CN112381665A - Multi-feature weighted household electrical load identification method based on chicken swarm algorithm - Google Patents
Multi-feature weighted household electrical load identification method based on chicken swarm algorithm Download PDFInfo
- Publication number
- CN112381665A CN112381665A CN202011036602.4A CN202011036602A CN112381665A CN 112381665 A CN112381665 A CN 112381665A CN 202011036602 A CN202011036602 A CN 202011036602A CN 112381665 A CN112381665 A CN 112381665A
- Authority
- CN
- China
- Prior art keywords
- value
- current
- attribute
- chicken
- matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 241000287828 Gallus gallus Species 0.000 title claims abstract description 92
- 238000000034 method Methods 0.000 title claims abstract description 56
- 244000144992 flock Species 0.000 claims abstract description 53
- 239000011159 matrix material Substances 0.000 claims abstract description 47
- 235000013330 chicken meat Nutrition 0.000 claims description 80
- 230000005611 electricity Effects 0.000 claims description 17
- 238000005457 optimization Methods 0.000 claims description 10
- 238000004458 analytical method Methods 0.000 claims description 9
- 238000000354 decomposition reaction Methods 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 5
- 230000000694 effects Effects 0.000 claims description 4
- 230000005484 gravity Effects 0.000 claims description 4
- 230000006978 adaptation Effects 0.000 claims description 3
- 230000003247 decreasing effect Effects 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 239000004576 sand Substances 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 abstract description 8
- 230000002431 foraging effect Effects 0.000 description 7
- 230000006399 behavior Effects 0.000 description 6
- 235000013305 food Nutrition 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004134 energy conservation Methods 0.000 description 2
- 244000144980 herd Species 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 241000272201 Columbiformes Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000004459 forage Substances 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 239000010931 gold Substances 0.000 description 1
- 229910052737 gold Inorganic materials 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 239000003960 organic solvent Substances 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Finance (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Data Mining & Analysis (AREA)
- General Business, Economics & Management (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Game Theory and Decision Science (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Tourism & Hospitality (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention relates to a multi-feature weighted household electrical load identification method based on a chicken flock algorithm, which comprises the following steps: step 1, installing a sensor at a user port, and counting active power, reactive power and harmonic current of each electrical attribute in the obtained user basic data; step 2, optimizing the objective function of the active power, the reactive power and the harmonic current of each electrical attribute in the obtained user basic data; obtaining an optimal value of a target function by using an improved discrete chicken flock algorithm; and 3, constructing an attribute decision matrix for the electrical operation result obtained by each electrical characteristic, acquiring an entropy value of the electrical characteristic quantity by using a structure entropy weight method, and combining the weight determined by the entropy value method with the assignment of the identification matrix corresponding to the identification matrix obtained in the step 2 to obtain the final load service condition. The invention solves the problem of accuracy of non-invasive load monitoring for load identification in the prior art.
Description
Technical Field
The invention relates to the field of electric power, is used in non-invasive electrical equipment, and particularly relates to a multi-feature weighted household electrical load identification method based on a chicken flock algorithm.
Background
The reasonable planning and use of energy resources are always a hot problem in the world, the situation of energy shortage is aggravated nowadays, the environmental problem is also increasingly severe, and all countries advocate energy conservation and emission reduction and improve the energy utilization rate. Therefore, the government of China is also a series of actively established policies for energy conservation and emission reduction. The electric power part also actively responds to the national call, further plans the interior of the power grid, optimizes the operation load and reduces the loss. At present, the living standard of people is greatly improved, and the specific gravity of the electricity consumption of residents is gradually increased. Therefore, the power utilization condition of residents is analyzed, the power utilization of the residents is reasonably arranged, low-carbon life is advocated, energy consumption is reduced, and the method has great significance for protecting natural environment and natural resources.
Under the background of the smart power grid, a large amount of electricity utilization data of residents are collected, and analysis and deep utilization of the electricity utilization habits of users become research hotspots, which is the actual basis for exploring frequent interaction between the smart power grid and the users and analyzing the electricity utilization behaviors of the users, reasonably planning and guiding the electricity utilization habits of the users. How to collect detailed data of residential electricity consumption becomes a primary problem. At first, intrusive load monitoring is the primary choice, but intrusive monitoring requires a sensor to be installed on each appliance, although the real-time power consumption of the appliance and related electrical parameters, such as voltage, current, frequency, active power, reactive power and power factor, can be accurately obtained when the appliance is used. However, a large number of sensors and data transmission devices need to be installed, on one hand, a large number of installation and maintenance works can affect the life of users, the popularization difficulty of equipment on the user side is increased, and on the other hand, the cost of the equipment is greatly increased due to the construction of a large number of sensors and data transmission channels.
Based on this, Hart teaches that a Non-invasive Load Monitoring NILM (Non-invasive Load Monitoring) method is proposed in 1982, and the operation conditions of each electrical device in a user room are calculated by detecting electrical information at a user port and decomposing the total power consumption of a user by using a mathematical algorithm according to the electrical information. Compared with invasive monitoring, the NILM has low cost, more convenient installation and maintenance and strong popularization and use value, and has wide application prospect in the load monitoring scene of which the resident family is the main body. However, the accuracy of non-intrusive load identification is still far from adequate compared to intrusive load identification.
Disclosure of Invention
The invention provides a non-invasive load identification method, which aims to at least optimize and improve the technical problem of low accuracy of load identification of non-invasive load monitoring in the prior art.
The invention is realized by at least one of the following technical schemes.
A household electrical load identification method based on multi-feature weighting of a chicken flock algorithm comprises the following steps:
step 1, basic data processing: installing a sensor at a user port, collecting user electrical data, and counting the electricity consumption habits of residential and civilian users to serve as a basic database; in the obtained user basic data, counting the active power, the reactive power and the harmonic current of each electrical attribute;
step 2, optimizing an objective function: optimizing the objective function of the active power, the reactive power and the harmonic current of each electrical attribute in the obtained user basic data; obtaining an optimal value of an objective function by using an improved discrete chicken flock algorithm;
step 3, weighted analysis of a structure entropy weight method: and (3) constructing an attribute decision matrix for the electric operation result obtained by each electric characteristic, acquiring an entropy value of the electric characteristic quantity by using a structure entropy weight method, and finally identifying the load use condition by combining the weight determined by the entropy value method with the assignment of the identification matrix according to the obtained identification matrix. According to the method, collected residential and civil ionization dispersion data are further classified and refined, and through weight proportion setting and optimization by combining a chicken flock algorithm, a non-invasive identification result is more consistent with residential electricity utilization habits, and the identification result is more accurate.
Further, the step 1 comprises:
the method comprises the following steps of collecting various electrical characteristics of the electric appliance used by a user by utilizing a sensor arranged on the side of a user terminal: performing load decomposition operation on active power, reactive power and steady harmonic current;
the active power formula is as follows:
wherein V (t) is the instantaneous voltage at time t, Ii(T) is the instantaneous current at time T, T is the alternating current cycle time, dt is the integration factor, IiRepresents the current value of the ith device;
the active power load power consumption decomposition establishes an optimized objective function:
in the formula PiThe real power of the ith device is recorded in the database; p is the current total active power; x is the number ofiRepresenting the switching state of the ith equipment, wherein when the switching state is 0, the ith equipment is not operated, when the switching state is 1, the ith equipment is operated, and N represents all the equipment quantity;
reactive power formula:
establishing an optimized objective function for reactive power load power consumption decomposition:
in the formula QiThe real power of the ith device is recorded in the database; q current total reactive power;
steady state harmonic current: when a certain family contains m types of main household electrical equipment and n types of independent working states, the stable-state operation of the electrical equipment has superposition, and the total current is equal to the linear superposition of the n working states, namely:
iL(t)＝x1is1(t)+x2is2(t)+…+xnisn(t)
wherein il (t) is the total current at the user side; i.e. is1、is2、is3…isnOperating currents for states 1,2, … n, respectively; x is the number of1、x2、…、xn∈{0,1}The operating states respectively represent 1 st and 2 … n home appliance states, the value of 0 represents the non-operating state, and the value of 1 represents the operating state;
when the above formula is expressed by a vector method, the expression is:
wherein, thetaL,hIs the phase angle, I, of the total currentL,hIs the value of the total current, thetasn,hPhase angle value of partial current, Isn,h
Is the value of the partial current, n, h is used for marking the information of the current, h represents that the dimension of odd harmonic waves describing the current is 1, 3, 5 and 7, n represents a current belonging to the partial current, IL,1∠θL,1Represents the sum of the first harmonics of all branch currents; x is the number of1，x2，…，xn∈{0,1}Is the variable to be solved of the equation set; the elements in the matrix on the left side of the equation are known quantities measured in real time, and the elements in the matrix on the right side of the equation are quantities which can be known through statistics of an offline basic database;
obtaining the actual current change value as a target function after the household appliance of the user is turned on and off:
mind(3)＝||IL-ISX||
in the formula: i isLCollecting the total current of the household user for the concentrator, ISParametrization for a user's home appliance modelAnd (4) counting a matrix, wherein X is a working state matrix of each household appliance load in the residential user, and the load identification is converted into the solution of the optimal combination problem.
Further, the step 2 comprises: aiming at the fact that the load working state of the household appliance is only in an on state and an off state, namely 0 and 1, discretization needs to be carried out on a chicken flock algorithm, the chicken flock algorithm is improved into a discrete binary chicken flock algorithm, namely after cluster positions are updated, a Sigmoid function is adopted, the Sigmoid function is abbreviated as sig (), the value range of the function is (0,1), and the expression of the improved discrete binary chicken flock algorithm is as follows:
the expression of the improved cock is as follows:
in the formula:
represents the position of the chicken at the time t + 1';
representing the position of the chicken after sigmoid function mapping at the time of t + 1; i. j represents xi.jPosition of serial number, randn (0, σ)2) Denotes a mean value of 0 and a standard deviation of σ2The gaussian distribution of (e), exp () represents an exponential function of e;
indicating that a numerical decision range is generated based on sigmoid;
the expression of the improved hen is as follows:
s2＝exp(fr2-fi)
wherein rand represents a random number between (0,1), r1＝[1,N]Is the index value of the cock, representing the head in the ith hen mass; and r2＝[1,N]Is the index value of the rooster, but this is randomly chosen from the rooster flock, and r1≠r2，fiTo represent
Correspondingly; f. ofr1And fr2Respectively represent
And
a corresponding adaptation value; s1、s2Determining the corresponding coefficient of hen position in the hen swarm algorithm, alpha is constant to prevent f from occurringiAt 0, zero divides the error.
when the rank order is updated every time, reinitializing the chickens so as to enhance the activity of the chicken flocks and improve the global search capability of the chicken flocks;
and respectively optimizing the three objective functions by using a chicken flock algorithm, wherein the actual current change value is taken as the working state of the household appliances of the residential users in the objective functions and only has two states of on and off, namely only 0 and 1 are needed to be adopted for representation, discretizing the chicken flock algorithm, namely a discrete binary chicken flock algorithm, after the chicken flock is iteratively updated, adopting a sigmoid function, and finally determining whether the position is 0 or 1 by using a mapping method.
Further, the structural entropy weight method weighting analysis in step 3 includes: data acquisition and processing; the weight of the entropy weight method needs to construct an attribute decision matrix, and the attribute decision matrix needs to obtain original data support; the original data is a target function of active power, reactive power and harmonic current characteristics which are optimized by a chicken flock algorithm, and the fitness value of an electric appliance state combination and the target function is obtained; the characteristics of active power, reactive power, harmonic current and the probability of residential electricity consumption behavior are taken as attributes and are expressed as xj(ii) a Then the switch states of the recognition results of the optimization functions are combined to form a scheme, which is expressed as Am(ii) a Finally, the fitness value is used as original data and is expressed as xijForming an attribute decision matrix, raw data xijCarrying out normalization treatment to obtain specific gravity FijAnd obtaining an entropy value:
for the jth attribute, calculating the entropy E of the attribute by using the data column under the attributej：
Wherein, K is a constant, and K is 1/ln (m); m is a scheme number, and E is ensured to be more than or equal to 0jLess than or equal to 1, calculating the deviation degree d of the jth attributejFor the determined j attribute, if the attribute has smaller influence on load identification, the data column under the attribute is closer to the completely unordered state, and EjThe larger the deviation of this influencing factor should be, the smaller it is, thus defining:
dj＝1-Ej；
weight ratio:
the value obtained by normalizing the obtained deviation degree is the weight proportion of each attribute, and reflects the degree of influence of the attribute on load identification, wherein the weight formula of the jth influence factor is as follows:
arranging the switch state combinations of the electric appliance identification results obtained by the characteristics of the electric appliances according to a matrix form, and assigning the identification matrix:
Dn＝[Di(1)Di(2)Di(3)…Di(h)]i is 1,2,3 … V, V is the number of features of the appliance, h is the number of results selected for each feature; i.e. Di(h) The identification characteristics of the electric appliance are i, and h types of results can be obtained;
for each di (j), i ═ 1,2, 3.. v; j is 1,2,3 … c, Wi is assigned, the weighted value of Di (1) is set as c, Di (2) is c-1 according to the similarity, Di (c) is 1 after sequentially decreasing, and the following steps are provided:
Wi(j)＝c+1-i
knowing that the weighted value of Di (j) is the weighted number of the corresponding matrix position in W;
combining the weight determined by the entropy method with the assignment of the identification matrix, multiplying the elements appearing in each row with the corresponding assignment and weight to obtain weights, and summing the weights of the same elements, wherein the largest weight is the determined electric appliance combination type.
Compared with the prior art, the invention has the beneficial effects that:
the chicken flock algorithm belongs to a [0,1] interval and takes any value for optimization, because the load switch state only actually takes 0/1 states, the non-invasive identification common chicken flock algorithm cannot be used, the optimized and improved chicken flock algorithm utilizes discrete thinking, the convergence speed is accelerated, and when the load type is combined with subsequent analysis, multi-feature weighting identification is adopted, load identification is carried out compared with common single features, and the identification reliability can be improved by the multi-feature weighting coefficient identification method.
Drawings
FIG. 1 is a schematic flow chart of a household electrical load identification method based on multi-feature weighting of a chicken flock algorithm according to the present embodiment;
fig. 2 is a schematic diagram of a multi-feature load identification flow of the structure entropy weight method according to the embodiment.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the first, second, etc. of the present invention are used for distinguishing similar objects, and are not necessarily used for describing a specific order or sequence. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Moreover, the terms "comprises," "comprising," and any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The technical terms involved in the present invention are explained first as follows:
load characteristics: the load characteristics refer to the electrical behavior of the load in the normal working process, and the load characteristics are related electrical parameters such as current, active power, reactive power and the like extracted after waveform transformation of voltage and current waveforms or data processing, which are known from the analysis of the load characteristics per se.
And (3) the electricity consumption behavior of residents: the resident electricity consumption behavior is the resident electricity consumption law, and the data are analyzed through the data acquisition of the resident user in a period of time to obtain the resident electricity consumption law. For example, when an appliance is turned on, how long it is turned on, and a probability statistic is made for it.
Chicken flock algorithm: assuming that the potential solution of the combinatorial optimization problem is one chicken, calculating the objective function value of each chicken, and dividing the chicken group into cocks, hens and chickens according to the size of the objective function value. The method comprises the steps that the target function value of the cock is the smallest and is closest to a target object, the cock serves as the chicken with the highest foraging capability in each colony, each chicken forages in the colony of the pigeons, and the position of the cock is adjusted according to own experience and the experience of the cock in the colony in the process of searching for food. In the iterative process, after the set times of intervals, the levels and the groups of the target function values are sorted and divided again according to the sequence of the target function values.
In the setting, the foraging capacity of the cocks is the strongest, and when the cocks are used for searching for food, the cocks have considerable advantages compared with other chicken breeds, namely, the chickens with strong foraging capacity are wider in the range of searching for food than the chickens with weak foraging capacity, and the following expression is used for expressing:
in the formula, the first and second organic solvents are,
representing the position of the cock at the time t; randn (0, sigma)2) Denotes a mean value of 0 and a standard deviation of σ2(ii) a gaussian distribution; ε is a very small constant (which may be 0.001) to avoid the denominator of 0 in the formula; k represents the sequence number of the cock and is randomly selected from the cock group; n represents the total number of roosters; f. ofiTo represent
Corresponding value of the objective function, fkTo represent
A corresponding objective function value;
for hens, the cocks in the herd will be tracked to find food together, but will also steal food that other herds have found, and the expression is as follows:
in the formula: rand is taken as [0,1]]A random number in between; r1 ═ 1, N]Is the index value of the cock, which represents the cock in the group in which the ith hen is located; and r2 ═ 1, N]Is the index value of the rooster, which is randomly selected from the rooster flock, and r1≠r2，fr1And fr2Respectively represent
And
a corresponding objective function value; f. ofiTo represent
A corresponding objective function value;
the foraging capacity of the chick is the worst, so the chick can only be foraged around the chick-mother, the chick-mother is randomly selected from the hen group, and the mother-child relationship between the chick and the chick-mother is also randomly established. The foraging process of chickens can be described by the following formula:
in the formula:
indicates the location of the chicken mother of the ith chick, and x ∈ [1, M ]]M is the total number of hens; FL is a parameter, FL e (0, 2) indicates that chicks will follow their mother to feed, and the FL value for each chick will be chosen randomly from 0 to 2, taking into account individual variability.
And grading at intervals according to the size of the objective function value of each chicken, and then searching for a global optimal solution according to the formula.
Improving a chicken flock algorithm: aiming at the fact that the load working state of the household appliance is only on and off, namely 0 and 1, the chicken swarm algorithm needs to be discretized, the chicken swarm algorithm is improved into a discrete binary chicken swarm algorithm, namely after the cluster position is updated, a Sigmoid function is adopted, the Sigmoid function is abbreviated as sig (), the value range of the function is (0,1), one real number can be mapped to the interval of (0,1), and the two-classification method can be used for two-classification. From this mapping it is determined whether the chicken position takes 0 or 1, and the expression of the improved discrete binary chicken flock algorithm is as follows:
the expression of the improved cock is as follows:
in the formula:
represents the position of the chicken at the time t + 1';
representing the position of the chicken after sigmoid function mapping at the moment of t + 1; i. j represents xi.jPosition of serial number, randn (0, σ)2) Denotes a mean value of 0 and a standard deviation of σ2The gaussian distribution of (e), exp () represents an exponential function of e;
indicating that a numerical decision range is generated based on sigmoid;
the expression of the improved hen is as follows:
s2＝exp(fr2-fi)
wherein rand represents a random number between (0,1), r1＝[1,N]Is the index value of the cock, representing the head in the ith hen mass; and r2＝[1,N]Is the index value of the rooster, but this is randomly chosen from the rooster flock, and r1≠r2，fiTo represent
Correspondingly; f. ofr1And fr2Respectively represent
And
a corresponding adaptation value; s1、s2For the corresponding coefficient for determining the hen position in the hen flock algorithm, alpha is 1 very small constant to prevent f from occurringiAt 0, zero divides the error.
the foraging capacity of the chickens is the weakest, so that the capacity of the chicken flocks for searching the global optimal solution is enhanced in order to follow the rule of natural object competition, the chicken flock algorithm is prevented from falling into the local optimal solution, and therefore, when the rank order is updated every time, the small gold is reinitialized, the activity of the chicken flocks is enhanced, and the global searching capacity of the chicken flocks is improved.
As shown in fig. 1 and fig. 2, the present embodiment provides a household electrical load identification method based on multi-feature weighting of a chicken swarm algorithm, which utilizes an improved chicken swarm algorithm to perform load decomposition identification operation on electrical appliance features of a household user, and then performs weighting analysis on identification results of various electrical features to obtain results, and the method includes the following steps:
step 1: obtaining basic data: installing a sensor at a user port, collecting user electrical data, and counting electricity utilization habits of residential users to serve as a basic database; in the obtained user basic data, counting the active power, the reactive power and the harmonic current of each electrical attribute;
step 2: data processing: and performing target optimization on active power, reactive power and harmonic current of each electrical attribute in possible user basic data.
Active power characteristic formula:
wherein V (t) is the instantaneous voltage at time t, Ii(T) is instantaneous current, T is alternating current cycle time, dt is an integral factor, IiRepresents the current value of the ith device;
establishing an optimized objective function for active power load power consumption decomposition:
in the formula PiIs the active power of the ith device of the database; and P is the current total active power. x is the number ofiThe element is {0,1}, which respectively represents the switching working condition of the state of the household appliance, the value of 0 represents that the household appliance is in a non-working state, and the value of 1 represents that the household appliance is in a working state; n represents the total number of home devices
Reactive power characteristics:
calculating the formula:
establishing an optimized objective function for reactive power load power consumption decomposition:
in the formula QiIs the active power of the ith device of the database; q current total reactive power;
harmonic current characteristics: in steady state operation, the steady state fundamental current and the harmonic current have the characteristics of periodicity and regularity, and the steady state current can be expressed as:
is(t)＝Is1cos(ωt+θs1)+···Iskcos(kωt+θsk)+···
is(t) is the current instantaneous value of a household electrical appliance under a certain stable working condition; i iss1Is the amplitude of the fundamental component of the operating current in the operating state, omega being the angular frequency of the fundamental component in the operating current, thetas1Is the initial phase angle of the fundamental component in the working current; k is an integer and represents the number of harmonics, and when k is 1, represents the fundamental wave; i isskIs the amplitude of the kth harmonic component in the operating current; θ sk is the initial phase angle of the kth harmonic component in the operating current.
When a certain household contains m types of main household electrical equipment, n types of independent working states, the working state with lower power and the household electrical equipment are ignored, the superposition is realized when the household electrical equipment is operated in a steady state, and the total current is approximately equal to the linear superposition of the n working states, namely:
iL(t)＝x1is1(t)+x2is2(t)+…+xnisn(t)
in the formula iL(t) is the total current on the user side; i.e. is1、is2、is3…isnOperating currents for states 1,2, … n, respectively; x is the number of1、x2、…、xn∈{0,1}The operating states respectively represent 1 st and 2 … n home appliance states, the value of 0 represents the non-operating state, and the value of 1 represents the operating state;
because even harmonic content in the power grid is low, odd harmonic components are mainly considered; when the above formula is expressed by a vector method, it can be expressed as:
wherein, thetaL,hIs the phase angle, I, of the total currentL,hIs the value of the total current, thetasn,hPhase angle value of partial current, Isn,hIs the value of the partial current (n, h is information for marking the current, h denotes that the dimension of odd harmonics describing the current is 1, 3, 5, 7, etc., n denotes which particular one of the partial currents belongs to, example Is3,5∠θs3,5Indicating at a partial current Is3∠θs3Value of the fifth harmonic of (I)LIs the abbreviation of total current, Is the abbreviation of branch current), IL,1∠θL,1Representing the sum of the first harmonics of all branch currents), x1，x2，…，xn∈{0,1}Is the variable to be solved of the equation set; the elements in the matrix on the left side of the equation are known quantities obtained by real-time measurement, and the elements in the matrix on the right side of the equation are known quantities which can be known through statistics of an offline basic database;
obtaining the actual current change value as a target function after the household appliance of the user is turned on and off:
mind(3)＝||IL-ISX||
in the formula: i isLCollecting the total current of the household user for the concentrator, ISAnd (3) converting the load identification into the solution of the optimal combination problem, wherein the parameter matrix of the user household appliance equipment model is used, and X is the working state matrix of each household appliance load in the residential user.
And respectively optimizing the three objective functions by using a chicken flock algorithm, wherein the household appliance working states of the residential users in the objective function III only have an on state and an off state, namely only 0 and 1 are needed for representation, and discretizing the chicken flock algorithm, namely the discrete binary chicken flock algorithm. After the chicken flock is updated iteratively, a sigmoid function is adopted, and finally whether the position is 0 or 1 is determined by a mapping method.
In order to more reliably improve the identification result of the non-invasive load and the influence of different characteristic quantities on the accuracy of the load identification result, the obtained target function is subjected to multi-characteristic weighting discrimination, namely, the obtained information is subjected to quantitative analysis by using a structure entropy weight method, and the final weight is determined by combining subjective assignment of qualitative analysis, so that the final load identification result is comprehensively obtained.
The information entropy is a quantity used for describing the disorder degree of information contained in an information theory, the greater the value of the entropy is, the higher the disorder degree is, and the lower the effect occupied by the corresponding information is through a disorder program of the information entropy measurement information, and based on the principle, the information entropy can be used for determining the judgment weight of each attribute in the load identification, and the accuracy of the load identification is optimized and improved.
In the multi-feature load recognition by using the structure entropy weight method, the optimization purpose of a target function is realized by adopting a chicken swarm algorithm, and compared with a discrete particle swarm algorithm, the chicken swarm algorithm carries out random initialization again when the rank order of a chicken is updated every time, so that the whole chicken swarm can carry out continuous optimization, the global search capability of the algorithm is improved, namely when the discrete particle swarm algorithm obtains a local optimal solution to obtain a conclusion, the chicken swarm algorithm still continues optimization search, the global optimal solution is finally obtained, and the working type and the working state of a household appliance are more accurately recognized.
Step 3, weighted analysis of a structure entropy weight method: and (3) constructing an attribute decision matrix for the electric operation result obtained by each electric characteristic, acquiring an entropy value of the electric characteristic quantity by using a structure entropy weight method, and combining the weight determined by the entropy value method and the assignment of the identification matrix corresponding to the identification matrix obtained in the step (2) to obtain the final load service condition.
As shown in fig. 2, the structural entropy weight method specifically includes the following steps:
1) and (6) data acquisition and processing. The weight of the entropy weight method needs to construct an attribute decision matrix, and the attribute decision matrix needs to obtain original data support. The original data is the real power, the reactive power and the harmonic of the chicken swarm algorithmAnd optimizing the objective function of the wave current characteristics to obtain the state combination of the electric appliance and the fitness value of the objective function. Active power, reactive power, harmonic current characteristics and residential electricity consumption behavior probability are taken as attributes and are expressed as xj(ii) a Then, the switch states of the recognition results of the optimization functions are combined as a scheme, which is expressed as Am(ii) a Finally, the fitness value is used as the original data and is expressed as xijForming an attribute decision matrix:
wherein n represents the types of the electric appliances and xinRepresents a correspondence AiConversion attribute value of nth electrical appliance in switch state matrix, original data xijNormalized by a specific gravity Fij：
2) Entropy value:
for the jth attribute, the data column under the attribute is used for obtaining the entropy E of the attributej：
Wherein, K is a constant, and K is 1/ln (m); m is a scheme number, and E is ensured to be more than or equal to 0jLess than or equal to 1, calculating the deviation degree d of the jth attributejFor the determined j attribute, if the attribute has smaller influence on load identification, the data column under the attribute is closer to the completely unordered state, and EjThe larger the deviation of this influencing factor should be, the smaller it is, thus defining:
dj＝1-Ej
3) weight proportion
The value obtained by normalizing the obtained deviation degree is the weight proportion of each attribute, and reflects the degree of influence of the attribute on load identification, wherein the weight formula of the jth influence factor is as follows:
4) forming an identification matrix: arranging the switch state combinations of the electric appliance identification results obtained by the characteristics of the electric appliances according to a matrix form:
Dn＝[Di(1)Di(2)Di(3)…Di(h)]i is 1,2,3 … V, V is the number of features of the appliance, h is the number of results selected for each feature; i.e. Di(h) The identification characteristics of the electric appliance are i, and h types of results can be obtained;
and (3) identifying the matrix for assignment: to each Di(j) (i-1, 2,3 … n; j-1, 2,3 … m) assigned a value of WiAccording to the similarity, Di(1) The weighted value is set to c, Di(2) Then is c-1, and D is decreased successivelyi(c) Is 1, has:
Wi(j)＝c+1-i
Di(j) the weighting value of (a) is the weighting at the corresponding matrix position in W.
Combining the weights determined by the entropy method with the assignments to the identification matrix. And multiplying the elements appearing in each row by the corresponding assignment and the weight to obtain weights, and summing the weights of the same elements, wherein the largest weight is the judged electric appliance combination type.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.
Claims (4)
1. A household electrical load identification method based on multi-feature weighting of a chicken flock algorithm is characterized by comprising the following steps:
step 1, basic data processing: installing a sensor at a user port, collecting user electrical data, and counting electricity consumption habits of residential users to serve as a basic database; in the obtained user basic data, counting the active power, the reactive power and the harmonic current of each electrical attribute;
step 2, optimizing an objective function: optimizing the objective function of the active power, the reactive power and the harmonic current of each electrical attribute in the obtained user basic data; obtaining an optimal value of an objective function by using an improved discrete chicken flock algorithm;
step 3, weighted analysis of a structure entropy weight method: and (3) constructing an attribute decision matrix for the electric operation result obtained by each electric characteristic, acquiring an entropy value of the electric characteristic quantity by using a structure entropy weight method, and finally identifying the load use condition by combining the weight determined by the entropy value method with the assignment of the identification matrix according to the obtained identification matrix.
2. The method for identifying the household electrical load based on the multiple characteristic weights of the chicken flock algorithm according to the claim 1, wherein the step 1 comprises the following steps:
the method comprises the following steps of collecting various electrical characteristics of the electric appliance used by a user by utilizing a sensor arranged on the side of a user terminal: performing load decomposition operation on active power, reactive power and steady harmonic current;
the active power formula is as follows:
wherein,v (t) is the instantaneous voltage at time t, Ii(T) is the instantaneous current at time T, T is the alternating current cycle time, dt is the integration factor, IiRepresents the current value of the ith device;
the active power load power consumption decomposition establishes an optimized objective function:
in the formula PiThe real power of the ith device is recorded in the database; p is the current total active power; x is the number ofiRepresenting the switching state of the ith equipment, wherein when the switching state is 0, the ith equipment is not operated, when the switching state is 1, the ith equipment is operated, and N represents all the equipment quantity;
reactive power formula:
establishing an optimized objective function for reactive power load power consumption decomposition:
in the formula QiThe real power of the ith device is recorded in the database; q current total reactive power;
steady state harmonic current: when a certain household contains m types of main household electrical equipment and n types of independent working states, the steady-state operation of the electrical equipment has superposition, and the total current is equal to the linear superposition of the n working states, namely:
iL(t)＝x1is1(t)+x2is2(t)+…+xnisn(t)
in the formula iL(t) is the total current on the user side; i.e. is1、is2、is3…isnAre respectively No. 1,2.… n; x is the number of1、x2、...、xnE {0,1} respectively represents the working conditions of 1 st and 2 … th home appliance states, the value of 0 represents the non-working state, and the value of 1 represents the working state;
when the above formula is expressed by a vector method, the expression is:
wherein, thetaL,hIs the phase angle, I, of the total currentL,hIs the value of the total current, thetasn,hPhase angle value of partial current, Isn,hIs the value of the partial current, n, h is used for marking the information of the current, h represents that the dimension of odd harmonic waves describing the current is 1, 3, 5 and 7, n represents a current belonging to the partial current, IL,1∠θL,1Represents the sum of the first harmonics of all branch currents; x is the number of1，x2，...，xnE {0,1} is a variable to be solved of the equation set; the elements in the matrix on the left side of the equation are known quantities obtained by real-time measurement, and the elements in the matrix on the right side of the equation are known quantities which can be known through statistics of an offline basic database;
obtaining the actual current change value as a target function after the household appliance of the user is turned on and off:
mind(3)＝||IL-ISX||
in the formula: i isLCollecting the total current of the household user for the concentrator, ISAnd (3) converting the load identification into the solution of the optimal combination problem, wherein the parameter matrix of the user household appliance equipment model is used, and X is the working state matrix of each household appliance load in the residential user.
3. The chicken flock algorithm-based household electrical load identification method based on multi-feature weighting according to claim 1, wherein the step 2 comprises the following steps: aiming at the fact that the load working state of the household appliance is only in an on state and an off state, namely 0 and 1, discretization needs to be carried out on a chicken flock algorithm, the chicken flock algorithm is improved into a discrete binary chicken flock algorithm, namely after cluster positions are updated, a Sigmoid function is adopted, the Sigmoid function is abbreviated as sig (), the value range of the function is (0,1), and the expression of the improved discrete binary chicken flock algorithm is as follows:
the expression of the improved cock is as follows:
in the formula:
represents the position of the chicken at the time t + 1';
representing the position of the chicken after sigmoid function mapping at the moment of t + 1; i. j represents xi.jPosition of serial number, randn (0, σ)2) Denotes a mean value of 0 and a standard deviation of σ2The gaussian distribution of (e), exp () represents an exponential function of e;
indicating that a numerical decision range is generated based on sigmoid;
the expression of the improved hen is as follows:
s2＝exp(fr2-fi)
wherein rand represents a random number between (0,1), r1＝[1,N]Is the index value of the cock, which represents the head in the ith hen group; and r2＝[1,N]Is the index value of the rooster, but this is randomly chosen from the rooster flock, and r1≠r2，fiTo represent
Correspondingly; f. ofr1And fr2Respectively represent
And
a corresponding adaptation value; s1、s2Determining the corresponding coefficient of hen position in the hen swarm algorithm, alpha is constant to prevent f from occurringiWhen the value is 0, zero divides the error;
when the rank order is updated every time, reinitializing the chickens so as to enhance the activity of the chicken flocks and improve the global search capability of the chicken flocks;
and respectively optimizing the three objective functions by using a chicken flock algorithm, wherein the actual current change value is taken as the working state of the household appliances of the residential users in the objective functions and only has two states of on and off, namely only 0 and 1 are needed to be adopted for representation, discretizing the chicken flock algorithm, namely a discrete binary chicken flock algorithm, after the chicken flock is iteratively updated, adopting a sigmoid function, and finally determining whether the position is 0 or 1 by using a mapping method.
4. The method for identifying the household electrical load based on the multi-feature weighting of the chicken flock algorithm according to the claim 1, wherein the structural entropy weighting analysis in the step 3 comprises: data acquisition and processing; the weight of the entropy weight method needs to construct an attribute decision matrix, and the attribute decision matrix needs to obtain original data support; the original data is the target function of the chicken flock algorithm for optimizing the characteristics of active power, reactive power and harmonic current, and the fitness value of the electric appliance state combination and the target function is obtained; the characteristics of active power, reactive power, harmonic current and the probability of residential electricity consumption behavior are taken as attributes and are expressed as xj(ii) a Then the switch states of the recognition results of the optimization functions are combined to form a scheme, which is expressed as Am(ii) a Finally, the fitness value is used as original data and is expressed as xijForming an attribute decision matrix, raw data xijCarrying out normalization treatment to obtain specific gravity FijAnd obtaining an entropy value:
for the jth attribute, calculating the entropy E of the attribute by using the data column under the attributej：
Wherein, K is a constant, and K is 1/ln (m); m is a scheme number, and E is ensured to be more than or equal to 0jLess than or equal to 1, calculating the deviation degree d of the jth attributejFor the determined j attribute, if the attribute has smaller influence on load identification, the data column under the attribute is closer to the completely unordered state, and EjThe larger the deviation of this influencing factor should be, the smaller it is, thus defining:
dj＝1-Ej；
weight ratio:
the value obtained by normalizing the obtained deviation degree is the weight proportion of each attribute, and reflects the degree of influence of the attribute on load identification, wherein the weight formula of the jth influence factor is as follows:
arranging the switch state combinations of the electric appliance identification results obtained by the characteristics of the electric appliances according to a matrix form, and assigning the identification matrix:
Dn＝[Di(1) Di(2) Di(3) … Di(h)]i is 1,2,3 … V, V is the number of features of the appliance, h is the number of results selected for each feature; i.e. Di(h) The identification characteristics of the electric appliance are i, and h types of results can be obtained;
for each di (j), i ═ 1,2,3 … v; j is 1,2,3 … c, Wi is assigned, the weighted value of Di (1) is set as c, Di (2) is c-1 according to the similarity, Di (c) is 1 after sequentially decreasing, and the following steps are provided:
Wi(j)＝c+1-i
knowing that the weighted value of Di (j) is the weighted number of the corresponding matrix position in W;
combining the weight determined by the entropy method with the assignment of the identification matrix, multiplying the elements appearing in each row with the corresponding assignments and weights to obtain weights, and summing the weights of the same elements, wherein the largest weight is the determined electric appliance combination type.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011036602.4A CN112381665A (en) | 2020-09-28 | 2020-09-28 | Multi-feature weighted household electrical load identification method based on chicken swarm algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011036602.4A CN112381665A (en) | 2020-09-28 | 2020-09-28 | Multi-feature weighted household electrical load identification method based on chicken swarm algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112381665A true CN112381665A (en) | 2021-02-19 |
Family
ID=74579954
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011036602.4A Pending CN112381665A (en) | 2020-09-28 | 2020-09-28 | Multi-feature weighted household electrical load identification method based on chicken swarm algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112381665A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113922373A (en) * | 2021-10-15 | 2022-01-11 | 国网安徽省电力有限公司肥西县供电公司 | Electrical load identification method based on multi-feature index fusion |
| CN114240105A (en) * | 2021-12-03 | 2022-03-25 | 南京南瑞信息通信科技有限公司 | Transformer area line loss analysis method, device and system |
| WO2023228298A1 (en) * | 2022-05-25 | 2023-11-30 | 三菱電機株式会社 | Power consumption estimation device, program, and power consumption estimation method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110009231A (en) * | 2019-04-04 | 2019-07-12 | 上海电力学院 | A kind of power load recognition methods and device based on repeatability and entropy weight distinction |
-
2020
- 2020-09-28 CN CN202011036602.4A patent/CN112381665A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110009231A (en) * | 2019-04-04 | 2019-07-12 | 上海电力学院 | A kind of power load recognition methods and device based on repeatability and entropy weight distinction |
Non-Patent Citations (3)
| Title |
|---|
| 刘兴杰 等: "基于改进鸡群算法的非侵入式负荷监测", 《电力自动化设备》 * |
| 许仪勋 等: "基于改进鸡群算法的非侵入式家电负荷分解", 《电力系统保护与控制》 * |
| 许仪勋 等: "基于结构熵权法的非侵入式家电识别研究", 《电测与仪表》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113922373A (en) * | 2021-10-15 | 2022-01-11 | 国网安徽省电力有限公司肥西县供电公司 | Electrical load identification method based on multi-feature index fusion |
| CN114240105A (en) * | 2021-12-03 | 2022-03-25 | 南京南瑞信息通信科技有限公司 | Transformer area line loss analysis method, device and system |
| WO2023228298A1 (en) * | 2022-05-25 | 2023-11-30 | 三菱電機株式会社 | Power consumption estimation device, program, and power consumption estimation method |
| WO2023228816A1 (en) * | 2022-05-25 | 2023-11-30 | 三菱電機株式会社 | Electric power consumption estimation apparatus, program, and electric power consumption estimation method |
| JP7511799B2 (en) | 2022-05-25 | 2024-07-05 | 三菱電機株式会社 | Power consumption estimation device, program, and power consumption estimation method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112381665A (en) | Multi-feature weighted household electrical load identification method based on chicken swarm algorithm | |
| Jin et al. | Comparison of clustering techniques for residential energy behavior using smart meter data | |
| CN110954744A (en) | Non-invasive load monitoring method based on event detection | |
| CN110991263B (en) | Non-invasive load identification method and system for resisting background load interference | |
| Dou et al. | Hybrid model for renewable energy and loads prediction based on data mining and variational mode decomposition | |
| Wang et al. | An isolation-based distributed outlier detection framework using nearest neighbor ensembles for wireless sensor networks | |
| CN111563827B (en) | Load decomposition method based on physical characteristics of electrical appliance and electricity utilization behaviors of residents | |
| CN109359665B (en) | Household appliance load identification method and device based on support vector machine | |
| CN111553444A (en) | Load identification method based on non-invasive load terminal data | |
| CN115796393B (en) | Energy management optimization method, system and storage medium based on multi-energy interaction | |
| CN106410781A (en) | Power consumer demand response potential determination method | |
| CN113361454A (en) | Deep learning non-invasive load monitoring method based on unsupervised optimization | |
| CN111242276A (en) | One-dimensional convolution neural network construction method for load current signal identification | |
| CN116681186A (en) | Power quality analysis method and device based on intelligent terminal | |
| CN115545333A (en) | Method for predicting load curve of multi-load daily-type power distribution network | |
| CN111815039A (en) | Weekly scale wind power probability prediction method and system based on weather classification | |
| Tang et al. | Semi-intrusive load monitoring for large-scale appliances | |
| CN117318051B (en) | Intelligent energy storage management and control system and method based on Internet of things technology | |
| CN117974395A (en) | Environment-friendly community management method and system based on carbon emission monitoring | |
| CN117540981A (en) | Building multi-energy scheduling method and system based on carbon emission constraint | |
| CN116298576A (en) | Time-segmentation-considered extensible non-invasive load monitoring method | |
| Singh et al. | Grey Wolf Optimization Based CNN-LSTM Network for the Prediction of Energy Consumption in Smart Home Environment | |
| CN116500386A (en) | Wind power plant collector line cable partial discharge signal acquisition and processing method | |
| CN111948446A (en) | Self-adaptive load identification method and intelligent electric energy meter | |
| CN117710153B (en) | Energy utilization auxiliary decision-making method and system based on multi-terminal equipment |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210219 |