device based on semi-supervised learning algorithm |
CN110443302A (en) * | 2019-08-02 | 2019-11-12 | 天津相和电气科技有限公司 | Load discrimination method and its application based on Fusion Features and deep learning |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050037515A1 (en) * | 2001-04-23 | 2005-02-17 | Nicholson Jeremy Kirk | Methods for analysis of spectral data and their applications osteoporosis |
US8664937B2 (en) * | 2011-02-18 | 2014-03-04 | Honeywell International Inc. | Non-invasive energy consumption monitoring |
US20140207398A1 (en) * | 2013-01-23 | 2014-07-24 | Samsung Electronics Co., Ltd | Transient Normalization for Appliance Classification, Disaggregation, and Power Estimation in Non-Intrusive Load Monitoring |
-
2019
- 2019-12-27 CN CN201911375093.5A patent/CN111242161B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105972761A (en) * | 2016-05-25 | 2016-09-28 | 华北电力大学(保定) | Non-invasive air conditioner load monitoring method |
WO2018099418A1 (en) * | 2016-11-30 | 2018-06-07 | The Chinese University Of Hong Kong | Analysis of cell-free dna in urine and other samples |
CN106908671A (en) * | 2017-01-13 | 2017-06-30 | 中国海洋大学 | A kind of non-intrusion type household loads intelligent detecting method and system |
WO2019025938A1 (en) * | 2017-08-04 | 2019-02-07 | 3M Innovative Properties Company | Confined space configuration and operations management system |
CN108054749A (en) * | 2017-11-27 | 2018-05-18 | 华北电力大学 | A kind of non-intrusive electrical load decomposition method and device |
CN109145949A (en) * | 2018-07-19 | 2019-01-04 | 山东师范大学 | Non-intrusive electrical load monitoring and decomposition method and system based on integrated study |
CN109767054A (en) * | 2018-11-22 | 2019-05-17 | 福建网能科技开发有限责任公司 | Efficiency cloud appraisal procedure and edge efficiency gateway based on deep neural network algorithm |
CN109752613A (en) * | 2018-12-31 | 2019-05-14 | 天津求实智源科技有限公司 | Default electricity use detection system and method based on non-intrusion type load monitoring |
CN110376457A (en) * | 2019-06-28 | 2019-10-25 | 同济大学 | Non-intrusion type load monitoring method and device based on semi-supervised learning algorithm |
CN110443302A (en) * | 2019-08-02 | 2019-11-12 | 天津相和电气科技有限公司 | Load discrimination method and its application based on Fusion Features and deep learning |
Non-Patent Citations (1)
Title |
---|
宋旭帆 ; 周明 ; 涂京 ; 李庚银 ; .基于k-NN结合核Fisher判别的非侵入式负荷监测方法.电力系统自动化.2018,(第06期),第73页-80页. * |
Also Published As
Publication number | Publication date |
---|---|
CN111242161A (en) | 2020-06-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111242161B (en) | Non-invasive non-resident user load identification method based on intelligent learning | |
Liu et al. | Energy disaggregation of appliances consumptions using ham approach | |
CN108280479B (en) | Power grid user classification method based on load characteristic index weighted clustering algorithm | |
CN109492667A (en) | A kind of feature selecting discrimination method for non-intrusive electrical load monitoring | |
CN110991786B (en) | 10kV static load model parameter identification method based on similar daily load curve | |
CN109598451B (en) | Non-invasive load identification method based on PCA (principal component analysis) and LSTM (least Square TM) neural network | |
CN104809658B (en) | A kind of rapid analysis method of low-voltage distribution network taiwan area line loss | |
CN111368904B (en) | Electrical equipment identification method based on electric power fingerprint | |
CN106096805A (en) | A kind of residential electricity consumption load classification method based on entropy assessment feature selection | |
CN112149873B (en) | Low-voltage station line loss reasonable interval prediction method based on deep learning | |
CN109146252A (en) | Based on the user power utilization behavior clustering method and system for adjusting potentiality index | |
CN105740635B (en) | A kind of cloud ideal solution evaluation method of transformer electromagnetic design scheme | |
CN112580471A (en) | Non-invasive load identification method based on AdaBoost feature extraction and RNN model | |
CN111489188A (en) | Resident adjustable load potential mining method and system | |
CN113780684A (en) | Intelligent building user energy consumption behavior prediction method based on LSTM neural network | |
CN111461921A (en) | Load modeling typical user database updating method based on machine learning | |
CN113887912A (en) | Non-invasive load identification method for deeply learning downward embedded equipment | |
CN110503145A (en) | A kind of typical load curve acquisition methods based on k-shape cluster | |
CN112508254B (en) | Method for determining investment prediction data of transformer substation engineering project | |
CN104360948A (en) | IEC 61850 configuration file engineering consistency test method based on fuzzy algorithm | |
CN111768066B (en) | Park electric heating load coupling relation analysis method and device based on fusion characteristics | |
CN114186631A (en) | Load identification method based on non-invasive intelligent terminal | |
CN106355515A (en) | Power load sorting method and system | |
Wang et al. | Identifying household EV models via weighted power recurrence graphs | |
Jangid et al. | Load Profile Segmentation of Various Load Categories using Clustering |
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 |