CN113158446B - Non-invasive electrical load identification method - Google Patents
Non-invasive electrical load identification method Download PDFInfo
- Publication number
- CN113158446B CN113158446B CN202110372297.4A CN202110372297A CN113158446B CN 113158446 B CN113158446 B CN 113158446B CN 202110372297 A CN202110372297 A CN 202110372297A CN 113158446 B CN113158446 B CN 113158446B
- Authority
- CN
- China
- Prior art keywords
- initial state
- load
- power
- loads
- characteristic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 239000002131 composite material Substances 0.000 claims abstract description 22
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 15
- 239000013598 vector Substances 0.000 claims abstract description 15
- 238000007781 pre-processing Methods 0.000 claims abstract description 13
- 230000005684 electric field Effects 0.000 claims abstract description 10
- 238000001228 spectrum Methods 0.000 claims abstract description 6
- 238000005070 sampling Methods 0.000 claims description 21
- 230000002068 genetic effect Effects 0.000 claims description 15
- 230000005611 electricity Effects 0.000 claims description 4
- 230000035772 mutation Effects 0.000 claims description 4
- 238000005457 optimization Methods 0.000 abstract description 6
- 230000006870 function Effects 0.000 description 27
- 238000010586 diagram Methods 0.000 description 10
- 238000012545 processing Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000003860 storage Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012806 monitoring device Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R21/00—Arrangements for measuring electric power or power factor
- G01R21/06—Arrangements for measuring electric power or power factor by measuring current and voltage
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Geometry (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Power Engineering (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention provides a non-invasive power load identification method, which is used for collecting load voltage data and current data and extracting total orthogonal current harmonic frequency spectrum characteristics; carrying out windowing pretreatment to obtain windowed power data; determining initial states of all loads based on the generated initial state library; establishing a composite characteristic objective function model according to the windowed power data and the power characteristic vectors and harmonic characteristic vectors of all loads; and determining the parameter initial value of the composite characteristic objective function model based on the initial state of each load, and solving the composite characteristic objective function model to obtain an optimal power load solution. According to the method, the data preprocessing and the recognition algorithm optimization are effectively combined, and the original data windowing preprocessing and the initial state pre-judgment are utilized, so that the operation complexity is reduced; the characteristic model is optimized by utilizing the characteristic of orthogonal current harmonic waves with larger difference between loads, so that the accuracy of load identification under the electric field scene is improved.
Description
Technical Field
The invention relates to the technical field of power, in particular to a non-invasive power load identification method.
Background
The non-invasive electric load identification technology is characterized in that the electric power consumption information is collected through a monitoring device at the entrance of an electric power user, and then the working states of various electric appliances of the user are identified through analyzing the information such as total current, total voltage and the like at the entrance. The technology can feed back analysis results to users, guide the electricity consumption behaviors of the users, scientifically formulate planning schemes and provide electricity consumption suggestions, thereby promoting the realization of energy conservation and consumption reduction. Compared with the traditional invasive monitoring technology, the method reduces equipment installation cost and later maintenance consumption. The technology has very important use value and research significance for green, continuous and coordinated development, energy production promotion and consumption revolution in China.
At present, as the types of the current household appliances are gradually increased, the original data volume of the load basic characteristics is larger, the identification processing process is more complex, and the identification efficiency is reduced; the similarity of load characteristics such as partial electric appliance power, current and the like becomes high, so that characteristic overlapping phenomenon is caused, the identification effect is poor, and under the actual electric field scene, the condition that a high-power electric appliance and a low-power electric appliance operate simultaneously usually exists, interference and noise influence of high-power non-stable load fluctuation exist, and the low-power load identification accuracy is low. According to the research state of a non-invasive power load identification technology, the technology mainly establishes a power load model according to the electrical characteristics of the power load, and then utilizes pattern identification and optimization technology to realize the decomposition of the power load, such as a clustering algorithm, a machine learning algorithm, an evolution algorithm and the like. However, the research of the existing fuzzy cluster recognition is carried out on the premise that the number and the type of loads, namely the cluster number, are known, and when the number and the type of loads are unknown or various electric appliances run simultaneously, the type and the running state of the electric appliances cannot be accurately recognized; the machine learning method has higher calculation efficiency, but the accuracy in load monitoring and recognition is greatly influenced by random initial results, weights and the like; the evolutionary algorithm can improve the accuracy of the load identification result in an iterative optimization mode, but the calculation efficiency is low. Comprehensive current research results are less related to non-invasive load recognition schemes for improving recognition algorithms and optimizing operation efficiency.
Disclosure of Invention
The invention aims to provide a high-efficiency and accurate non-invasive charge identification method, which reduces the operation complexity of a load identification algorithm and improves the accuracy and the anti-interference capability.
In order to achieve the technical purpose, the invention adopts the following technical scheme.
The invention provides a non-invasive power load identification method, which comprises the following steps:
Collecting load voltage data and current data and extracting load power characteristics and total orthogonal current harmonic spectrum characteristics; carrying out windowing pretreatment based on load voltage data, current data and load power characteristics to obtain windowed power data;
Determining initial states of all loads based on the generated initial state library;
establishing a composite characteristic objective function model according to the windowed power data and the power characteristic vectors and harmonic characteristic vectors of all loads; and determining the parameter initial value of the composite characteristic objective function model based on the initial state of each load, and solving the composite characteristic objective function model to obtain an optimal power load solution.
Further, the total orthogonal current harmonic frequency characteristic H (k) is expressed as follows:
Where U (T) is the load voltage at the sampling time T, i (T) is the load current at the sampling time T, U rms is the root mean square voltage of the voltage U (T), N is the total number of loads in the power utilization scene, N is the sequence number of the loads in the power utilization scene, and T is the sampling time period.
Further, the expression for windowing preprocessing the collected voltage data, current data and orthogonal current harmonic frequency characteristics is as follows:
Wherein P (t) is the total load power at the sampling time t, and P m (t) is the power data obtained after windowing and preprocessing the load power P (t) at the sampling time t; w (T) is the window function, D is the total number of windowed windows, T w is the window width of the window function, and β i is the window overlap ratio.
Further, the method for determining the initial states of all loads based on the generated initial state library is as follows:
Step a, randomly generating an initial state library State0={S0_1,S0_2,…,S0_Q},S0_j={δ0_j1,δ0_j2,…,δ0_jN},j=1,2,…,Q,S0_j at the moment t to represent the j-th operation initial state of all loads in the initial state library, wherein Q is the total number of states in the initial state library; delta 0_ji epsilon {0,1}, i=1, 2, …, N being the total number of loads in the scene with electric field, delta 0_ji representing the jth initial State of the ith load in initial State library State0,
Step b: calculating the distance difference between each initial State in the initial State library State0 at the moment t and the power data obtained after windowing pretreatment of the load power P (t) at the sampling moment t, and determining the minimum distance difference delta, wherein the minimum distance difference delta is expressed as follows:
Wherein δ 0_ji ∈ {0,1}, i=1, 2, …, N is the total number of loads in the scene of the electric field, δ 0_ji represents the j-th initial state of the i-th load in the initial state library, j=1, 2, …, Q is the total number of initial states in the initial state library; p ji is the power of a certain load in its jth initial state;
Step c: determining an initial state S 0_min corresponding to the minimum value of the deviation;
Step d: judging whether delta is less than or equal to epsilon, and updating epsilon value if the delta is less than or equal to epsilon, wherein epsilon is a threshold value; otherwise, the initial state S 0_min determined in the step c is replaced by the initial state generated again at random;
step e: the determined initial state S 0_min is taken as an all-load initial state.
Still further, the composite feature objective function model is represented as follows:
Wherein D (j, t) represents an objective function associated with the sampling time t and the initial state; p m (t) is the power data obtained after windowing and preprocessing the load power P (t) of the original sampling time t; h (k) is the characteristic value of the total harmonic load at the sampling time t; the power eigenvector p= [ P j1,pj2,…,pjN ] of all loads under the load state S 0_j, the harmonic eigenvector h= [ H j1,hj2,…,hjN ] of all loads, ω represents the weight value of the model focusing on the power characteristic, ω' represents the weight value of the model focusing on the orthogonal current harmonic characteristic.
Still further, the method for solving the composite characteristic objective function model is as follows:
Taking initial State library State0 as a genetic initial population { S 0_1,S0_2,…,S0_Q }, wherein each State individual is an N-dimensional binary vector S 0_j; setting genetic iteration parameters; determining initial parameter values of the composite characteristic objective function model according to the determined power characteristic vectors and harmonic characteristic vectors of all loads corresponding to the initial states S 0_min of all loads and the initial states of all loads;
Calculating the fitness of the population according to the objective function, determining an optimal value in the population, judging whether the maximum genetic algebra is reached, and performing selection, crossing and mutation operations if the maximum genetic algebra is not reached to generate a new initial next-generation population, and repeatedly calculating the fitness of the population according to the objective function; when the maximum genetic algebra is reached, the individual with the highest fitness is selected from the optimal individuals of each generation and is used as the optimal solution output of the algorithm.
Based on the technical scheme, the method further comprises the following steps: the initial State library State0 is updated by the following steps: and replacing the State S 0_max corresponding to the maximum value of the distance difference in the initial State library State0 with the State with the highest occurrence frequency in the selected identification period to form a new initial State library State0.
Further, the method for updating the threshold epsilon is as follows: and replacing the original threshold epsilon by adopting the median value of the distance difference in each state in the initial state library.
The beneficial technical effects obtained by the invention are as follows:
According to the method, the data preprocessing and the recognition algorithm optimization are effectively combined, and the original data windowing preprocessing and the initial state pre-judgment are utilized, so that the operation complexity is reduced; the characteristic model is optimized by utilizing the characteristic of orthogonal current harmonic waves with larger difference between loads, so that the accuracy of load identification under the electric field scene is improved. The method solves the problem of poor identification effect when the load active and reactive power characteristics are close in the prior art, and simultaneously solves the problem of remarkably increased operation time caused in the algorithm optimization process;
The invention carries out pretreatment on each consistent initial state, optimizes the initial state of each load, improves the recognition efficiency of the non-invasive power load recognition method and ensures that the method converges more quickly.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute a limitation on the application. In the drawings:
FIG. 1 is a schematic diagram of a non-intrusive power load identification method according to an embodiment of the present invention;
FIG. 2 is a flow chart of a non-intrusive power load identification method according to an embodiment of the present invention;
FIG. 3 is a block diagram of a collection device used in accordance with an embodiment of the present invention;
FIG. 4 is a schematic diagram of windowing preprocessing of raw data of load characteristics in an embodiment of the invention;
FIG. 5 is a graph showing the relationship between window width, overlap ratio and operation time according to an embodiment of the present invention.
Detailed Description
The application will be described in detail hereinafter with reference to the drawings in conjunction with embodiments. It should be noted that, without conflict, the embodiments of the present application and features of the embodiments may be combined with each other.
An embodiment one, the present embodiment provides a non-invasive power load identification method, including the following steps, and a schematic structural diagram of the method provided in the present embodiment is shown in fig. 1:
Step 1: load voltage data and current data are collected. In the embodiment, an intelligent non-invasive load characteristic acquisition device shown in fig. 2 is adopted to acquire the basic characteristics of load voltage and current signals. The intelligent non-invasive load characteristic acquisition device is shown in fig. 3, and comprises a monitoring device, a data acquisition device and a peripheral device. The monitoring equipment comprises an intelligent monitor, a concentrator, a controller and the like, the data acquisition equipment comprises a communication bus, a total control module, a storage unit and the like, and the peripheral equipment comprises a display module and an input device. The signals such as voltage and current of the electric appliance are obtained through the device.
Step 2: based on the basic characteristic data in the step 1, extracting load characteristics such as power and current harmonic of the electric appliance, including extracting load characteristics such as power and current harmonic of the electric appliance (i.e. load), and extracting total orthogonal current harmonic spectrum characteristics.
The difference of the orthogonal current harmonic frequency domain characteristics of different loads is larger, which is helpful to improve the accuracy of load identification, so that besides the power characteristics, the orthogonal current harmonic frequency spectrum H (k) is extracted in the step 2 as follows:
Wherein the load voltage and current are respectively represented as U (t) and i (t), U rms is root mean square voltage of the voltage U (t), and N is total load in the field scene.
Step 3: windowing pretreatment is carried out on the collected voltage data, current data and orthogonal current harmonic frequency characteristics, and power data after windowing is obtained;
determining initial states of all loads based on the generated initial state library; the step is based on the load characteristics in the step 2, the original data preprocessing is carried out, and the operation efficiency is improved.
3.1 Windowing pretreatment.
The windowing pretreatment of the original data takes the principle that effective data is not lost, so that the data volume is reduced, and the operation complexity is reduced. In the fixed electric field scene, the load state change is usually periodic, a period value can be judged through the original data acquisition and analysis, then effective data is extracted by adding a time window to the load power characteristic data in the whole time period range, and the windowing mode is shown in fig. 4. Important parameters related to the windowing pretreatment process comprise windowing positions, windowing sizes and efficiency amplification, wherein the windowing positions are selected to be positions with mutation or characteristic value fluctuation; the initial window size is larger, the subsequent window size is decreased, and the window length can be further reduced through long-time data accumulation; the efficiency increase is related to the window size and window overlap ratio. The windowed load power signature data may be expressed as:
Wherein P (T) is the total power of the sampling time T, w (T) is a window function, D is the total number of windowing, T is the window width, and beta is the window overlap ratio. As shown in fig. 5, the larger the overlap ratio, that is, the higher the window overlap ratio, the longer the calculation time.
3.2 Initial state prejudging, namely determining the initial states of all loads based on the generated initial state library. The estimated initial state is used as the input of the next recognition algorithm, so that the subsequent processing time can be reduced.
The estimated initial state is used as the input of the next step, so that the subsequent processing time can be reduced. For non-intrusive load identification, the initial state of operation of all appliances may be denoted as S 0={δ01,δ02,…,δ0N},δ0i e {0,1}, i=1, 2, …, N. According to the data after windowing pretreatment in the step 3.1, outputting the initial state S 0 of each load through the single power target characteristic value delta and the preset threshold epsilon, wherein the estimation method comprises the following steps:
Step a, randomly generating an initial state library State0={S0_1,S0_2,…,S0_Q},S0_j={δ0_j1,δ0_j2,…,δ0_jN},j=1,2,…,Q,S0_j at the moment t to represent the j-th operation initial state of all loads in the initial state library, wherein Q is the total number of states in the initial state library; delta 0_ji epsilon {0,1}, i=1, 2, …, N being the total number of loads in the scene of the electric field, delta 0_ji representing the jth initial State of the ith load in the initial State library State 0;
Step b: calculating the distance difference between each initial State in the initial State library State0 at the moment t and the power data obtained after windowing pretreatment of the load power P (t) at the sampling moment t, and determining the minimum distance difference delta, wherein the minimum distance difference delta is expressed as follows:
wherein p ji is the power of a certain electric appliance in the j-th state, the larger the Q value is, the higher the complexity of data preprocessing is, but the lower the complexity of subsequent recognition processing is. Delta 0_ji epsilon {0,1}, i=1, 2, …, N is the total number of loads in the scene of the electric field, delta 0_ji represents the j-th initial state of the i-th load in the initial state library, j=1, 2, …, Q is the total number of initial states in the initial state library; p ji is the power of a certain load (i-th load) in its j-th initial state;
Step c: determining an initial state S 0_min corresponding to the minimum value of the deviation;
Step d: judging whether delta is less than or equal to epsilon, and updating epsilon value if the delta is less than or equal to epsilon, wherein epsilon is a threshold value; otherwise, the initial state S 0_min determined in the step c is replaced by the initial state generated again at random;
step e: the determined initial state S 0_min is taken as an all-load initial state.
Step 4: establishing a composite characteristic objective function model according to the windowed power data and the power characteristic vectors and harmonic characteristic vectors of all loads; and determining the parameter initial value of the composite characteristic objective function model based on the initial state of each load, and solving the composite characteristic objective function model to obtain an optimal power load solution.
The purpose of the step 4 is to improve accuracy, improve and optimize a single objective function on the basis of a traditional genetic algorithm, comprehensively utilize load power characteristics and orthogonal current harmonic characteristics, and establish a composite characteristic objective function model. And then obtaining an optimal solution through genetic iteration based on the composite characteristic objective function model according to the load characteristic data obtained after pretreatment and the initial state.
Because the frequency domain characteristics of the orthogonal current harmonics of different loads have larger difference, the characteristic of the orthogonal current harmonics is introduced in the step to enhance the difference between the load characteristics, and the identification accuracy and the anti-interference performance are improved. The load power characteristic and the orthogonal current harmonic characteristic are combined, and an objective function model based on the composite characteristic is established as follows:
Using D (j, t) to represent an objective function related to the sampling time t and the initial state; the initial state library randomly generated in the step 3 is used as a genetic initial population { S 0_1,S0_2,…,S0_Q }, wherein each state individual is an N-dimensional binary vector S 0_j={δ0_j1,δ0_j2,,…,δ0_jN }; the state S 0_min generated in step 3 determines the initial parameter values of the parameters δ 0_ji、pji and h ji, that is, the initial value of δ 0_ji is the value of the state S 0_min corresponding to the ith load, and the initial value of p ji takes the power value of the ith load in the ith state S 0_min; the initial value of h ji is the harmonic wave (namely the orthogonal current harmonic spectrum) of the ith load in the state S 0_min;
Monitoring the extracted total power and total harmonic load characteristic values at a certain moment, and respectively representing the total power and the total harmonic load characteristic values as P m (t) and H k (t) after pretreatment; the values of the power eigenvectors p= [ P j1,pj2,…,pjN ], the harmonic eigenvectors h= [ H j1,hj2,…,hjN ], ω and ω 'of all the loads in the state S 0_j represent whether the model is focused on the power characteristic or the orthogonal current harmonic characteristic, ω represents the weight value of the model focused on the power characteristic, and ω' represents the weight value of the model focused on the orthogonal current harmonic characteristic.
Performing multi-objective optimization according to the model, and performing multiple iterations to obtain an optimal individual: (1) Substituting the initial value of the parameter determined according to the initial state S 0_min according to the objective function model D (j, t), and setting a genetic iteration parameter; (2) Calculating the fitness of the population according to the objective function, recording the optimal value in the population, judging whether the maximum genetic algebra is reached, executing the step (4), otherwise, continuing; (3) Performing selection, crossing and mutation operations to generate a new next generation population, and repeatedly executing the step (2); (4) And selecting an individual with highest fitness from the optimal individuals of each generation as an optimal solution output of the algorithm.
In order to adaptively adjust the value of the update threshold epsilon on the basis of the first embodiment, the second embodiment enables the set threshold to adapt to the change of the initial state and be more reasonable, and the embodiment further includes: the original threshold epsilon is replaced by the median value of the distance difference in each state in the initial state library, and the calculation formula of the distance difference in each state in the initial state library is as follows:
The third embodiment, on the basis of the above embodiments, further includes: the initial State library State0 is updated by the following steps: and replacing the State S 0_max corresponding to the maximum value of the distance difference in the initial State library State0 with the State with the highest occurrence frequency in the selected identification period to form a new initial State library State0.
In step a in the initial state pre-judging step 3.2, after the initial state library is randomly generated, the method provided by the embodiment is adopted to update the random initial state library, and the updated initial state library of each load is used as the initial state library for the next round of recognition, so that the recognition efficiency and the convergence rate are further improved.
In addition, if the initial population adopts the initial state library of each load generated randomly in the process of solving the composite characteristic objective function model in the step 4, the initial population can be updated by adopting the steps in the embodiment, so that the recognition efficiency and the convergence rate are further improved. The flow chart of the method provided by the embodiment is shown in fig. 2.
In the above embodiments, the steps are numbered for convenience of description, and the order of execution of the steps is not to be construed as being limited.
The invention relates to a high-efficiency accurate non-invasive power load identification method, which adopts an intelligent non-invasive load characteristic acquisition device to acquire and store voltage and current signals under a power field scene, and can obtain a load identification result efficiently and accurately through key steps such as data windowing pretreatment, initial state generation, identification algorithm analysis, genetic iteration and the like.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The embodiments of the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those having ordinary skill in the art without departing from the spirit of the present invention and the scope of the claims, which are all within the protection of the present invention.
Claims (4)
1. A method of non-intrusive electrical load identification, comprising the steps of:
Collecting load voltage data and current data and extracting load power characteristics and total orthogonal current harmonic spectrum characteristics;
carrying out windowing pretreatment based on load voltage data, current data and load power characteristics to obtain windowed power data;
Determining initial states of all loads based on the generated initial state library;
Establishing a composite characteristic objective function model according to the windowed power data and the power characteristic vectors and harmonic characteristic vectors of all loads; determining a parameter initial value of the composite characteristic objective function model based on the initial state of each load, and solving the composite characteristic objective function model to obtain an optimal power load solution;
the total quadrature current harmonic frequency characteristic H (k) is expressed as follows:
Wherein U (T) is the load voltage at the sampling time T, i (T) is the load current at the sampling time T, U rms is the root mean square voltage of the voltage U (T), N is the total number of loads in the electricity scene, N is the sequence number of the loads in the electricity scene, and T is the sampling time period;
the expression for windowing preprocessing the collected voltage data, current data and orthogonal current harmonic frequency characteristics is as follows:
Wherein P (t) is the total load power at the sampling time t, and P m (t) is the power data obtained after windowing and preprocessing the load power P (t) at the sampling time t; w (T) is a window function, D is the total number of windowed windows, T w is the window width of the window function, and beta i is the window overlap ratio;
the method for determining the initial states of all loads based on the generated initial state library is as follows:
Step a, randomly generating an initial state library State0={S0_1,S0_2,…,S0_Q},S0_j={δ0_j1,δ0_j2,…,δ0_jN},j=1,2,…,Q,S0_j at the moment t to represent the j-th operation initial state of all loads in the initial state library, wherein Q is the total number of states in the initial state library; delta 0_ji epsilon {0,1}, i=1, 2, …, N being the total number of loads in the scene with electric field, delta 0_ji representing the jth initial State of the ith load in initial State library State0,
Step b: calculating the distance difference between each initial State in the initial State library State0 at the moment t and the power data obtained after windowing pretreatment of the load power P (t) at the sampling moment t, and determining the minimum distance difference delta, wherein the minimum distance difference delta is expressed as follows:
Wherein δ 0_ji ∈ {0,1}, i=1, 2, …, N is the total number of loads in the scene of the electric field, δ 0_ji represents the j-th initial state of the i-th load in the initial state library, j=1, 2, …, Q is the total number of initial states in the initial state library; p ji is the power of a certain load in its jth initial state;
Step c: determining an initial state S 0_min corresponding to the minimum value of the deviation;
Step d: judging whether delta is less than or equal to epsilon, and updating epsilon value if the delta is less than or equal to epsilon, wherein epsilon is a threshold value; otherwise, the initial state S 0_min determined in the step c is replaced by the initial state generated again at random;
step e: taking the determined initial state S 0_min as an initial state of all loads;
the composite characteristic objective function model is expressed as follows:
Wherein D (j, t) represents an objective function associated with the sampling time t and the initial state; p m (t) is the power data obtained after windowing and preprocessing the load power P (t) of the original sampling time t; h (k) is the characteristic value of the total harmonic load at the sampling time t; the power eigenvector p= [ P j1,pj2,…,pjN ] of all loads under the load state S 0_j, the harmonic eigenvector h= [ H j1,hj2,…,hjN ] of all loads, ω represents the weight value of the model focusing on the power characteristic, ω' represents the weight value of the model focusing on the orthogonal current harmonic characteristic.
2. The non-intrusive electrical load identification method of claim 1, wherein the method of solving the composite characteristic objective function model is as follows:
Taking initial State library State0 as a genetic initial population { S 0_1,S0_2,…,S0_Q }, wherein each State individual is an N-dimensional binary vector S 0_j; setting genetic iteration parameters; determining initial parameter values of the composite characteristic objective function model according to the determined power characteristic vectors and harmonic characteristic vectors of all loads corresponding to the initial states S 0_min of all loads and the initial states of all loads;
Calculating the fitness of the population according to the objective function, determining an optimal value in the population, judging whether the maximum genetic algebra is reached, and performing selection, crossing and mutation operations if the maximum genetic algebra is not reached to generate a new initial next-generation population, and repeatedly calculating the fitness of the population according to the objective function; reaching the maximum number of genetics, selecting the most fitness from the optimal individuals for each generation
High individuals, as the optimal solution output of the algorithm.
3. The non-intrusive electrical load identification method of claim 2, further comprising: the initial State library State0 is updated by the following steps: and replacing the State S 0_max corresponding to the maximum value of the distance difference in the initial State library State0 with the State with the highest occurrence frequency in the selected identification period to form a new initial State library State0.
4. The non-intrusive electrical load identification method of claim 1, in which the method of updating the threshold epsilon is as follows:
and replacing the original threshold epsilon by adopting the median value of the distance difference in each state in the initial state library.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110372297.4A CN113158446B (en) | 2021-04-07 | 2021-04-07 | Non-invasive electrical load identification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110372297.4A CN113158446B (en) | 2021-04-07 | 2021-04-07 | Non-invasive electrical load identification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113158446A CN113158446A (en) | 2021-07-23 |
CN113158446B true CN113158446B (en) | 2024-06-11 |
Family
ID=76888574
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110372297.4A Active CN113158446B (en) | 2021-04-07 | 2021-04-07 | Non-invasive electrical load identification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113158446B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107525964B (en) * | 2017-10-23 | 2023-11-21 | 云南电网有限责任公司电力科学研究院 | Non-invasive load identification method and device based on fusion decision |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104777383A (en) * | 2015-04-16 | 2015-07-15 | 武汉阿帕科技有限公司 | Non-invasive electrical load monitoring and load decomposing device |
CN105429135A (en) * | 2015-12-08 | 2016-03-23 | 河南许继仪表有限公司 | Distinguishing and decision-making method and system for noninvasive power load decomposition |
CN106022645A (en) * | 2016-06-07 | 2016-10-12 | 李祖毅 | Non-intruding type on-line and real-time electric power load identification method and identification system |
CN106655160A (en) * | 2016-10-27 | 2017-05-10 | 国家电网公司 | Non-intrusion electric power load decomposition identification decision method and system |
CN106786534A (en) * | 2016-12-28 | 2017-05-31 | 天津求实智源科技有限公司 | A kind of non-intrusive electrical load transient process discrimination method and system |
CN109541346A (en) * | 2018-11-23 | 2019-03-29 | 四川长虹电器股份有限公司 | A kind of non-intrusion type electrical load under steady state condition identifies method for improving |
CN110866214A (en) * | 2019-11-07 | 2020-03-06 | 昆明理工大学 | Non-invasive load decomposition based on improved differential evolution algorithm |
CN112101110A (en) * | 2020-08-13 | 2020-12-18 | 西安理工大学 | Non-invasive load identification method for user side of power system |
-
2021
- 2021-04-07 CN CN202110372297.4A patent/CN113158446B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104777383A (en) * | 2015-04-16 | 2015-07-15 | 武汉阿帕科技有限公司 | Non-invasive electrical load monitoring and load decomposing device |
CN105429135A (en) * | 2015-12-08 | 2016-03-23 | 河南许继仪表有限公司 | Distinguishing and decision-making method and system for noninvasive power load decomposition |
CN106022645A (en) * | 2016-06-07 | 2016-10-12 | 李祖毅 | Non-intruding type on-line and real-time electric power load identification method and identification system |
CN106655160A (en) * | 2016-10-27 | 2017-05-10 | 国家电网公司 | Non-intrusion electric power load decomposition identification decision method and system |
CN106786534A (en) * | 2016-12-28 | 2017-05-31 | 天津求实智源科技有限公司 | A kind of non-intrusive electrical load transient process discrimination method and system |
CN109541346A (en) * | 2018-11-23 | 2019-03-29 | 四川长虹电器股份有限公司 | A kind of non-intrusion type electrical load under steady state condition identifies method for improving |
CN110866214A (en) * | 2019-11-07 | 2020-03-06 | 昆明理工大学 | Non-invasive load decomposition based on improved differential evolution algorithm |
CN112101110A (en) * | 2020-08-13 | 2020-12-18 | 西安理工大学 | Non-invasive load identification method for user side of power system |
Non-Patent Citations (2)
Title |
---|
基于竞争聚集和神经网络的非侵入式负载监测方法;陈于锋;;江苏通信(06);全文 * |
非侵入式电力负荷分解算法综述;张玲玲;;安徽电子信息职业技术学院学报(01);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN113158446A (en) | 2021-07-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110956220B (en) | Non-invasive household appliance load identification method | |
CN108616120B (en) | Non-invasive power load decomposition method based on RBF neural network | |
CN111160798B (en) | Non-invasive household appliance load identification method based on bee colony algorithm | |
CN109633301B (en) | Non-invasive electrical appliance load identification method based on quantum genetic optimization | |
CN108599172B (en) | Transmission and distribution network global load flow calculation method based on artificial neural network | |
CN111563827B (en) | Load decomposition method based on physical characteristics of electrical appliance and electricity utilization behaviors of residents | |
Zhou et al. | Non-intrusive load monitoring using a CNN-LSTM-RF model considering label correlation and class-imbalance | |
CN114444539A (en) | Power load identification method, apparatus, device, medium, and program product | |
CN113158446B (en) | Non-invasive electrical load identification method | |
CN111563631A (en) | Wind power generation power prediction method and device based on variational modal decomposition | |
Jiang et al. | A new method for recognition and classification of power quality disturbances based on IAST and RF | |
CN113408622A (en) | Non-invasive load identification method and system considering characteristic quantity information expression difference | |
CN116756575A (en) | Non-invasive load decomposition method based on BGAIN-DD network | |
CN112434783A (en) | Non-invasive load decomposition method and system | |
CN115528684A (en) | Ultra-short-term load prediction method and device and electronic equipment | |
CN113595242B (en) | Non-invasive load identification method based on depth CNN-HMM | |
CN113033775B (en) | Non-invasive load identification network architecture based on supervised learning | |
Liu et al. | Nonintrusive load disaggregation combining with external attention mechanism and Seq2Piont | |
CN112016684B (en) | Electric power terminal fingerprint identification method of deep parallel flexible transmission network | |
Yan et al. | A clustering method for power time series curves based on improved self-organizing mapping algorithm | |
Ding et al. | A novel non-intrusive load monitoring method based on quantum particle swarm optimization algorithm | |
CN111948446A (en) | Self-adaptive load identification method and intelligent electric energy meter | |
CN115048854B (en) | Household electrical appliance load decomposition method | |
Qiu et al. | A semi-supervised load identification method with class incremental learning | |
Yang et al. | Research on Non-intrusive Load Decomposition Based on FHMM |
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 |