CN116051910A - Non-invasive load identification method, device, terminal equipment and storage medium - Google Patents
Non-invasive load identification method, device, terminal equipment and storage medium Download PDFInfo
- Publication number
- CN116051910A CN116051910A CN202310226288.3A CN202310226288A CN116051910A CN 116051910 A CN116051910 A CN 116051910A CN 202310226288 A CN202310226288 A CN 202310226288A CN 116051910 A CN116051910 A CN 116051910A
- Authority
- CN
- China
- Prior art keywords
- current data
- load
- model
- load identification
- invasive
- 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 68
- 238000013145 classification model Methods 0.000 claims abstract description 54
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 53
- 238000012549 training Methods 0.000 claims description 39
- 230000005611 electricity Effects 0.000 claims description 25
- 238000013528 artificial neural network Methods 0.000 claims description 14
- 238000004422 calculation algorithm Methods 0.000 claims description 14
- 238000013527 convolutional neural network Methods 0.000 claims description 13
- 230000005856 abnormality Effects 0.000 claims description 11
- 238000007635 classification algorithm Methods 0.000 claims description 11
- 125000004122 cyclic group Chemical group 0.000 claims description 11
- 238000005070 sampling Methods 0.000 claims description 11
- 238000004891 communication Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 10
- 230000002159 abnormal effect Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 description 7
- 238000012544 monitoring process Methods 0.000 description 6
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000005684 electric field Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Biophysics (AREA)
- Water Supply & Treatment (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Public Health (AREA)
- Computational Linguistics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a non-invasive load identification method, a device, a terminal device and a storage medium, wherein the non-invasive load identification method comprises the following steps: acquiring total load current data; decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data; obtaining respective corresponding time sequence images according to at least one single load current data; and classifying the time sequence images based on a preset classification model to obtain a load identification result. Based on the scheme, classification based on the time sequence image can effectively improve the accuracy of non-invasive load identification.
Description
Technical Field
The present disclosure relates to the field of power demand side management technologies, and in particular, to a non-invasive load identification method, device, terminal equipment, and storage medium.
Background
Nowadays, the application of non-invasive load identification technology on user demand side management is becoming wider and wider, and related information of each electric load can be obtained through collecting and analyzing total load data, namely, the running condition of each electric equipment is obtained.
Typical non-invasive load identification procedures include data acquisition, data preprocessing, load classification, and decomposition. However, due to the high complexity of practical electric field scenarios, the accuracy of traditional non-invasive load identification methods is not high.
Disclosure of Invention
The main purpose of the application is to provide a non-invasive load identification method, a device, a terminal device and a storage medium, which aim to solve the problem of low accuracy of the traditional non-invasive load identification method.
To achieve the above object, the present application provides a non-invasive load identification method, including:
acquiring total load current data;
decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data;
obtaining respective corresponding time sequence images according to the at least one single load current data;
and classifying the time sequence images based on a preset classification model to obtain a load identification result.
Optionally, before the step of acquiring the total load current data, the method further includes:
based on a preset high-frequency sampling rule, acquiring single-load current data samples corresponding to a plurality of electric equipment respectively, wherein each single-load current data sample comprises current data of a plurality of periods;
selecting current data of one period from the current data of the plurality of periods as single-period current data;
model training is carried out based on the single-period current data corresponding to the electric equipment respectively to obtain an identification model, wherein the type of the identification model comprises the classification model and the decomposition model.
Optionally, the step of performing model training based on the monocycle current data corresponding to each of the plurality of electric devices to obtain the identification model includes:
drawing a corresponding time sequence image sample according to the single-period current data corresponding to each of the plurality of electric devices;
model training is carried out based on a preset convolutional neural network algorithm and time sequence image samples corresponding to the electric equipment respectively, and the classification model is obtained.
Optionally, the step of performing model training based on the monocycle current data corresponding to each of the plurality of electric devices to obtain the identification model includes:
randomly combining the single-period current data corresponding to each of the plurality of electric devices to obtain a mixed current data sample;
model training is carried out based on a preset cyclic neural network algorithm and the mixed current data sample, and the decomposition model is obtained.
Optionally, the classification model further includes a dynamic load classification algorithm, and the step of classifying the time-series image based on a preset classification model to obtain a load recognition result includes:
and classifying the time sequence images based on the dynamic load classification algorithm to obtain a load identification result corresponding to the electric equipment with the load dynamically changed.
Optionally, the non-invasive load identification method is applied to an edge computing device, the edge computing device establishes communication connection with at least one intelligent breaker, the step of classifying the time sequence image based on a preset classification model to obtain a load identification result further includes:
judging whether the load identification result has load abnormality or not;
if yes, controlling the intelligent circuit breaker corresponding to the abnormal load to execute a power-off task.
Optionally, the step of establishing communication connection between the edge computing device and the intelligent electricity management platform, and judging whether the load identification result has a load abnormality further includes:
if not, uploading the load identification result to the intelligent electricity consumption management platform so that the intelligent electricity consumption management platform can monitor electricity consumption of the electric equipment connected with the edge computing equipment.
The embodiment of the application also provides a non-invasive load identification device, which comprises:
the acquisition module is used for acquiring the total load current data;
the decomposition module is used for decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data;
the drawing module is used for obtaining respective corresponding time sequence images according to the at least one single-load current data;
and the classification module is used for carrying out classification processing on the time sequence images based on a preset classification model to obtain a load identification result.
The embodiment of the application also provides a terminal device, which comprises a memory, a processor and a non-invasive load identification program stored on the memory and capable of running on the processor, wherein the non-invasive load identification program realizes the steps of the non-invasive load identification method when being executed by the processor.
The embodiments of the present application also propose a computer readable storage medium on which a non-invasive load identification program is stored, which when executed by a processor implements the steps of the non-invasive load identification method as described above.
The non-invasive load identification method, the non-invasive load identification device, the terminal equipment and the storage medium provided by the embodiment of the application are realized by acquiring total load current data; decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data; obtaining respective corresponding time sequence images according to the at least one single load current data; and classifying the time sequence images based on a preset classification model to obtain a load identification result. Based on the scheme, the method comprises the steps of firstly obtaining total load current data, decomposing at least one single load current data based on a decomposition model, drawing a corresponding time sequence image according to the single load current data, and taking the time sequence image as input of a classification model to obtain a load identification result. Therefore, classification based on time sequence images can effectively improve the accuracy of non-invasive load identification.
Drawings
FIG. 1 is a schematic diagram of functional modules of a terminal device to which a non-invasive load identification apparatus of the present application belongs;
FIG. 2 is a flow chart of a first exemplary embodiment of a non-intrusive load identification method of the present application;
FIG. 3 is a flow chart of a second exemplary embodiment of a non-intrusive load identification method of the present application;
FIG. 4 is a flow chart of a third exemplary embodiment of a non-intrusive load identification method of the present application;
FIG. 5 is a flow chart of a fourth exemplary embodiment of a non-intrusive load identification method of the present application;
FIG. 6 is a flow chart of a fifth exemplary embodiment of a non-intrusive load identification method of the present application;
FIG. 7 is a flowchart of a sixth exemplary embodiment of a non-intrusive load identification method of the present application;
fig. 8 is a flowchart of a seventh exemplary embodiment of a non-invasive load identification method according to the present application.
The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
The main solutions of the embodiments of the present application are: acquiring total load current data; decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data; obtaining respective corresponding time sequence images according to the at least one single load current data; and classifying the time sequence images based on a preset classification model to obtain a load identification result. Based on the scheme, the method comprises the steps of firstly obtaining total load current data, decomposing at least one single load current data based on a decomposition model, drawing a corresponding time sequence image according to the single load current data, and taking the time sequence image as input of a classification model to obtain a load identification result. Therefore, classification based on time sequence images can effectively improve the accuracy of non-invasive load identification.
Specifically, referring to fig. 1, fig. 1 is a schematic functional block diagram of a terminal device to which a non-invasive load identification apparatus of the present application belongs. The non-invasive load identification means may be a device independent of the terminal equipment capable of non-invasive load identification, which may be carried on the terminal equipment in the form of hardware or software. The terminal equipment can be an intelligent mobile terminal with a data processing function such as a mobile phone and a tablet personal computer, and can also be a fixed terminal equipment or a server with a data processing function.
In this embodiment, the terminal device to which the non-invasive load identification apparatus belongs at least includes an output module 110, a processor 120, a memory 130, and a communication module 140.
The memory 130 stores an operating system and a non-invasive load recognition program, and the non-invasive load recognition device can acquire total load current data; decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data; respective corresponding time sequence images obtained according to at least one single load current data; classifying the time sequence images based on a preset classification model, and storing the obtained information such as load identification results and the like in the memory 130; the output module 110 may be a display screen or the like. The communication module 140 may include a WIFI module, a mobile communication module, a bluetooth module, and the like, and communicates with an external device or a server through the communication module 140.
Wherein the non-intrusive load identification program in the memory 130 when executed by the processor performs the steps of:
acquiring total load current data;
decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data;
obtaining respective corresponding time sequence images according to the at least one single load current data;
and classifying the time sequence images based on a preset classification model to obtain a load identification result.
Further, the non-invasive load identification procedure in the memory 130, when executed by the processor, also implements the steps of:
based on a preset high-frequency sampling rule, acquiring single-load current data samples corresponding to a plurality of electric equipment respectively, wherein each single-load current data sample comprises current data of a plurality of periods;
selecting current data of one period from the current data of the plurality of periods as single-period current data;
model training is carried out based on the single-period current data corresponding to the electric equipment respectively to obtain an identification model, wherein the type of the identification model comprises the classification model and the decomposition model.
Further, the non-invasive load identification procedure in the memory 130, when executed by the processor, also implements the steps of:
drawing a corresponding time sequence image sample according to the single-period current data corresponding to each of the plurality of electric devices;
model training is carried out based on a preset convolutional neural network algorithm and time sequence image samples corresponding to the electric equipment respectively, and the classification model is obtained.
Further, the non-invasive load identification procedure in the memory 130, when executed by the processor, also implements the steps of:
randomly combining the single-period current data corresponding to each of the plurality of electric devices to obtain a mixed current data sample;
model training is carried out based on a preset cyclic neural network algorithm and the mixed current data sample, and the decomposition model is obtained.
Further, the non-invasive load identification procedure in the memory 130, when executed by the processor, also implements the steps of:
and classifying the time sequence images based on the dynamic load classification algorithm to obtain a load identification result corresponding to the electric equipment with the load dynamically changed.
Further, the non-invasive load identification procedure in the memory 130, when executed by the processor, also implements the steps of:
judging whether the load identification result has load abnormality or not;
if yes, controlling the intelligent circuit breaker corresponding to the abnormal load to execute a power-off task.
Further, the non-invasive load identification procedure in the memory 130, when executed by the processor, also implements the steps of:
if not, uploading the load identification result to the intelligent electricity consumption management platform so that the intelligent electricity consumption management platform can monitor electricity consumption of the electric equipment connected with the edge computing equipment.
According to the scheme, the embodiment specifically obtains the total load current data; decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data; obtaining respective corresponding time sequence images according to the at least one single load current data; and classifying the time sequence images based on a preset classification model to obtain a load identification result. In this embodiment, the total load current data is first obtained, at least one single load current data is decomposed from the total load current data based on the decomposition model, a corresponding time sequence image is further drawn according to the single load current data, and the time sequence image is used as the input of the classification model, so that the load recognition result can be obtained. Therefore, classification based on time sequence images can effectively improve the accuracy of non-invasive load identification.
Referring to fig. 2, a first embodiment of a non-invasive load identification method according to the present application provides a flowchart, where the non-invasive load identification method includes:
step S10, acquiring total load current data.
Specifically, the non-invasive load identification method according to the present embodiment may be applied to an edge computing device, for example, a smart meter or a dedicated non-invasive load identification device, and the edge computing device is used for load identification of a specific power grid.
In a certain area range, multiple electric devices can exist at the same time, and the load characteristics of different electric devices are different. For example, a common resident house may have multiple electric devices, such as an air conditioner, a television, a refrigerator, a computer, an electric vehicle charger, and the like, located in the same home network. The edge computing device first obtains the total load current data of the specific grid for subsequent non-intrusive load identification.
And step S20, decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data.
The total load current data represents the load condition of all electric equipment in the corresponding power grid, and the load current data corresponding to each electric equipment, namely single load current data, needs to be extracted from the total load current data. Specifically, the edge computing device is pre-provided with a decomposition model, and the decomposition model is obtained based on machine learning training. The total load current data is used as the input of a decomposition model, and at least one single load current data (in the case of electricity consumption) can be obtained through decomposition processing of the decomposition model.
For example, the total load current data is decomposed based on a decomposition model to obtain single load current data A1, single load current data A2, and single load current data A3. The single load current data A1, the single load current data A2 and the single load current data A3 respectively correspond to the electric equipment B1, the electric equipment B2 and the electric equipment B3.
And step S30, obtaining respective corresponding time sequence images according to the at least one single-load current data.
Specifically, after at least one single load current data is obtained through decomposition, respectively corresponding time sequence images are drawn according to each single load current data, wherein the time sequence images reflect the condition that the load of corresponding electric equipment changes along with time, and the conditions comprise current characteristics and time characteristics. More specifically, each single load current data corresponds to one time series image.
And step S40, classifying the time sequence images based on a preset classification model to obtain a load identification result.
Specifically, the edge computing device is pre-provided with a classification model, and the decomposition model is obtained based on machine learning training. The time sequence images corresponding to the single load current data are used as the input of the classification model, and the classification processing of the classification model can identify the electric equipment corresponding to each time sequence image (or each single load current data), namely, which electric equipment exists in the power grid. On the basis, the real-time load condition corresponding to each electric equipment can be obtained by combining the single load current data obtained by the decomposition.
The embodiment provides the edge computing equipment supporting the non-invasive load identification, a user can identify and monitor the load of a certain area or a power grid without installing a load measuring device for each electric equipment, the cost of load identification and monitoring is effectively reduced, and the trouble of modifying the existing line is avoided.
According to the scheme, the embodiment specifically obtains the total load current data; decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data; obtaining respective corresponding time sequence images according to the at least one single load current data; and classifying the time sequence images based on a preset classification model to obtain a load identification result. In this embodiment, the total load current data is first obtained, at least one single load current data is decomposed from the total load current data based on the decomposition model, a corresponding time sequence image is further drawn according to the single load current data, and the time sequence image is used as the input of the classification model, so that the load recognition result can be obtained. Therefore, classification based on time sequence images can effectively improve the accuracy of non-invasive load identification.
Further, referring to fig. 3, a flow chart is provided in a second embodiment of the non-invasive load identification method according to the present application, based on the embodiment shown in fig. 2, step S10, before obtaining the total load current data, further includes:
step S001, based on a preset high-frequency sampling rule, obtaining single-load current data samples corresponding to a plurality of electric equipment respectively, wherein each single-load current data sample comprises current data of a plurality of periods.
The present embodiment is mainly used for training a recognition model, and the training process is generally completed by a special device (such as a PC, a server, etc.) other than an edge computing device. Specifically, firstly, a plurality of electric devices are started, and the current data of the plurality of electric devices are sampled in a high-frequency sampling mode, so that various corresponding single-load current data samples of the plurality of electric devices are obtained. It is noted that for the accuracy of the sampling result, the sampling time will generally exceed a single current data period of the powered device, i.e. each single load current data sample actually comprises several periods of current data.
Compared with the low-frequency sampling, the high-frequency sampling can more comprehensively capture the load characteristic information of the electric equipment.
Step S002, selecting the current data of one period from the current data of the plurality of periods as single period current data.
Specifically, a large amount of redundant information exists in the single-load current data sample, and only a single period of current data is needed in the training process of the identification model. For this purpose, current data of one cycle is selected from the single load current data samples as single cycle current data, and current data of other cycles is deleted.
And step S003, performing model training based on the single-period current data corresponding to the electric equipment to obtain an identification model, wherein the type of the identification model comprises the classification model and the decomposition model.
Specifically, after obtaining single-period current data corresponding to each of a plurality of electric devices, building a corresponding training set by utilizing the single-period current data corresponding to each of the plurality of electric devices, performing model training based on the training set to obtain a corresponding identification model, wherein the type of the identification model comprises a classification model and a decomposition model,
further, the classification model and the decomposition model are migrated to the edge computing device to enable the edge computing device to support non-intrusive load identification functionality.
According to the scheme, particularly, based on a preset high-frequency sampling rule, single-load current data samples corresponding to a plurality of electric equipment are obtained, wherein each single-load current data sample comprises current data of a plurality of periods; selecting current data of one period from the current data of the plurality of periods as single-period current data; model training is carried out based on the single-period current data corresponding to the electric equipment respectively to obtain an identification model, wherein the type of the identification model comprises the classification model and the decomposition model. In this embodiment, a high-frequency sampling manner is adopted to obtain single-load current data samples corresponding to a plurality of electric devices, so that load characteristic information of the electric devices can be more comprehensively captured, current data of other periods except single-period current data are deleted, data redundancy can be reduced, and efficiency and effect of model training are improved.
Further, referring to fig. 4, a flow chart is provided in a third embodiment of the non-invasive load identification method of the present application, based on the embodiment shown in fig. 3, step S003, model training is performed based on the monocycle current data corresponding to each of the plurality of electric devices, so as to obtain an identification model for further refinement, including:
step S0031, drawing to obtain a corresponding time sequence image sample according to the single-period current data corresponding to each of the plurality of electric equipment.
Specifically, in a period of one current data, current characteristics of electric equipment change along with time change, and after obtaining single-period current data corresponding to each of a plurality of electric equipment, the embodiment draws corresponding time sequence image samples based on the current characteristics and the time characteristics of the single-period current data.
Step S0032, performing model training based on a preset convolutional neural network algorithm and time sequence image samples corresponding to the electric equipment respectively to obtain the classification model.
For different electric equipment, the corresponding time sequence image samples are different, so that the time sequence image samples can be used for training a classification model. Specifically, a convolutional neural network algorithm is preset, and a convolutional neural network (Convolutional Neural Networks, CNN) is a type of feedforward neural network which includes convolutional calculation and has a depth structure. Convolutional neural networks have a characteristic learning capability and can perform translation-invariant classification on input information according to a hierarchical structure of the convolutional neural networks, so the convolutional neural networks are also called as 'translation-invariant artificial neural networks'.
Further, a classification model training set is constructed according to time sequence image samples corresponding to the electric equipment respectively, model training is conducted on the basis of a preset convolutional neural network algorithm and the classification model training set, and a classification model is obtained.
According to the scheme, the corresponding time sequence image sample is drawn according to the single-period current data corresponding to the electric equipment; model training is carried out based on a preset convolutional neural network algorithm and time sequence image samples corresponding to the electric equipment respectively, and the classification model is obtained. In the embodiment, the time sequence image samples corresponding to the electric equipment are drawn, the classification model is trained based on the time sequence image samples, the training efficiency of the classification model can be effectively improved, and the accuracy of the classification result of the classification model is improved.
Further, referring to fig. 5, a flow chart is provided in a fourth embodiment of the non-invasive load identification method of the present application, based on the embodiment shown in fig. 3, step S003, model training is performed based on the monocycle current data corresponding to each of the plurality of electric devices, so as to obtain an identification model for further refinement, including:
step S0033, randomly combining the single-period current data corresponding to each of the plurality of electric devices to obtain a mixed current data sample.
Specifically, in the actual electric field scene, multiple electric devices often use electricity simultaneously, and if the decomposition model is trained by adopting single-period current data corresponding to a single electric device, the effectiveness of the decomposition model may be reduced. Therefore, the embodiment randomly combines the single-period current data corresponding to the electric equipment to obtain a mixed current data sample, and the mixed current data sample reflects more complicated electricity consumption conditions.
And step S0034, performing model training based on a preset cyclic neural network algorithm and the mixed current data sample to obtain the decomposition model.
Specifically, a cyclic neural network algorithm is preset, and the cyclic neural network (Recurrent Neural Network, RNN) is a recursive neural network which takes sequence data as input, performs recursion in the evolution direction of the sequence, and all nodes (cyclic units) are connected in a chained manner. The cyclic neural network has memory, parameter sharing and complete graphics, so that the cyclic neural network has certain advantages in learning the nonlinear characteristics of the sequence.
Further, a decomposition model training set is constructed according to the mixed current data sample, model training is carried out based on a preset cyclic neural network algorithm and the decomposition model training set, and a decomposition model is obtained.
According to the scheme, specifically, the single-period current data corresponding to the electric equipment are randomly combined to obtain a mixed current data sample; model training is carried out based on a preset cyclic neural network algorithm and the mixed current data sample, and the decomposition model is obtained. In this embodiment, the single-period current data corresponding to each of the plurality of electric devices is randomly combined to obtain the mixed current data sample, and the decomposition model is obtained based on the training of the mixed current data sample, so that the decomposition model can be effectively compatible with various complicated electricity utilization environments, and the accuracy of the decomposition result of the decomposition model is improved.
Further, referring to fig. 6, a flow chart is provided in a fifth embodiment of the non-invasive load identification method of the present application, based on the embodiment shown in fig. 2, the classification model further includes a dynamic load classification algorithm, and step S40, performing classification processing on the time-series image based on a preset classification model, to obtain further refinement of the load identification result, where the method includes:
and step S401, classifying the time sequence images based on the dynamic load classification algorithm to obtain a load identification result corresponding to the electric equipment with the load dynamically changed.
Considering that the dynamic load characteristics of some electric devices are easier to identify than the transient load characteristics, for example, the charging current of an electric vehicle charger gradually decreases with time, the embodiment optimizes the load identification of the electric devices. Specifically, the classification model further comprises a preset dynamic load classification algorithm, the time sequence image is used as the input of the classification model, and the dynamic load classification algorithm with the classification model performs classification processing, so that a load identification result corresponding to the electric equipment with the load dynamically changed can be obtained. As with the electric vehicle charger exemplified above, the electric vehicle charger is determined only if the corresponding time-series image reflects that the charging current gradually decreases with the lapse of time and satisfies the preset condition.
According to the scheme, the time sequence images are classified based on the dynamic load classification algorithm, so that a load identification result corresponding to the electric equipment with the load dynamically changed is obtained. The embodiment improves the electric equipment with obvious dynamic load characteristics, adopts a dynamic load classification algorithm to classify the time sequence images, can accurately identify the electric equipment with obvious dynamic load characteristics such as an electric vehicle charger and the like, and improves the accuracy of the classification result of the classification model.
Further, referring to fig. 7, a flowchart is provided in a sixth embodiment of the non-invasive load identification method according to the present application, based on the embodiment shown in fig. 2, the non-invasive load identification method is applied to an edge computing device, the edge computing device establishes a communication connection with at least one intelligent circuit breaker, and step S40, after performing classification processing on the time-series image based on a preset classification model, further includes:
step S004, judging whether the load recognition result has load abnormality.
Specifically, the edge computing device related to the embodiment is matched with at least one intelligent breaker to operate, the edge computing device performs load identification and monitoring on each electric equipment in a specific power grid, and judges whether load abnormality exists in a load identification result, wherein the load abnormality comprises short circuit, overhigh load or possible power theft and the like.
And step S005, if so, controlling the intelligent circuit breaker corresponding to the abnormal load to execute a power-off task.
Specifically, if the edge computing device determines that the load identification result has a load abnormality, the edge computing device may further send a power-off instruction to the corresponding intelligent circuit breaker. Correspondingly, after the intelligent circuit breaker receives the power-off instruction, the power-off task is executed according to the power-off instruction, so that the power-on safety is ensured.
According to the scheme, whether the load identification result is abnormal or not is judged; if yes, controlling the intelligent circuit breaker corresponding to the abnormal load to execute a power-off task. According to the embodiment, on the basis of accurately identifying the load, the edge computing equipment and the intelligent circuit breaker are controlled to realize intelligent management on the power grid, if the load is found to be abnormal, the intelligent circuit breaker can execute a power-off task, so that the power consumption safety risk is eliminated in time, and the power consumption safety is guaranteed.
Further, referring to fig. 8, a flow chart is provided in a seventh embodiment of the non-invasive load identification method of the present application, based on the embodiment shown in fig. 7, the edge computing device establishes a communication connection with the intelligent power consumption management platform, and step S004, after determining whether the load identification result has a load abnormality, further includes:
step S006, if not, uploading the load identification result to the intelligent electricity management platform so that the intelligent electricity management platform monitors electricity consumption of the electric equipment connected with the edge computing equipment.
The embodiment provides an on-line load monitoring mode, specifically, an edge computing device is connected with an intelligent electricity consumption management platform in a wired or wireless mode, and when the situation that a load identification result does not have load abnormality is judged, the edge computing device can upload the load identification result to the intelligent electricity consumption management platform. The intelligent electricity utilization management platform provides a man-machine interaction interface, and is convenient for relevant management personnel to check. Or the intelligent electricity management platform establishes communication connection with mobile terminals such as mobile phones and the like, and further sends the load identification result to the mobile terminals, so that management personnel using the mobile terminals can check the load identification result conveniently.
Similarly, when the load identification is abnormal, the edge computing device can upload the load identification result containing abnormal information to the intelligent power consumption management platform.
According to the scheme, particularly if not, the load identification result is uploaded to the intelligent electricity consumption management platform so that the intelligent electricity consumption management platform can monitor electricity consumption of electric equipment connected with the edge computing equipment. The embodiment provides an on-line load monitoring mode, wherein an edge computing device uploads a load identification result to an intelligent electricity management platform, so that remote load monitoring is realized, and the intelligent level of load monitoring is improved.
In addition, the embodiment of the application also provides a non-invasive load identification device, which comprises:
the acquisition module is used for acquiring the total load current data;
the decomposition module is used for decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data;
the drawing module is used for obtaining respective corresponding time sequence images according to the at least one single-load current data;
and the classification module is used for carrying out classification processing on the time sequence images based on a preset classification model to obtain a load identification result.
The principle and implementation process of non-invasive load identification are implemented in this embodiment, please refer to the above embodiments, and are not described herein.
In addition, the embodiment of the application also provides a terminal device, which comprises a memory, a processor and a non-invasive load identification program stored on the memory and capable of running on the processor, wherein the non-invasive load identification program realizes the steps of the non-invasive load identification method when being executed by the processor.
Because the non-invasive load identification program is executed by the processor, all the technical solutions of all the embodiments are adopted, and therefore, at least all the beneficial effects brought by all the technical solutions of all the embodiments are provided, and are not described in detail herein.
In addition, the embodiment of the application also provides a computer readable storage medium, wherein the computer readable storage medium stores a non-invasive load identification program, and the non-invasive load identification program realizes the steps of the non-invasive load identification method when being executed by a processor.
Because the non-invasive load identification program is executed by the processor, all the technical solutions of all the embodiments are adopted, and therefore, at least all the beneficial effects brought by all the technical solutions of all the embodiments are provided, and are not described in detail herein.
Compared with the prior art, the non-invasive load identification method, the non-invasive load identification device, the terminal equipment and the storage medium provided by the embodiment of the application are used for acquiring the total load current data; decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data; obtaining respective corresponding time sequence images according to the at least one single load current data; and classifying the time sequence images based on a preset classification model to obtain a load identification result. Based on the scheme, the method comprises the steps of firstly obtaining total load current data, decomposing at least one single load current data based on a decomposition model, drawing a corresponding time sequence image according to the single load current data, and taking the time sequence image as input of a classification model to obtain a load identification result. Therefore, classification based on time sequence images can effectively improve the accuracy of non-invasive load identification.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present application are merely for describing, and do not represent advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as above, including several instructions for causing a terminal device (which may be a mobile phone, a computer, a server, a controlled terminal, or a network device, etc.) to perform the method of each embodiment of the present application.
The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the claims, and all equivalent structures or equivalent processes using the descriptions and drawings of the present application, or direct or indirect application in other related technical fields are included in the scope of the claims of the present application.
Claims (10)
1. A method of non-intrusive load identification, the method comprising:
acquiring total load current data;
decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data;
obtaining respective corresponding time sequence images according to the at least one single load current data;
and classifying the time sequence images based on a preset classification model to obtain a load identification result.
2. The non-intrusive load identification method of claim 1, wherein prior to the step of obtaining the total load current data, further comprising:
based on a preset high-frequency sampling rule, acquiring single-load current data samples corresponding to a plurality of electric equipment respectively, wherein each single-load current data sample comprises current data of a plurality of periods;
selecting current data of one period from the current data of the plurality of periods as single-period current data;
model training is carried out based on the single-period current data corresponding to the electric equipment respectively to obtain an identification model, wherein the type of the identification model comprises the classification model and the decomposition model.
3. The method for non-intrusive load identification as defined in claim 2, wherein the step of performing model training based on the monocycle current data corresponding to each of the plurality of electric devices to obtain the identification model comprises:
drawing a corresponding time sequence image sample according to the single-period current data corresponding to each of the plurality of electric devices;
model training is carried out based on a preset convolutional neural network algorithm and time sequence image samples corresponding to the electric equipment respectively, and the classification model is obtained.
4. The method for non-intrusive load identification as defined in claim 2, wherein the step of performing model training based on the monocycle current data corresponding to each of the plurality of electric devices to obtain the identification model comprises:
randomly combining the single-period current data corresponding to each of the plurality of electric devices to obtain a mixed current data sample;
model training is carried out based on a preset cyclic neural network algorithm and the mixed current data sample, and the decomposition model is obtained.
5. The non-invasive load recognition method according to claim 1, wherein the classification model further comprises a dynamic load classification algorithm, and the step of classifying the time-series image based on a preset classification model to obtain a load recognition result comprises:
and classifying the time sequence images based on the dynamic load classification algorithm to obtain a load identification result corresponding to the electric equipment with the load dynamically changed.
6. The non-invasive load identification method according to claim 1, wherein the non-invasive load identification method is applied to an edge computing device, the edge computing device establishes a communication connection with at least one intelligent circuit breaker, and the step of classifying the time-series image based on a preset classification model to obtain a load identification result further comprises:
judging whether the load identification result has load abnormality or not;
if yes, controlling the intelligent circuit breaker corresponding to the abnormal load to execute a power-off task.
7. The non-intrusive load identification method of claim 6, wherein the edge computing device establishes a communication connection with an intelligent power management platform, and after the step of determining whether the load identification result has a load abnormality, further comprises:
if not, uploading the load identification result to the intelligent electricity consumption management platform so that the intelligent electricity consumption management platform can monitor electricity consumption of the electric equipment connected with the edge computing equipment.
8. A non-invasive load identification apparatus, the non-invasive load identification apparatus comprising:
the acquisition module is used for acquiring the total load current data;
the decomposition module is used for decomposing the total load current data based on a preset decomposition model to obtain at least one single load current data;
the drawing module is used for obtaining respective corresponding time sequence images according to the at least one single-load current data;
and the classification module is used for carrying out classification processing on the time sequence images based on a preset classification model to obtain a load identification result.
9. A terminal device, characterized in that it comprises a memory, a processor and a non-invasive load identification program stored on the memory and executable on the processor, which non-invasive load identification program, when executed by the processor, implements the steps of the non-invasive load identification method according to any of claims 1-7.
10. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a non-invasive load identification program, which when executed by a processor, implements the steps of the non-invasive load identification method according to any of claims 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310226288.3A CN116051910A (en) | 2023-03-10 | 2023-03-10 | Non-invasive load identification method, device, terminal equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310226288.3A CN116051910A (en) | 2023-03-10 | 2023-03-10 | Non-invasive load identification method, device, terminal equipment and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116051910A true CN116051910A (en) | 2023-05-02 |
Family
ID=86133386
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310226288.3A Pending CN116051910A (en) | 2023-03-10 | 2023-03-10 | Non-invasive load identification method, device, terminal equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116051910A (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106056879A (en) * | 2016-06-30 | 2016-10-26 | 国网山东省电力公司无棣县供电公司 | Electric meter remote control system based on intelligent identification |
CN108767847A (en) * | 2018-02-02 | 2018-11-06 | 华北电力大学扬中智能电气研究中心 | A kind of non-intrusion type power load decomposition method and device |
WO2019154050A1 (en) * | 2018-02-09 | 2019-08-15 | 国网江苏省电力有限公司电力科学研究院 | Non-intrusive load identification-based submetering smart electricity meter |
CN111985824A (en) * | 2020-08-25 | 2020-11-24 | 安徽南瑞中天电力电子有限公司 | Non-invasive load monitoring method and monitoring equipment for intelligent ammeter box |
CN114818827A (en) * | 2022-05-19 | 2022-07-29 | 国网江苏省电力有限公司营销服务中心 | Non-invasive load decomposition method based on seq2point network |
CN114859169A (en) * | 2022-04-20 | 2022-08-05 | 深圳供电局有限公司 | Intelligent identification method and system for distribution transformer outgoing line load and storage medium |
CN114977176A (en) * | 2022-07-19 | 2022-08-30 | 深圳江行联加智能科技有限公司 | Power load decomposition method, device, equipment and storage medium |
CN115758246A (en) * | 2022-11-21 | 2023-03-07 | 东南大学 | Non-invasive load identification method based on EMD and AlexNet |
-
2023
- 2023-03-10 CN CN202310226288.3A patent/CN116051910A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106056879A (en) * | 2016-06-30 | 2016-10-26 | 国网山东省电力公司无棣县供电公司 | Electric meter remote control system based on intelligent identification |
CN108767847A (en) * | 2018-02-02 | 2018-11-06 | 华北电力大学扬中智能电气研究中心 | A kind of non-intrusion type power load decomposition method and device |
WO2019154050A1 (en) * | 2018-02-09 | 2019-08-15 | 国网江苏省电力有限公司电力科学研究院 | Non-intrusive load identification-based submetering smart electricity meter |
CN111985824A (en) * | 2020-08-25 | 2020-11-24 | 安徽南瑞中天电力电子有限公司 | Non-invasive load monitoring method and monitoring equipment for intelligent ammeter box |
CN114859169A (en) * | 2022-04-20 | 2022-08-05 | 深圳供电局有限公司 | Intelligent identification method and system for distribution transformer outgoing line load and storage medium |
CN114818827A (en) * | 2022-05-19 | 2022-07-29 | 国网江苏省电力有限公司营销服务中心 | Non-invasive load decomposition method based on seq2point network |
CN114977176A (en) * | 2022-07-19 | 2022-08-30 | 深圳江行联加智能科技有限公司 | Power load decomposition method, device, equipment and storage medium |
CN115758246A (en) * | 2022-11-21 | 2023-03-07 | 东南大学 | Non-invasive load identification method based on EMD and AlexNet |
Non-Patent Citations (2)
Title |
---|
任智仁 等: "基于CNN-BiGRU-Attention的非侵入式负荷分解", 《电机与控制应用》, vol. 49, no. 8, pages 41 - 46 * |
贾子越: "基于深度学习的非侵入式负荷监测方法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》, no. 12, pages 042 - 294 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111835582B (en) | Configuration method and device of Internet of things inspection equipment and computer equipment | |
CN110430260B (en) | Robot cloud platform based on big data cloud computing support and working method | |
CN112114986B (en) | Data anomaly identification method, device, server and storage medium | |
Yang | A novel cloud information agent system with Web service techniques: Example of an energy-saving multi-agent system | |
WO2023151523A1 (en) | Deep learning programming method and system based on digital twin daas platform | |
CN111835584A (en) | Inspection method and device for products of Internet of things, computer equipment and storage medium | |
CN113222170B (en) | Intelligent algorithm and model for AI collaborative service platform of Internet of things | |
CN111314707B (en) | Data mapping identification method, device and equipment and readable storage medium | |
Xiao | Machine learning in smart home energy monitoring system | |
CN115549313A (en) | Electricity utilization monitoring method and system based on artificial intelligence | |
KR102622634B1 (en) | System for non-intrusive load monitoring of electronic device and method thereof | |
CN114189506A (en) | Equipment inspection method, device and system | |
CN107729144B (en) | Application control method and device, storage medium and electronic equipment | |
CN116051910A (en) | Non-invasive load identification method, device, terminal equipment and storage medium | |
CN117057935A (en) | Data processing method, device, equipment and storage medium based on field design | |
CN116521344A (en) | AI algorithm scheduling method and system based on resource bus | |
CN113011482B (en) | Non-invasive load identification method, terminal device and storage medium | |
CN116184851A (en) | Sleep mode adjusting method, device, terminal and storage medium | |
CN114895609A (en) | Machine room monitoring method, device, equipment and medium | |
CN113222401A (en) | Non-invasive load type intelligent identification method, device, server and system | |
CN107357703A (en) | A kind of terminal applies power consumption detection method and server | |
CN113328520A (en) | Intelligent operation and maintenance method, system and device for power transmission and storage medium | |
CN113362069A (en) | Dynamic adjustment method, device and equipment of wind control model and readable storage medium | |
CN111859370A (en) | Method, apparatus, electronic device and computer-readable storage medium for identifying service | |
CN116582468B (en) | Internet traffic monitoring method, device, equipment and storage medium |
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 |