CN113036759B - Fine granularity identification method and identification system for power consumer load - Google Patents
Fine granularity identification method and identification system for power consumer load Download PDFInfo
- Publication number
- CN113036759B CN113036759B CN202110351379.0A CN202110351379A CN113036759B CN 113036759 B CN113036759 B CN 113036759B CN 202110351379 A CN202110351379 A CN 202110351379A CN 113036759 B CN113036759 B CN 113036759B
- Authority
- CN
- China
- Prior art keywords
- load
- current
- straight line
- track
- power
- 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
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2310/00—The network for supplying or distributing electric power characterised by its spatial reach or by the load
- H02J2310/70—Load identification
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
Abstract
The invention relates to a fine granularity identification method and an identification system for loads of electric power users, wherein the identification method comprises the steps of acquiring steady-state voltage and current data of various loads in real time by using high-frequency sampling equipment; after various load steady-state voltage and current data are obtained, decomposing the high-frequency sampling current into active current and reactive current by utilizing the Fryze power theory, and carrying out standardized processing on the high-frequency sampling voltage and the reactive current to construct a two-dimensional V-I track image; after a two-dimensional V-I track image is obtained, the two-dimensional V-I track is processed through an RGB color coding technology, and active current, instantaneous power and change information thereof are respectively fused in three R, G, B channels to obtain a color V-I track image; and constructing a convolutional neural network, adjusting the resolution of the color V-I track image, inputting the color V-I track image as the neural network, and extracting the characteristics of the color V-I track image to realize the identification of the load. Compared with the prior art, the method has the advantages of high identification precision, strong universality and the like.
Description
Technical Field
The invention relates to a fine granularity identification method for electric power load, in particular to a fine granularity identification method and system for electric power user load.
Background
With the continuous increase of the electricity consumption of the family houses and commercial buildings, the reduction of the energy consumption waste of the buildings is of great significance, and the Non-invasive load monitoring (Non-Intrusive Load Monitoring, NILM) technology can realize the energy consumption monitoring and management with the lowest cost, so as to achieve the purposes of energy conservation and emission reduction. The NILM technology analyzes the internal load type, the running state and the energy consumption condition of the user in real time by acquiring the load data of the power user bus, helps the user to know own power consumption habit, guides the user to optimize the power consumption mode and reduces the household power consumption; meanwhile, the method is beneficial to the fine management of the power grid enhancement demand side, the dynamic demand response potential of the power consumer is mined, and the bidirectional interactive intelligent electricity service is provided.
At present, a single V-I track can only transmit image information, so that information such as power, harmonic waves and the like of equipment cannot be reflected in principle, and the phenomenon of overlapping of the characteristics of the V-I track exists among different loads with similar working principles due to various electric appliances. Although feature fusion can solve the problems of single feature and feature overlapping, steady-state features are used for fusion in the prior art, the attention to transient features is insufficient, and the transient features are more suitable for load identification tasks due to the uniqueness of the transient features, so that the identification precision of the power user load identification method in the prior art is lower, and the algorithm processing speed is also lower.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a fine granularity identification method and system for the load of the power consumer, which have high identification precision and strong universality.
The aim of the invention can be achieved by the following technical scheme:
a method for fine granularity identification of power consumer load, said identification method comprising:
step 1: acquiring steady-state voltage and current data of various loads in real time by using high-frequency sampling equipment;
step 2: after various load steady-state voltage and current data are obtained, decomposing the high-frequency sampling current into active current and reactive current by utilizing the Fryze power theory, and carrying out standardized processing on the high-frequency sampling voltage and the reactive current to construct a two-dimensional V-I track image;
step 3: after a two-dimensional V-I track image is obtained, the two-dimensional V-I track is processed through an RGB color coding technology, and active current, instantaneous power and change information thereof are respectively fused in three R, G, B channels to obtain a color V-I track image;
step 4: and constructing a convolutional neural network, adjusting the resolution of the color V-I track image, inputting the color V-I track image as the neural network, and extracting the characteristics of the color V-I track image to realize the identification of the load.
Preferably, the sampling frequency of the high-frequency sampling device in the step 1 to the load voltage and the load current is greater than or equal to 20KHz.
Preferably, when the high-frequency sampling device in step 1 collects the steady-state voltage of the load, the voltage waveform is kept substantially unchanged before and after the switching event of the single electric appliance, and the current waveform is obtained by calculating the steady-state current difference value of the buses before and after the switching event.
More preferably, the method for judging the switching event of the single electrical appliance is as follows:
and calculating the active power difference value of adjacent periods of the load at a certain moment, and detecting that a switching event occurs if the difference value is larger than a set threshold value.
Preferably, in the step 2, the method for performing current decomposition by adopting the Fryze power theory specifically includes:
i f (t)＝i(t)-i a (t)
wherein P is a Active power in a steady-state period; v rms Is the effective value of the load voltage; t (T) s Is a steady state period; v (t) is load steady-state high-frequency sampling voltage data at the moment t; i (t) is at time tLoad raw total current data; i.e a (t) is active current data obtained after current decomposition at time t; i.e f And (t) is reactive current data obtained after decomposition at time t.
Preferably, the method for constructing the two-dimensional V-I track by using the standardized voltage and the reactive current in the step 2 specifically includes:
wherein v (k) and i f (k) The method comprises the steps of acquiring original voltage and reactive current data; Δi k Is the reactive current data after standardization; deltav k Is the voltage data after normalization processing.
Preferably, in the step 3, the two-dimensional V-I track is processed by a color coding technology, and active current, instantaneous power and change information thereof are respectively fused into three R, G, B channels, which comprises the following specific steps:
step 3-1: in the two-dimensional V-I track, connecting two adjacent sampling points by using straight line segments, and directly performing color coding on the straight line segments;
step 3-2: for each straight line segment, fusing active current amplitude information on an R channel, coloring the straight line segment between two adjacent sampling points by taking the active current average value of two adjacent sampling points as the straight line segment, wherein the range of the depth of the R channel is between (0 and 1), so that the average active current is required to be subjected to standardized treatment;
the standardized processing method comprises the following steps:
wherein j is the sampling point sequence number; r is R j The depth value of the R channel is the j-th straight line segment; i.e a Is active current data;
step 3-3: for each straight line segment, fusing V-I track motion information on a G channel, introducing the slope of the straight line segment between two adjacent sampling points of the V-I track into the G channel, and mapping the slope of the straight line segment into a (0, 1) interval by using an arctangent function, wherein the specific method comprises the following steps:
wherein K is j Is the slope of the jth straight line segment; g j The depth value of the G channel of the jth straight line segment; v is high frequency steady state voltage data; i.e f Is reactive current data;
step 3-4: and fusing instantaneous power amplitude information on the B channel for each straight line segment, carrying out standardization processing on the instantaneous power average value of two adjacent sampling points, taking the instantaneous power average value as a depth value of the B channel, and judging the power grade of each device through the change range of the depth value of the B channel, wherein the formula is as follows:
M＝max{p 1 ,p 2 ,…,p m }
wherein B is j The depth value of the channel B is the j-th straight line segment; m is the maximum value of instantaneous power in sampling points of all devices; m is the total number of the devices; p is p 1 ,p 2 ,…,p m The instantaneous power maximum for a single device sampling point.
Preferably, the method for constructing the convolutional neural network in the step 4 specifically includes:
step 4-1: constructing an original VGG16 convolutional neural network, which comprises 13 convolutional layers, 5 pooling layers and three full-connection layers; the convolution kernel size of each convolution layer is set to be 3 multiplied by 3, each pooling layer adopts a maximum pooling mode, and the pooling kernel size is 2 multiplied by 2;
step 4-2: modifying the number of neurons of the last full-connection layer from 1000 to the actual device class number;
step 4-3: using a global averaging pooling layer to replace the first fully connected layer and adjusting the number of neurons of the second fully connected layer to 512;
step 4-4: a Dropout layer was added after the second fully connected layer, the parameter set to 0.5.
Preferably, the constraint of adjusting the resolution of the color V-I track image in the step 4 is:
the image resolution is set so that the image features are clearly expressed while avoiding noise and interference.
The power load fine granularity recognition system comprises a multifunctional acquisition card, a singlechip, a wireless communication module and a cloud server; the multifunctional acquisition card and the wireless communication module are respectively connected with the singlechip; the cloud server stores the fine granularity identification method for the power user load according to any one of the above; the wireless communication module transmits the load data acquired by the multifunctional acquisition card to the cloud server, and the data processing and load identification tasks are completed in the cloud server.
Compared with the prior art, the invention has the following beneficial effects:
1. the identification precision is high: after the electric power user load fine granularity identification method adopts the Fryze power theory to carry out current decomposition, the V-I track is constructed by voltage and reactive current, and compared with the traditional V-I track, the shape difference of the V-I track among different types of equipment can be increased, so that the uniqueness of the V-I track is improved; the multi-feature fusion is realized on the R, G, B three channels in a color coding mode, so that the information quantity carried by the tracks is increased, the similarity of the tracks of different devices is further reduced, and the problem that different types of devices are difficult to identify due to similar load features can be solved; meanwhile, the convolutional neural network adopts a VGG16 model with a more complex structure, so that the characteristic extraction capability of the model is improved, the accuracy of load identification is improved, and the accuracy can reach 97.39%.
2. The universality is strong: according to the power user load fine granularity identification system, the embedded equipment is used for acquiring load data, the cloud server is used for processing the data and utilizing the non-invasive load identification device for carrying out load identification by deep learning, so that the load identification is not limited to the embedded equipment with lower computing capability, and the universality of the identification method is greatly improved.
Drawings
FIG. 1 is a flow chart of a fine granularity identification method for electric power consumer load in the invention;
FIG. 2 is a schematic diagram of a color coding model process flow according to the present invention;
FIG. 3 is a schematic diagram of a convolutional neural network according to the present invention;
FIG. 4 is a graph of a theoretical current split for a partial load based on Fryze power in an embodiment of the invention;
FIG. 4 (a) is a graph of theoretical current decomposition of air conditioning load based on Fryze power;
fig. 4 (b) is a graph of theoretical current decomposition of the electric fan load based on Fryze power;
FIG. 4 (c) is a graph of the theoretical current split of a notebook computer load based on Fryze power;
FIG. 4 (d) is a graph of theoretical current split for a washer load based on Fryze power;
FIG. 5 is a partial loaded color V-I trace gray scale plot in an embodiment of the invention;
FIG. 5 (a) is a color V-I trace gray scale plot of air conditioning load;
FIG. 5 (b) is a gray scale plot of a color V-I trace for an electric fan load;
FIG. 5 (c) is a gray scale plot of a color V-I trace for a notebook computer load;
FIG. 5 (d) is a gray scale plot of a color V-I trace for a washing machine load.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
A fine granularity identification method and an identification system for power consumer loads are shown in figure 1, and the method comprises the following steps:
step 1: acquiring steady-state voltage and current data of various loads in real time by using high-frequency sampling equipment;
in the embodiment, the high-frequency sampling equipment consisting of the ATT7053 multifunctional acquisition card and the STM32F103 singlechip is used for acquiring steady-state voltage and current data of the home subscriber bus, and the wireless communication module (WIFI module) is used for transmitting the data to the cloud server, so that the data processing and load identification tasks are completed in the cloud server.
In the step 1, the sampling frequency of the high-frequency sampling equipment to the load voltage and current is more than or equal to 20KHz.
When the high-frequency sampling equipment in the step 1 collects the steady-state voltage of the load, the voltage waveform is basically kept unchanged before and after a switching event of a single electric appliance, and the current waveform is obtained by calculating the steady-state current difference value of buses before and after the switching event.
The method for judging the load switch event comprises the following steps: (1) Calculating the active power difference value of adjacent periods of the load at a certain moment; (2) If the difference is greater than a set threshold, a switching event is detected. If the load switch event occurs, calculating the steady-state current difference value of the buses before and after the switching switch to obtain the steady-state current of the equipment.
Step 2: and decomposing the high-frequency sampling current into active current and reactive current by utilizing the Fryze power theory, and carrying out standardization processing on the high-frequency sampling voltage and the reactive current to construct a two-dimensional V-I track image. The calculation steps for current decomposition using the Fryze power theory are as follows:
i f (t)＝i(t)-i a (t)
wherein P is a Active power in a steady-state period; v rms Is the effective value of the load voltage; t (T) s Is a steady state period; v (t) is load steady-state high-frequency sampling voltage data at the moment t; i (t) is the original total current data of the load at the moment t; i.e a (t) is active current data obtained after current decomposition at time t; i.e f And (t) is reactive current data obtained after decomposition at time t.
The formula for constructing the two-dimensional V-I track by using the standardized voltage and the reactive current is as follows:
wherein v (k) and i f (k) The method comprises the steps of acquiring original voltage and reactive current data; Δi k Is the reactive current data after standardization; deltav k Is the voltage data after normalization processing.
Step 3: after a two-dimensional V-I track image is obtained, the two-dimensional V-I track is processed through an RGB color coding technology, and active current, instantaneous power and change information thereof are respectively fused in three R, G, B channels to obtain a color V-I track image;
the color coding process is shown in fig. 3 and includes:
step 3-1: in the two-dimensional V-I track, connecting two adjacent sampling points by using straight line segments, and directly performing color coding on the straight line segments;
step 3-2: for each straight line segment, fusing active current amplitude information on an R channel, coloring the straight line segment between two adjacent sampling points by taking the active current average value of two adjacent sampling points as the straight line segment, wherein the range of the depth of the R channel is between (0 and 1), so that the average active current is required to be subjected to standardized treatment;
the standardized processing method comprises the following steps:
wherein j is the sampling point sequence number; r is R j The depth value of the R channel is the j-th straight line segment; i.e a Is active current data;
step 3-3: for each straight line segment, fusing V-I track motion information on a G channel, introducing the slope of the straight line segment between two adjacent sampling points of the V-I track into the G channel, and mapping the slope of the straight line segment into a (0, 1) interval by using an arctangent function, wherein the specific method comprises the following steps:
wherein K is j Is the slope of the jth straight line segment; g j The depth value of the G channel of the jth straight line segment; v is high frequency steady state voltage data; i.e f Is reactive current data;
step 3-4: and fusing instantaneous power amplitude information on the B channel for each straight line segment, carrying out standardization processing on the instantaneous power average value of two adjacent sampling points, taking the instantaneous power average value as a depth value of the B channel, and judging the power grade of each device through the change range of the depth value of the B channel, wherein the formula is as follows:
M＝max{p 1 ,p 2 ,…,p m }
wherein B is j The depth value of the channel B is the j-th straight line segment; m is the maximum value of instantaneous power in sampling points of all devices; m is the total number of the devices; p is p 1 ,p 2 ,…,p m The instantaneous power maximum for a single device sampling point.
As shown in fig. 2, in the present embodiment, the operation in step 3 is adopted for each device to construct a device color V-I track image database, and a color V-I track gray scale map of a partial load in an example is shown in fig. 4, in which fig. 4 (a) is a theoretical current decomposition graph of an air conditioner load based on fryz power, fig. 4 (b) is a theoretical current decomposition graph of an electric fan load based on fryz power, fig. 4 (c) is a theoretical current decomposition graph of a notebook computer load based on fryz power, and fig. 4 (d) is a theoretical current decomposition graph of a washing machine load based on fryz power.
Step 4: and constructing and improving a VGG16 convolutional neural network, adjusting the resolution of the color V-I track image, inputting the color V-I track image as a neural network, and extracting the characteristics of the color V-I track image to realize the identification of the load.
The method for constructing the convolutional neural network specifically comprises the following steps:
step 4-1: constructing an original VGG16 convolutional neural network, which comprises 13 convolutional layers, 5 pooling layers and three full-connection layers; the convolution kernel size of each convolution layer is set to be 3 multiplied by 3, each pooling layer adopts a maximum pooling mode, and the pooling kernel size is 2 multiplied by 2;
step 4-2: in order to adapt the convolutional neural network to the classification number of the actual load identification task, modifying the number of neurons of the last full-connection layer from 1000 to the actual equipment class number;
step 4-3: to reduce model parameters, save computational resources, replace the first fully connected layer with a global averaging pooling layer and adjust the number of neurons of the second fully connected layer to 512;
step 4-4: to suppress the model overfitting, a Dropout layer was added after the second fully connected layer, with the parameter set to 0.5.
And finally, taking the image as the input of the improved VGG16 convolutional neural network, classifying the image, and completing the fine granularity identification of the equipment.
The embodiment also relates to a power load fine granularity identification system, which comprises a multifunctional acquisition card, a singlechip, a wireless communication module and a cloud server. The multifunctional acquisition card and the wireless communication module are respectively connected with the singlechip, any power user load fine-granularity identification method is stored in the cloud server, the wireless communication module transmits load data acquired by the multifunctional acquisition card to the cloud server, and data processing and load identification tasks are completed in the cloud server.
The present example verifies the proposed method against the public dataset PLAID, which contains a total of 1074 sets of data of current and voltage measurements sampled at 30kHz by 11 different types of consumers. The image resolution was set to 64×64, the VGG16 convolutional neural network was migration learned with training weights of the ImageNet dataset, and the softmax multi-classifier number was set to 11. In the training process, if the loss value does not drop in 5 epochs, the learning rate is adjusted to 1/2 of the original learning rate so as to optimize the model. The overall recognition accuracy obtained by the test result is 97.39%, and compared with the existing advanced algorithm, the recognition rate is greatly improved. The non-invasive load identification device which collects load data by the embedded equipment, processes the data by the cloud server and performs load identification by deep learning is realized, so that the load identification is not limited to the embedded equipment with lower computing capacity.
While the invention has been described with reference to certain preferred embodiments, it will be understood by those skilled in the art that various changes and substitutions of equivalents may be made and equivalents will be apparent to those skilled in the art without departing from the scope of the invention. Therefore, the protection scope of the invention is subject to the protection scope of the claims.
Claims (7)
1. The fine granularity identification method for the power consumer load is characterized by comprising the following steps of:
step 1: acquiring steady-state voltage and current data of various loads in real time by using high-frequency sampling equipment;
step 2: after various load steady-state voltage and current data are obtained, decomposing the high-frequency sampling current into active current and reactive current by utilizing the Fryze power theory, and carrying out standardized processing on the high-frequency sampling voltage and the reactive current to construct a two-dimensional V-I track image;
step 3: after a two-dimensional V-I track image is obtained, the two-dimensional V-I track is processed through an RGB color coding technology, and active current, instantaneous power and change information thereof are respectively fused in three R, G, B channels to obtain a color V-I track image;
step 4: constructing a convolutional neural network, adjusting the resolution of the color V-I track image, inputting the color V-I track image as the neural network, and extracting the characteristics of the color V-I track image to realize the identification of the load;
the method for carrying out current decomposition by adopting the Fryze power theory in the step 2 specifically comprises the following steps:
i f (t)＝i(t)-i a (t)
wherein P is a Active power in a steady-state period; v rms Is the effective value of the load voltage; t (T) s Is a steady state period; v (t) is load steady-state high-frequency sampling voltage data at the moment t; i (t) is the original total current data of the load at the moment t; i.e a (t) is active current data obtained after current decomposition at time t; i.e f (t) is reactive current data obtained after decomposition at time t;
the method for constructing the two-dimensional V-I track by using the standardized voltage and the reactive current in the step 2 specifically comprises the following steps:
wherein v (k) and i f (k) The method comprises the steps of acquiring original voltage and reactive current data; Δi k Is the reactive current data after standardization; deltav k Is voltage data after standardized treatment;
in the step 3, the two-dimensional V-I track is processed by a color coding technology, and the active current, the instantaneous power and the change information thereof are respectively integrated into three R, G, B channels, and the specific method is as follows:
step 3-1: in the two-dimensional V-I track, connecting two adjacent sampling points by using straight line segments, and directly performing color coding on the straight line segments;
step 3-2: for each straight line segment, fusing active current amplitude information on an R channel, coloring the straight line segment between two adjacent sampling points by taking the active current average value of two adjacent sampling points as the straight line segment, wherein the range of the depth of the R channel is between (0 and 1), so that the average active current is required to be subjected to standardized treatment;
the standardized processing method comprises the following steps:
wherein j is the sampling point sequence number; r is R j The depth value of the R channel is the j-th straight line segment; i.e a Is active current data;
step 3-3: for each straight line segment, fusing V-I track motion information on a G channel, introducing the slope of the straight line segment between two adjacent sampling points of the V-I track into the G channel, and mapping the slope of the straight line segment into a (0, 1) interval by using an arctangent function, wherein the specific method comprises the following steps:
wherein K is j Is the slope of the jth straight line segment; g j The depth value of the G channel of the jth straight line segment; v is high frequency steady state voltage data; i.e f Is reactive current data;
step 3-4: and fusing instantaneous power amplitude information on the B channel for each straight line segment, carrying out standardization processing on the instantaneous power average value of two adjacent sampling points, taking the instantaneous power average value as a depth value of the B channel, and judging the power grade of each device through the change range of the depth value of the B channel, wherein the formula is as follows:
M＝max{p 1 ,p 2 ,…,p m }
wherein B is j The depth value of the channel B is the j-th straight line segment; m is the maximum value of instantaneous power in sampling points of all devices; m is the total number of the devices; p is p 1 ,p 2 ,…,p m The instantaneous power maximum for a single device sampling point.
2. The method for identifying the fine granularity of the load of the electric power consumer according to claim 1, wherein the sampling frequency of the high-frequency sampling device to the load voltage and the load current in the step 1 is more than or equal to 20KHz.
3. The method for identifying fine granularity of load of electric power consumer according to claim 1, wherein when the high-frequency sampling device in step 1 collects the steady-state voltage of the load, the voltage waveform is kept substantially unchanged before and after the switching event of the single electric appliance, and the current waveform is obtained by calculating the steady-state current difference value of the bus before and after the switching event.
4. The method for identifying fine granularity of power consumer load according to claim 3, wherein the method for judging the single electrical switching event is as follows:
and calculating the active power difference value of adjacent periods of the load at a certain moment, and detecting that a switching event occurs if the difference value is larger than a set threshold value.
5. The method for identifying fine granularity of power consumer load according to claim 1, wherein the method for constructing the convolutional neural network in the step 4 specifically comprises the following steps:
step 4-1: constructing an original VGG16 convolutional neural network, which comprises 13 convolutional layers, 5 pooling layers and three full-connection layers; the convolution kernel size of each convolution layer is set to be 3 multiplied by 3, each pooling layer adopts a maximum pooling mode, and the pooling kernel size is 2 multiplied by 2;
step 4-2: modifying the number of neurons of the last full-connection layer from 1000 to the actual device class number;
step 4-3: using a global averaging pooling layer to replace the first fully connected layer and adjusting the number of neurons of the second fully connected layer to 512;
step 4-4: a Dropout layer was added after the second fully connected layer, the parameter set to 0.5.
6. The method for identifying fine granularity of power consumer load according to claim 1, wherein the constraint condition for adjusting the resolution of the color V-I track image in step 4 is:
the image resolution is set so that the image features are clearly expressed while avoiding noise and interference.
7. The power load fine granularity recognition system is characterized by comprising a multifunctional acquisition card, a singlechip, a wireless communication module and a cloud server; the multifunctional acquisition card and the wireless communication module are respectively connected with the singlechip; the cloud server stores the fine granularity identification method for the power consumer load according to any one of claims 1 to 6; the wireless communication module transmits the load data acquired by the multifunctional acquisition card to the cloud server, and the data processing and load identification tasks are completed in the cloud server.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110351379.0A CN113036759B (en) | 2021-03-31 | 2021-03-31 | Fine granularity identification method and identification system for power consumer load |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110351379.0A CN113036759B (en) | 2021-03-31 | 2021-03-31 | Fine granularity identification method and identification system for power consumer load |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113036759A CN113036759A (en) | 2021-06-25 |
| CN113036759B true CN113036759B (en) | 2023-09-26 |
Family
ID=76453181
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110351379.0A Active CN113036759B (en) | 2021-03-31 | 2021-03-31 | Fine granularity identification method and identification system for power consumer load |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113036759B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113537327A (en) * | 2021-07-06 | 2021-10-22 | 江阴长仪集团有限公司 | Non-invasive load identification method and system based on Alexnet neural network and color coding |
| CN113723479A (en) * | 2021-08-18 | 2021-11-30 | 南京工程学院 | Non-invasive load identification method based on GRNN and mean shift algorithm |
| CN114372979A (en) * | 2022-02-23 | 2022-04-19 | 吉林化工学院 | Transferable electric power fingerprint depth identification method |
| CN114330461B (en) * | 2022-03-16 | 2022-06-24 | 北京智芯微电子科技有限公司 | V-I track generation method and device for non-invasive load identification and neural network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110188771A (en) * | 2019-05-17 | 2019-08-30 | 湖南大学 | Non-intrusive electrical load feature extraction, recognition methods, system and medium based on image procossing |
| CN111864896A (en) * | 2019-04-29 | 2020-10-30 | 清华大学 | Power load monitoring method and system |
| CN112180193A (en) * | 2020-09-28 | 2021-01-05 | 华中科技大学 | Non-invasive load identification system and method based on track image identification |
| CN112418722A (en) * | 2020-12-08 | 2021-02-26 | 浙江大学 | Non-invasive load identification method based on V-I (velocity-amplitude) trajectory graph and neural network |
-
2021
- 2021-03-31 CN CN202110351379.0A patent/CN113036759B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111864896A (en) * | 2019-04-29 | 2020-10-30 | 清华大学 | Power load monitoring method and system |
| CN110188771A (en) * | 2019-05-17 | 2019-08-30 | 湖南大学 | Non-intrusive electrical load feature extraction, recognition methods, system and medium based on image procossing |
| CN112180193A (en) * | 2020-09-28 | 2021-01-05 | 华中科技大学 | Non-invasive load identification system and method based on track image identification |
| CN112418722A (en) * | 2020-12-08 | 2021-02-26 | 浙江大学 | Non-invasive load identification method based on V-I (velocity-amplitude) trajectory graph and neural network |
Non-Patent Citations (1)
| Title |
|---|
| DAWIT FEKADU TESHOME等.Distinctive Load Feature Extraction Based on Fryze's Time-Domain Power Theory.《IEEE Power and Energy Technology Systems Journal》.2016,第03卷(第02期),60-70. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113036759A (en) | 2021-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113036759B (en) | Fine granularity identification method and identification system for power consumer load | |
| CN106022645B (en) | Non-invasive online real-time power load identification method and identification system | |
| CN112180193B (en) | Non-invasive load identification system and method based on track image identification | |
| CN112434799B (en) | Non-invasive load identification method based on full convolution neural network | |
| CN111864896B (en) | Power load monitoring method and system | |
| CN113011481B (en) | Electric energy meter function abnormality assessment method and system based on decision tree algorithm | |
| CN109470957B (en) | Non-invasive electrical appliance identification method based on Frey's distance | |
| CN108197817A (en) | A kind of method of the non-intrusion type load transient state monitoring based on big data | |
| Quek et al. | Load disaggregation using one-directional convolutional stacked long short-term memory recurrent neural network | |
| CN114970633B (en) | LSTM-based non-invasive electrical appliance identification method, system and equipment | |
| CN113902104A (en) | Non-invasive load monitoring method combining unsupervised domain self-adaptive strategy and attention mechanism | |
| Du et al. | Support vector machine based methods for non-intrusive identification of miscellaneous electric loads | |
| CN111563827A (en) | Load decomposition method based on electrical appliance physical characteristics and residential electricity consumption behaviors | |
| CN115456034A (en) | Automatic identification and monitoring method and system for electric bicycle charging | |
| Gurbuz et al. | Comprehensive non-intrusive load monitoring process: Device event detection, device feature extraction and device identification using KNN, random forest and decision tree | |
| Gurbuz et al. | A brief review of non-intrusive load monitoring and its impact on social life | |
| Dan et al. | Review of non-intrusive load appliance monitoring | |
| CN110244150B (en) | Non-invasive electrical appliance identification method based on root mean square and standard deviation | |
| Zhou et al. | Research on non-invasive load monitoring based on convolutional neural network | |
| CN113537330B (en) | Non-invasive load identification method | |
| CN111539275B (en) | Electrical load classification method and system based on load characteristic visualization | |
| CN114662576A (en) | Non-invasive storage battery car charging detection method and system based on supervised classification | |
| Tito et al. | Image segmentation-based event detection for non-intrusive load monitoring using Gramian Angular Summation Field | |
| CN110146758B (en) | Non-invasive electrical appliance identification method based on cross entropy | |
| Li et al. | A data augmentation method for distributed photovoltaic electricity theft using wasserstein generative adversarial network |
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 |