CN112968519B - Intelligent power load identification method - Google Patents
Intelligent power load identification method Download PDFInfo
- Publication number
- CN112968519B CN112968519B CN202110053866.9A CN202110053866A CN112968519B CN 112968519 B CN112968519 B CN 112968519B CN 202110053866 A CN202110053866 A CN 202110053866A CN 112968519 B CN112968519 B CN 112968519B
- Authority
- CN
- China
- Prior art keywords
- data
- graph node
- load
- circuit
- frequency domain
- 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
Images
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
- H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
-
- 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
-
- 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
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/40—Display of information, e.g. of data or controls
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses an intelligent power load identification method, which realizes lossless low-cost load identification without additionally arranging a sensor before each load. The method realizes the change of the total load of the circuit by increasing or decreasing the electric appliances in the circuit, and records the electric parameter variable quantity x of the total incoming line of the circuit when the load changes(i)While simultaneously converting x(i)The corresponding current circuit load condition is recorded as graph node data g(i). Graph node data g(i)Carrying out fast Fourier transform and converting the fast Fourier transform into graph node frequency domain data h with fixed dimension(i),x(i)And h(i)And forming a training sample. Constructing a self-circulation neural network, wherein the input of the self-circulation neural network is the electrical parameter variable quantity x at the moment i(i)And graph node frequency domain data h at the time of i-1(i‑1)And outputting graph node frequency domain data h at the moment i(i)(ii) a The training sample is adopted to carry out circular training on the self-circulation neural network, and the trained self-circulation neural network can be used for intelligent power load identification.
Description
Technical Field
The invention relates to the field of power load identification, in particular to an intelligent power load identification method based on deep learning.
Background
The restriction of resource environment is a prominent contradiction faced by the development of the economic society of China, how to promote energy conservation and emission reduction is a core key for solving the problems of energy conservation and environmental protection, and has very important significance. The premise of reasonably optimizing the energy structure for energy conservation and emission reduction lies in identifying load equipment in the circuit, and the traditional identification method for the load in the circuit needs to additionally install a sensor before each load, so that the cost is high, and the original circuit needs to be modified.
Disclosure of Invention
In view of this, the invention provides an intelligent power load identification method, which can record and identify the change of the current and the voltage of the main incoming line by using a deep learning method without additionally arranging a sensor in front of each load, thereby realizing lossless low-cost load identification.
In order to solve the above-mentioned technical problems, the present invention has been accomplished as described above.
An intelligent electrical load identification method, comprising:
Wherein, at the ith moment when the electrical parameter changes, the change data of the electrical parameter is added with the time and environment additional data at that moment and is recorded as the electrical parameter change amount x at the ith moment(i)(ii) a Graph node data g(i)By n sub-nodesN is the variable total load number in the circuit; each child nodeCorresponding to a single electric load in the circuit, k is equal to [0, n-1 ]],Records the load typeWith the current load powerGraph node data g(i)From the graph node data g of the previous moment(i-1)Increasing or decreasing child nodes to obtain;
step 3, utilizing a hidden Markov model to carry out electric parameter variation x(h)And graph node frequency domain data h(h)Modeling is carried out, the graph node frequency domain data are regarded as hidden variables of the variation of the electrical parameter, and the graph node frequency domain data h(i)Graph node frequency domain data h only subjected to last moment(i-1)The influence of (a);
step 4, realizing a hidden Markov model by using a self-circulation neural network; the input of the self-circulation neural network is the electrical parameter variable quantity x at the moment i(i)And graph node frequency domain data h at the time of i-1(i-1)And outputting graph node frequency domain data h at the moment i(i)(ii) a Performing cyclic training on the self-circulation neural network by adopting the training sample obtained in the step 2;
step 5, during actual load identification, constructing initial graph node data g according to the initial load state in the circuit(0)And converted into initial graph node frequency domain data h(0)(ii) a When the load change occurs in the circuit, i is made to be 1, and the electrical parameter change x in the circuit is obtained(1)X is to be(1)And h(0)Input into a self-circulation neural network, output h(1)To h is aligned with(1)Obtaining graph node data g by inverse Fourier transform(1)(ii) a When the load in the circuit changes, i is added with 1, and the cyclic calculation of the self-circulation neural network is performed, wherein the self-circulation neural network utilizes x(i)And h(i-1)Calculate the output h(i)To h is aligned with(i)Carrying out inverse Fourier transform to obtain graph node data g(i)According to g(i)A current load condition in the circuit is identified.
Preferably, in the step 2, the graph node data g is processed(i)Fourier transform is carried out, and the Fourier transform is converted into graph node frequency domain data h with fixed dimension(i)Comprises the following steps:
let g(i)Wherein n current contains n child nodes, each child node is composed of 2 data, then g(i)Can be expressed as an n x 2 dimensional numberGroup, using fast Fourier transform algorithm to transform g(i)The n x 2 dimensional array is converted into m x 2 dimensional frequency domain data, wherein m is the preset frequency domain sampling number, and the value is ensured to be always larger than g in actual use(i)The number of child nodes n.
Preferably, in the step 4, h is measured(i)Carrying out inverse Fourier transform to obtain graph node data g(i)Comprises the following steps:
Preferably, m is set to 100 in a home scenario and 1000 in an industrial power environment.
Preferably, only one appliance load is changed at step 1 on a single change of the total load of the circuit.
Preferably, the load change in the circuit is identified by: the intelligent electric meter is installed at the general incoming line of the circuit, so that the collection of electric parameters in the circuit is realized, and the load change is identified according to the sudden change of the electric parameters.
Preferably, the electrical parameter variation x(i)The variation data of said electrical parameter in (a) includes total active power, reactive power, apparent power, voltage, current, voltage, and voltage,The true effective value of the voltage current, the phase angle and the amount of change in frequency.
Preferably, for x(1)Wherein the variation data of the electrical parameter is the first recorded electrical parameter data without the variation.
Preferably, the electrical parameter variation x(i)The environment additional data in (1) includes a current time, an air temperature, and an air humidity.
Has the advantages that:
(1) the method obtains the electrical parameter variation from the circuit general incoming line, obtains the current circuit load condition, and trains the self-circulation neural network by forming a training sample by the data, so that the self-circulation neural network can reflect the mapping relation between the electrical parameter variation and the circuit load condition. By the deep learning method, the intelligent identification of the load can be realized only by acquiring the total electric parameter data of the circuit without additionally arranging a sensor in front of each load.
(2) According to the invention, the identification of the load in the circuit is assisted by using environmental state data such as time, temperature, humidity and the like, so that the identification precision is effectively improved;
(3) the invention records the load change caused by the change of the electrical parameter in the circuit by using the encoding mode of the graph and the sub-nodes, and can quickly establish a training data set.
(4) The invention utilizes a Fourier transform method to map the unstructured sub-nodes in the graph to the frequency domain space, realizes the structuralization of data and is convenient for the training and the use of a subsequent self-circulation data network.
Drawings
FIG. 1 is a schematic diagram of the composition of graph nodes according to the present invention.
Figure 2 is a schematic diagram of the hidden markov model of the present invention.
Detailed Description
The invention is described in detail below by way of example with reference to the accompanying drawings.
The invention provides an intelligent power load identification method, which is characterized in that the method obtains the variable quantity of an electric parameter from a circuit general inlet, obtains the current circuit load condition, and forms a training sample by the data to train a self-circulation neural network, so that the self-circulation neural network can reflect the mapping relation between the variable quantity of the electric parameter and the circuit load condition.
The invention provides an intelligent power load identification method, which comprises the following steps:
The electrical parameters selected by the invention comprise active power, reactive power, apparent power, voltage and current true effective values, phase angles, frequencies and other electrical parameters.
The invention adds time and environmental data to the electric parameter data, and takes the problem of electricity utilization habits into consideration. The washing machine may be turned on at a certain time or the microwave oven may be used at a certain time, and the air conditioner may be used at a high temperature. Thus, the addition of environmental and temporal data can make the learning of the neural network more accurate. Preferably, the environmental additional data is selected from the current time (in seconds), air temperature, air humidity.
Step 3, changing the electrical parameter variable x(i)The corresponding current circuit load condition is marked as graph node g(i)。
As shown in FIG. 1, the graph node includes n sub-nodesn is the variable total load number in the circuit at this time; each one of which isCorresponding to a single electric load, k is equal to 0, n-1]。The load type of the load is recordedAnd the load power at that time
Graph node data g(i)Is obtained by comparing the graph node data g of the previous moment(i-1)Adding and deleting child nodes. Graph node g(i)Only with the graph node g at the previous moment(i-1)It is relevant. For example, increasing the use of an electric water heater, shutting down a washing machine, or changing the power of a refrigerator, may cause the graph nodes to change.
Step 4, simulating daily electricity utilization conditions, increasing and decreasing electric appliances in the circuit to realize the change of the total load of the circuit, and recording the electrical parameter variable quantity x of the total incoming line of the circuit when the load changes(i)While simultaneously converting x(i)The corresponding current circuit load condition is recorded as graph node data g(i). To facilitate subsequent neural network training, only one appliance load is changed at a single change in the total load of the circuit.
Wherein x is(i)Corresponding graph node data g(i)From g(i-1)The middle increase and decrease child nodes are obtained, and the load types of the corresponding increase and decrease child nodes are recordedAnd load powerThrough this step, the original data set is generated.
Step 5, because the number of the child nodes is not fixed, the graph node data g is caused(i)Is not fixed in dimensionThus, for graph node data g(i)Fourier transform is carried out, and the Fourier transform is converted into graph node frequency domain data h with fixed dimension(i)(ii) a For subsequent algorithmic model calculations. The electrical parameter variation x(i)And graph node frequency domain data h(i)And forming a training sample.
Because the original data centralizes the graph node g(i)The number of the sub-nodes included in (1) is not fixed, and therefore, the data of the sub-nodes is converted into the frequency domain data h with fixed length by using fourier transform(i)Then the training is carried out by sending the training data to a neural network model.
The specific method comprises the following steps: suppose g(i)The method comprises n sub-nodes, wherein each sub-node comprises two data of load type and load power, namely g(i)Can be represented as oneN is a different value at different times. G is divided by fast Fourier transform algorithm fft(i)Is/are as followsOf fixed dimensionThe frequency domain data of (1), wherein m is the number of artificially set frequency domain samples and is a fixed value, and the value of m is ensured to be always larger than g in actual use(i)The number of child nodes n. Preferably, the setting may be 100 in a scenario with a small number of power consumption loads, such as a home scenario, and 1000 in a scenario with a large number of power consumption loads, such as an industrial power consumption environment.
Step 6, utilizing a hidden Markov model to carry out electric parameter variation x(i)And graph node frequency domain data h(i)Modeling is carried out, the graph node frequency domain data are regarded as hidden variables of the variation of the electrical parameter, and the graph node frequency domain data h(i)Graph node frequency domain data h only subjected to last moment(i-1)The influence of (c). As shown in fig. 2.
And 7, realizing a hidden Markov model by using the self-circulation neural network, and training the self-circulation neural network.
The input of the self-circulation neural network is the electrical parameter variable quantity x at the moment i(i)And graph node frequency domain data h at the time of i-1(i-1)And outputting graph node frequency domain data h at the moment i(i)(ii) a And (4) performing cyclic training on the self-circulation neural network by adopting the training samples obtained in the step (3). In training, because x(i)And h(i)Is i gradually increasing continuous data, firstly inputting x(1)And h(0)Output h of neural network(1)Comparing the input value with the real value in the training sample and calculating a loss function, thereby realizing one-time circulation; then x is put(2)And the output value h of the previous round(1)Re-input into the self-circulation neural network and then output h(2)And by analogy, after all samples in a certain sequence in the training data are finished, updating the parameters of the self-circulation neural network by using a back propagation algorithm, and finally finishing the training of the self-circulation neural network.
Step 8, during actual load identification, constructing initial graph node data g according to the initial load state in the circuit(0)And converted into initial graph node frequency domain data h(0)(ii) a When the load change occurs in the circuit, i is made to be 1, and the electrical parameter change x in the circuit is obtained(1)X is to(1)And h(0)Input into a self-circulation neural network, and output h(1)To h is aligned with(1)Obtaining graph node data g by performing inverse Fourier transform(1)(ii) a When the load in the circuit changes, i is added with 1, and the cyclic calculation of the self-circulation neural network is performed, wherein the self-circulation neural network utilizes x(i)And h(i-1)Calculate the output h(i)To h is aligned with(i)Carrying out inverse Fourier transform to obtain graph node data g(i)According to g(i)A current load condition in the circuit is identified.
Wherein, for h(i)Carrying out inverse Fourier transform to obtain graph node data g(i)The calculation process of (2) is as follows: the following constraints are added to the scheme, from graph g(i)To figure g(i+1)In the course of load change, increase or decrease of child nodesThe number being at most 1, i.e. g(i)And g(i+1)The number of n in (1) cannot be changed more than 1. To achieve this, the neural network is first outputArray h of(i)By inverse Fourier transform intoData of (2) as original dataAt this time, the number of child nodes included in the original data is not yet determined. According to constraints, if the previous time g(i-1)The number of the child nodes in the node is j, and the value at the moment is only three possible, j-1, j, j +1, according to the constraintRespectively reserving the first j-1, j, j +1 data to obtain three arraysAnd performing 0-complementing operation at the rear part of the three arrays, and expanding all the operations toThen Fourier transform is carried out on the expanded array, and the transform result and h are compared(i)Comparing, and selecting the result with the minimum square difference as the final g(i)Thereby enabling identification of the current load condition in the circuit.
This flow ends by this point.
The above embodiments only describe the design principle of the present invention, and the shapes and names of the components in the description may be different without limitation. Therefore, a person skilled in the art of the present invention can modify or substitute the technical solutions described in the foregoing embodiments; such modifications and substitutions do not depart from the spirit and scope of the present invention.
Claims (9)
1. An intelligent electrical load identification method, comprising:
step 1, simulating daily electricity utilization conditions, increasing and decreasing electric appliances in a circuit to realize the change of a circuit total load, and recording the electric parameter variable quantity x of a circuit total incoming line when the load changes(i)While simultaneously converting x(i)The corresponding current circuit load condition is recorded as graph node data g(i);
Wherein, at the ith moment when the electrical parameter changes, the change data of the electrical parameter is added with the current time and environment additional data and is recorded as the electrical parameter change x at the ith moment(i)(ii) a Graph node data g(i)By n sub-nodesN is the variable total load number in the circuit; each child nodeCorresponding to a single electric load in the circuit, k is equal to [0, n-1 ]],Records the load typeWith the current load powerGraph node data g(i)From the graph node data g of the previous moment(i-1)Increasing or decreasing child nodes to obtain;
step 2, because the number of the child nodes is not fixed, the graph node data g is caused(i)Dimension is not fixed, and graph node data g is subjected to(i)Carrying out fast Fourier transform and converting the fast Fourier transform into graph node frequency domain data h with fixed dimension(i)(ii) a The electrical parameter variation x(i)And graph node frequency domain data h(i)Forming a training sample;
step 3, utilizing a hidden Markov model to carry out electric parameter variation x(i)And graph node frequency domain data h(i)Modeling is carried out, the graph node frequency domain data are regarded as hidden variables of the electrical parameter variable quantity, and the graph node frequency domain data h(i)Graph node frequency domain data h only subjected to last moment(i-1)The influence of (a);
step 4, realizing a hidden Markov model by using a self-circulation neural network; the input of the self-circulation neural network is the electrical parameter variable quantity x at the moment i(i)And graph node frequency domain data h at the time of i-1(i-1)And outputting graph node frequency domain data h at the moment i(i)(ii) a Performing cyclic training on the self-circulation neural network by adopting the training sample obtained in the step 2;
step 5, during actual load identification, constructing initial graph node data g according to the initial load state in the circuit(0)And converted into initial graph node frequency domain data h(0)(ii) a When the load change occurs in the circuit, the i is equal to 1, and the electrical parameter change x in the circuit is obtained(1)X is to be(1)And h(0)Input into a self-circulation neural network, and output h(1)To h is aligned with(1)Obtaining graph node data g by inverse Fourier transform(1)(ii) a When the load in the circuit changes, i is added with 1, and the cyclic calculation of the self-circulation neural network is performed, wherein the self-circulation neural network utilizes x(i)And h(i-1)Calculate the output h(i)To h is aligned with(i)Carrying out inverse Fourier transform to obtain graph node data g(i)According to g(i)A current load condition in the circuit is identified.
2. The method according to claim 1, wherein in step 2, the graph node data g is compared with(i)Fourier transform is carried out, and the Fourier transform is converted into graph node frequency domain data h with fixed dimensions(i)Comprises the following steps:
let g(i)Wherein n current contains n child nodes, each child node is composed of 2 data, then g(i)Can be expressed as an array of n x 2 dimensions, using the speedFourier transform algorithm will g(i)The n x 2 dimensional array is converted into m x 2 dimensional frequency domain data, wherein m is the preset frequency domain sampling number, and the value is ensured to be always larger than g in actual use(i)The number of child nodes n.
3. The method of claim 2, wherein in step 4, h is measured(i)Carrying out inverse Fourier transform to obtain graph node data g(i)Comprises the following steps:
m x 2 dimensional h for self-circulation neural network output(i)After inverse Fourier transform, the data with m × 2 dimensions is obtained and recorded as original dataIf the previous moment g(i-1)If the number of child nodes in the node is j, the original data isRespectively reserving the first j-1, j, j +1 data to obtain three arraysAnd performing 0 complementing operation at the rear part of the three arrays, expanding the arrays to mx 2, performing fast Fourier transform on the expanded arrays, and outputting the transform result and h output by the self-circulation neural network(i)Comparing, and selecting the result with the minimum square difference as the final g(i)For identifying the current load condition in the circuit.
4. The method of claim 2, wherein m is set to 100 in a home scenario and 1000 in an industrial power environment.
5. The method of claim 1, wherein only one appliance load is changed at a single change in the total load of the circuit at step 1.
6. A method as claimed in claim 1, characterized in that the occurrence of a load change in the circuit is identified by: the intelligent electric meter is installed at the general incoming line of the circuit, so that the collection of electric parameters in the circuit is realized, and the load change is identified according to the sudden change of the electric parameters.
7. The method of claim 1, wherein the electrical parameter variation x(i)The variation data of said electrical parameter in (a) comprises the amount of variation of total active power, reactive power, apparent power, voltage current true rms value, phase angle and frequency in the electrical circuit.
8. The method of claim 7, wherein for x, x(1)Wherein the variation data of the electrical parameter is the first recorded electrical parameter data without the variation.
9. The method of claim 1, wherein the electrical parameter variation x(i)The environment additional data in (1) includes a current time, an air temperature, and an air humidity.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110053866.9A CN112968519B (en) | 2021-01-15 | 2021-01-15 | Intelligent power load identification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110053866.9A CN112968519B (en) | 2021-01-15 | 2021-01-15 | Intelligent power load identification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112968519A CN112968519A (en) | 2021-06-15 |
CN112968519B true CN112968519B (en) | 2022-05-13 |
Family
ID=76271510
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110053866.9A Active CN112968519B (en) | 2021-01-15 | 2021-01-15 | Intelligent power load identification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112968519B (en) |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106022645B (en) * | 2016-06-07 | 2021-02-12 | 深圳市芝电能源科技有限公司 | Non-invasive online real-time power load identification method and identification system |
CN107122790B (en) * | 2017-03-15 | 2021-06-11 | 华北电力大学 | Non-invasive load identification algorithm based on hybrid neural network and ensemble learning |
CN108960180A (en) * | 2018-07-17 | 2018-12-07 | 泉州装备制造研究所 | Object detecting method based on the inseparable wavelet character of omnidirection two dimension |
-
2021
- 2021-01-15 CN CN202110053866.9A patent/CN112968519B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN112968519A (en) | 2021-06-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107591844B (en) | Active power distribution network robust reconstruction method considering node injection power uncertainty | |
Dinesh et al. | Residential appliance identification based on spectral information of low frequency smart meter measurements | |
CN110380444B (en) | Capacity planning method for distributed wind power orderly access to power grid under multiple scenes based on variable structure Copula | |
CN109633301B (en) | Non-invasive electrical appliance load identification method based on quantum genetic optimization | |
Fan et al. | Uncertainty evaluation algorithm in power system dynamic analysis with correlated renewable energy sources | |
Yang et al. | Monitoring data factorization of high renewable energy penetrated grids for probabilistic static voltage stability assessment | |
CN109412161A (en) | A kind of Probabilistic Load calculation method and system | |
Dejamkhooy et al. | Non–intrusive appliance load disaggregation in smart homes using hybrid constrained particle swarm optimization and factorial hidden markov model | |
Yao et al. | Short‐Term Power Load Forecasting of Integrated Energy System Based on Attention‐CNN‐DBILSTM | |
Han et al. | Non‐intrusive load monitoring by voltage–current trajectory enabled asymmetric deep supervised hashing | |
Xu et al. | Short-term wind speed prediction based on GRU | |
Ramadan et al. | Towards energy‐efficient smart homes via precise nonintrusive load disaggregation based on hybrid ANN–PSO | |
CN112968519B (en) | Intelligent power load identification method | |
CN106250658A (en) | On-load switch electromagnetic mechanism quick calculation method based on RBF | |
Paraskevas et al. | Domestic smart metering infrastructure and a method for home appliances identification using low‐rate power consumption data | |
Xu et al. | A real‐time state estimation framework for integrated energy system considering measurement delay | |
Zheng et al. | Non-intrusive load monitoring based on the graph least squares reconstruction method | |
Liu et al. | Non-intrusive load monitoring method considering the time-segmented state probability | |
CN115859190A (en) | Non-invasive household electrical classification method based on causal relationship | |
Zeng et al. | Load component decomposition and online modelling based on small disturbance response characteristic matching | |
Ming et al. | Electrical load prediction in energy internet via linear correlation coefficient approach | |
CN105930613A (en) | Equivalent modeling method for distributed power generation system | |
CN117559450B (en) | Method for improving non-intrusive load monitoring precision based on feedback model | |
Hong et al. | Deep‐Learning‐Assisted Topology Identification and Sensor Placement for Active Distribution Network | |
Zhang et al. | Wind energy forecasting framework based on new perspective |
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 |