CN115310999B - Enterprise electricity behavior analysis method and system based on multi-layer perceptron and sequencing network - Google Patents
Enterprise electricity behavior analysis method and system based on multi-layer perceptron and sequencing network Download PDFInfo
- Publication number
- CN115310999B CN115310999B CN202210737510.1A CN202210737510A CN115310999B CN 115310999 B CN115310999 B CN 115310999B CN 202210737510 A CN202210737510 A CN 202210737510A CN 115310999 B CN115310999 B CN 115310999B
- Authority
- CN
- China
- Prior art keywords
- enterprise
- model
- electricity
- enterprises
- consumption behavior
- 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
- 230000005611 electricity Effects 0.000 title claims abstract description 167
- 238000004458 analytical method Methods 0.000 title claims abstract description 76
- 238000012163 sequencing technique Methods 0.000 title claims abstract description 11
- 230000006399 behavior Effects 0.000 claims abstract description 184
- 239000013598 vector Substances 0.000 claims abstract description 86
- 238000012549 training Methods 0.000 claims abstract description 77
- 238000000034 method Methods 0.000 claims abstract description 30
- 238000012545 processing Methods 0.000 claims abstract description 11
- 238000011156 evaluation Methods 0.000 claims description 92
- 230000006870 function Effects 0.000 claims description 32
- 238000005457 optimization Methods 0.000 claims description 12
- 238000010276 construction Methods 0.000 claims description 7
- 230000003993 interaction Effects 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 6
- 230000001174 ascending effect Effects 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 6
- 230000007547 defect Effects 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 238000007477 logistic regression Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000013210 evaluation model Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/80—Management or planning
- Y02P90/82—Energy audits or management systems therefor
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Finance (AREA)
- Marketing (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Tourism & Hospitality (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an enterprise electricity behavior analysis method and system based on a multi-layer perceptron and a sorting network, wherein the method comprises the following steps: constructing a power utilization characteristic index set to obtain a model index characteristic vector, and scoring power utilization behaviors of the indexes; constructing a training data set, wherein the training data set comprises a feature vector set formed by model index feature vectors of a plurality of enterprises and supervision information; respectively carrying out multi-layer perceptron processing on model index feature vectors of a plurality of enterprises in the training data set to obtain corresponding model index feature coding sets, and training an enterprise electricity consumption behavior analysis model based on a sequencing network; and analyzing the electricity consumption behavior of the enterprise according to the enterprise electricity consumption behavior analysis model and the reference enterprise screened from the training data set. The method can better utilize multidimensional electric power big data, improve the efficiency, accuracy and expandability of the model for the analysis of the enterprise electric behavior, and greatly reduce the cost.
Description
Technical Field
The invention belongs to the technical field of power big data processing, and relates to an enterprise power consumption behavior analysis method and system based on a multi-layer perceptron and a sequencing network.
Background
In the prior art, a traditional expert experience model is generally used for constructing enterprise credit, and the traditional expert experience model usually selects few important data indexes, and judgment is made by using expert experience to obtain the evaluation of enterprise electricity consumption behaviors. However, expert experience models have disadvantages: firstly, when a large amount of power data needs to be analyzed and processed, the labor cost of an expert experience model is overlarge; secondly, the expert experience model can only evaluate the electric power behaviors of enterprises for the scene under the current condition, and the model is difficult to expand; thirdly, when the dimension of the power big data is large, the subjectivity of the expert experience model is strong, and the obtained analysis result of the power consumption behavior of the enterprise is often inaccurate.
Based on this, when an evaluation model is constructed to evaluate the power consumption behavior of an enterprise, the following problems are often faced: the manual labeling of a large amount of electric power data requires quantitative analysis labeling, the cost is high, the subjectivity of the labeled data is high, the accuracy cannot be ensured, and the evaluation model of the enterprise electric behavior is difficult to obtain by using methods such as logistic regression, numerical regression, linear regression and the like.
Disclosure of Invention
In order to solve the defects in the prior art, the invention aims to provide the enterprise electricity behavior analysis method and system based on the multi-layer perceptron and the ordering network, which can better utilize multi-dimensional large electric power data, improve the efficiency, accuracy and expandability of the model on the enterprise electricity behavior analysis and greatly reduce the cost.
In order to achieve the above object, the present invention adopts the following technical scheme:
the enterprise electricity consumption behavior analysis method based on the multi-layer perceptron and the ordering network comprises the following steps:
step 1, constructing a power utilization characteristic index set to obtain a model index characteristic vector, and grading power utilization behaviors of the indexes;
step 2, a training data set is constructed, wherein the training data set comprises a feature vector set formed by model index feature vectors of a plurality of enterprises and monitoring information which is obtained based on the power consumption behavior scores of the indexes and represents preliminary evaluation results of power consumption behaviors among different enterprises;
step 3, respectively carrying out multi-layer perceptron processing on model index feature vectors of a plurality of enterprises in the training data set to obtain corresponding model index feature coding sets, and training an enterprise electricity consumption behavior analysis model based on the ordering network by adopting the model index feature coding sets and the supervision information;
and 4, analyzing the electricity consumption behavior of the enterprise according to the enterprise electricity consumption behavior analysis model and the reference enterprise screened from the training data set.
The invention further comprises the following preferable schemes:
preferably, in step 1, a power consumption characteristic index set including a power consumption load level index, a power consumption standard index and a power consumption interaction capacity index of an enterprise is constructed, and all indexes in the power consumption characteristic index set are spliced in sequence to obtain a model index characteristic vector; and scoring the electricity consumption behavior of each index according to expert experience to construct supervision information.
Preferably, step 2 specifically includes:
step 2.1, constructing a feature vector set of the training data set: taking the model index feature vector of each enterprise as the feature vector of the training data set, and collecting the model index feature vectors of a plurality of enterprises to form a feature vector set;
step 2.2, constructing supervision information of the training data set: pairing and sampling are carried out on enterprises in the training data set, each pair of enterprises is endowed with a label representing the preliminary evaluation result of the power consumption behavior among the enterprises, and the labels representing the preliminary evaluation result of the power consumption behavior among different enterprises form supervision information;
the value mode of the label is as follows:
accumulating the scores of the power consumption behaviors of the enterprises obtained by the scores of the power consumption behaviors of the indexes in the step 1, and using the scores for primarily evaluating the power consumption behaviors of different enterprises;
comparing the primary evaluation scores of the first enterprise and the second enterprise in each pair of enterprises, and if the first enterprise is higher than the second enterprise by a set score, marking the label as 1 to indicate that the electricity utilization performance of the first enterprise is better; if the second enterprise is higher than the first enterprise by a set score, the label is marked as-1, which indicates that the second enterprise has better electricity utilization performance; otherwise, the tag is marked with 0, indicating that the first enterprise and the second enterprise behave identically in terms of electricity.
Preferably, step 3 specifically includes:
step 3.1, respectively performing multi-layer perceptron processing on model index feature vectors a of a plurality of enterprises in the training data set to obtain a model index feature coding set b:
b＝mlp(a,w)
wherein w is the characteristic weight set of each index in the parameter a of the multilayer perceptron mlp;
step 3.2, constructing a sorting network model, and constructing a loss function based on the sorting network model, the supervision information and the symmetry of the sorting relation;
step 3.3, constructing a random gradient descent optimization model based on the loss function so as to learn weights in the model index feature coding set;
and 3.4, constructing and training an enterprise electricity behavior analysis model by adopting a random gradient descent optimization method according to the model index feature coding set and the supervision information.
Preferably, in step 3.2, the ordered network model is constructed as expressed in the following formula,
wherein:
P ij the probability that the ordering network model considers that the electricity utilization behavior of the enterprise i is better than that of the enterprise j is represented;
the power consumption behavior evaluation of the assumption enterprise i is better than that of the assumption enterprise j;
sigma is a super-parameter constant;
s i ＝s(b i ) The enterprise electricity consumption behavior analysis model is represented to give enterprise electricity consumption behavior evaluation values of enterprise i;
s j ＝s(b j ) And the enterprise electricity consumption behavior analysis model is used for giving enterprise j an enterprise electricity consumption behavior evaluation value.
Preferably, in step 3.2, the loss function is constructed based on the ordering network model, the supervision information and the ordering relation symmetry, specifically as follows:
the cross entropy loss function is constructed, expressed in the following formula,
and (3) formulating and simplifying the cross entropy loss function to obtain:
wherein:
C ij values representing the loss functions of enterprise i and enterprise j;
S ij belonging to [ -1,0,1]The relative good-bad relationship of the enterprise electricity consumption behavior between the enterprise i and the enterprise j obtained according to the supervision information is represented;
defining a pairing set I according to the symmetry of the ordering relation, namely the symmetry of the relatively good relation between the first enterprise and the second enterprise, wherein the first enterprise is required to have better electricity consumption behavior than the second enterprise;
the loss function C of set I is defined as follows:
wherein s is satisfied by i >s j 。
Preferably, step 3.3, a random gradient descent optimization model is constructed based on the loss function, expressed in the following formula,
wherein: w (w) k The kth weight of the feature weight set w in the model index feature code set b is represented;
η represents a learning rate;
λ i is a factor for enterprise i.
Preferably, in step 3.4, an enterprise electricity behavior analysis model is constructed based on the random gradient descent optimization model, and is expressed by the following formula,
s i ,s j ＝f(a i ,a j ；w) (4)
wherein: w is a characteristic weight set in the model index characteristic coding set b obtained in the step 3.3;
f(a i ,a j the method comprises the steps of carrying out a first treatment on the surface of the w) represents a model index feature vector a according to the electricity behavior of enterprise i and enterprise j i ,a j Outputting a function of the evaluation value;
s i ＝s(b i ) The enterprise electricity consumption behavior analysis model is represented to give enterprise electricity consumption behavior evaluation values of enterprise i;
s j ＝s(b j ) And the enterprise electricity consumption behavior analysis model is used for giving enterprise j an enterprise electricity consumption behavior evaluation value.
Preferably, in step 4, a reference enterprise is introduced, and the evaluation value output by the enterprise to be analyzed by the enterprise electricity behavior analysis model is regulated to 0-100 evaluation score, which is specifically as follows:
step 4.1, obtaining evaluation values of all enterprises in the training data set by adopting an enterprise electricity behavior analysis model, sorting all enterprises in the training data set in ascending order according to the evaluation values, wherein the first enterprise definition evaluation is divided into 0 points, the last enterprise definition evaluation is divided into 100 points, and sampling 100 enterprises as reference enterprises b according to average intervals 1 -b 100 The model index feature set is A= { a 1 ,a 2 ,……a 100 And define its electricity behavior evaluation score S 1 -S 100 Sequentially {1,2, … …,100};
step 4.2, assuming that the model index feature vector of the enterprise to be evaluated is a;
firstly, sequentially combining a model index feature vector a of an enterprise to be evaluated with each reference enterprise b in a feature set A of the reference enterprise i Vector a of (2) i Forming a pair of enterprise electricity consumption behavior analysis models input into the step 3.4 to obtain corresponding electricity consumption behavior evaluation values s and s i Based on this, a pair of adjacent reference enterprises b is found i And b i+1 So that s is at s i Sum s i+1 Obtaining an interval where the evaluation score of the enterprise to be evaluated is located;
obtaining weights according to the evaluation values, and obtaining a reference enterprise b i And b i+1 Evaluation score S of (2) i And S is i+1 Weighting to obtain the electricity consumption behavior evaluation score S of the enterprise to be evaluated:
S＝S i+1 *(s-s i )/(s i+1 -s i )+S i *(s i+1 -s)/(s i+1 -s i )。
the invention also provides an enterprise electricity behavior analysis system based on the multi-layer perceptron and the ordering network, which comprises:
the model index feature vector construction and index scoring module is used for constructing an electricity utilization feature index set to obtain model index feature vectors and scoring electricity utilization behaviors of the indexes;
the training data set construction module is used for constructing a training data set, and comprises a feature vector set formed by model index feature vectors of a plurality of enterprises and supervision information which is obtained based on the power consumption behavior scores of the indexes and represents the preliminary evaluation results of the power consumption behaviors of different enterprises;
the enterprise electricity behavior analysis model training module is used for respectively carrying out multi-layer perceptron processing on model index feature vectors of a plurality of enterprises in the training data set to obtain corresponding model index feature coding sets, and training an enterprise electricity behavior analysis model based on a sequencing network by adopting the model index feature coding sets and supervision information;
and the enterprise electricity consumption behavior analysis module is used for analyzing the electricity consumption behavior of the enterprise according to the enterprise electricity consumption behavior analysis model and the reference enterprise screened from the training data set.
The invention has the beneficial effects that compared with the prior art:
the method extracts the enterprise electricity utilization characteristics through a data-based classification method, and converts the data requirements on the enterprise electricity utilization behavior from quantitative to qualitative, so that the expandability of the model is greatly improved; secondly, by utilizing the model index feature vector and the supervision information, an enterprise electricity behavior analysis model is constructed and trained by a multi-layer perceptron and sequencing network method, and compared with methods such as logistic regression, numerical regression, linear regression and the like, the method has the advantages that:
the invention is based on the enterprise electricity behavior analysis model of the multi-layer perceptron and the ordering network, the step 1 index grading only needs to help an expert to primarily evaluate the electricity behaviors among different enterprises to obtain the supervision information, the specific index grading is not used for model training, such as 400 minutes of a first enterprise and 250 minutes of a second enterprise, the index grading is not used for model training, the model training only uses the label information with 400 minutes more than 250 minutes and the model index feature vector (the model index feature vector comprises a growth rate u, a fluctuation rate v, a difference degree w and the like and does not comprise a score), so the index grading can not be very accurate, and the primary judgment is more accurate as long as the difference of the indexes is larger. The accurate quantitative electricity consumption behavior evaluation score of the given training sample is not needed, and only qualitative evaluation of the electricity consumption behavior of each pair of enterprises in the training data set is needed, so that the cost of data marking is greatly reduced.
The enterprise electricity consumption behavior analysis model can overcome the defect of strong subjectivity of the traditional expert experience model, can better utilize multi-dimensional large electric power data, ensures the accuracy of enterprise electricity consumption characteristic extraction, and overcomes the subjectivity and inaccuracy of the expert manual evaluation method.
Drawings
FIG. 1 is a flow chart of an enterprise electricity behavior analysis method based on a multi-layer perceptron and a ranking network of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for more clearly illustrating the technical aspects of the present invention, and are not intended to limit the scope of the present invention.
As shown in fig. 1, the present invention provides a method for analyzing power consumption behavior of an enterprise based on a multi-layer perceptron and a ranking network, and in a further preferred but non-limiting embodiment of the present invention, the method comprises the following steps 1-4:
step 1, constructing a power utilization characteristic index set to obtain a model index characteristic vector, and grading power utilization behaviors of the indexes;
further preferably, a power utilization characteristic index set comprising power utilization load level indexes, power utilization standard indexes and power utilization interaction capacity indexes of an enterprise is constructed, and all indexes in the power utilization characteristic index set are spliced in sequence to obtain model index characteristic vectors; and scoring the electricity consumption behavior of each index according to expert experience to construct supervision information. The method comprises the following steps:
1) Extracting power consumption load level indexes of enterprises, including power consumption increase rate indexes and power consumption stability indexes of the enterprises;
the electricity consumption increase rate index comprises an increase rate index of the electricity consumption of the enterprise of 12 months to the same rate of 12 months, and the electricity consumption behavior score calculation method comprises the following steps:
the power consumption of the enterprise is about 12 months higher than the power consumption of the enterprise by 12 months at the same rate of increase=u,
u >0.8, score: 100 minutes;
u= [0.4,0.8], score: 50 minutes;
u= (0,0.4), score: 30 minutes;
the electricity consumption stability index comprises a level index of electric power consumption of the enterprise in the local peer for 12 months and an electric power difference index of the enterprise for 12 months, and the electricity consumption behavior scoring calculation method comprises the following steps:
the electric wave power of the enterprise for 12 months=v, the electric quantity difference of the enterprise for 12 months=w,
v <0.2, score: 100 minutes;
v= [0.2,0.6], score: 60 minutes;
v >0.6, score 0;
w <0.2, score 100;
w= [0.2,0.7], score 60 points;
w >0.7, score 0;
2) Extracting power utilization specification indexes of enterprises, including power stealing behavior indexes and default power utilization behavior indexes;
the electricity larceny index comprises an index of the number of times of electricity larceny in the last 24 months and an index of the total cost of electricity larceny in the last 24 months, and the electricity consumption behavior score calculation method comprises the following steps:
number of electricity larceny in the last 24 months=q, total cost of electricity larceny in the last 24 months=rten thousand yuan,
q <2, score: 100 minutes;
q= [2,10], score; dividing into 40;
q >10, score: 0 minutes;
r <1, score: 100 minutes;
r= [1,10], score: dividing into 40;
r >10, score: 0 minutes;
the electricity consumption performance index comprises an electricity consumption frequency index of the infraction of the last 24 months and a total electricity consumption cost index of the infraction of the last 24 months, and the electricity consumption performance scoring calculation method comprises the following steps:
the number of times of the illegal use of electricity of nearly 24 months=s, the total cost of the illegal use of electricity of nearly 24 months=t,
s <2, score: 100 minutes;
s= [2,10], score; dividing into 40;
s >10, score: 0 minutes;
t <1, score: 100 minutes;
t= [1,10], score: dividing into 40;
t >10, score: 0 minutes.
3) Extracting power consumption interaction capability indexes of enterprises, including capability indexes of equipment expansion by interaction of the enterprises and operators;
the capacity indexes of the enterprise and the operator interaction for equipment capacity expansion comprise an accumulated capacity expansion frequency index of nearly 24 months and an accumulated capacity reduction frequency index of nearly 24 months, and the electricity consumption behavior scoring calculation method comprises the following steps:
the capacity increasing times z1 are accumulated for about 24 months, the capacity decreasing times z2 are accumulated for about 24 months,
z1>8, score: 100 minutes;
z1= [4,8], score: 70 minutes;
z1<4, score: 20 minutes;
z2<4, score: 100 minutes;
z2= [4,10], score: 70 minutes;
z2>10, score: 0 minutes.
The extracted enterprise electricity load level, electricity behavior mode and electricity interaction capability indexes are spliced according to the index sequence to form a model index feature vector; (i.e., the model index feature vector contains the same ratio of growth rate u, fluctuation rate v, difference degree w, etc., and does not contain index scores)
The obtained scores are used for subsequent construction of supervision information.
The above is a fixed flow in the neural network, and a series of steps such as weights are implemented in the multi-layer perceptron below.
Step 2, a training data set is constructed, wherein the training data set comprises a feature vector set formed by model index feature vectors of a plurality of enterprises and monitoring information which is obtained based on the power consumption behavior scores of the indexes and represents preliminary evaluation results of power consumption behaviors among different enterprises, and the training data set specifically comprises the following steps:
step 2.1, constructing a feature vector set of the training data set: taking the model index feature vector of each enterprise as the feature vector of the training data set, and collecting the model index feature vectors of a plurality of enterprises to form a feature vector set;
step 2.2, constructing supervision information of the training data set: pairing and sampling are carried out on enterprises in the training data set, each pair of enterprises is endowed with a label representing the preliminary evaluation result of the power consumption behavior among the enterprises, and the labels representing the preliminary evaluation result of the power consumption behavior among different enterprises form supervision information;
the value mode of the label is as follows:
accumulating the scores of the power consumption behaviors of the enterprises obtained by the scores of the power consumption behaviors of the indexes in the step 1, and using the scores for primarily evaluating the power consumption behaviors of different enterprises;
comparing the primary evaluation scores of the first enterprise and the second enterprise in each pair of enterprises, and if the first enterprise is higher than the second enterprise by a set value, such as 150 points, marking the label as 1, so that the electricity utilization performance of the first enterprise is better; if the second enterprise is 150 points higher than the first enterprise, the label is marked as-1, which indicates that the second enterprise has better electricity utilization performance; otherwise, the tag is marked with 0, indicating that the first enterprise and the second enterprise behave identically in terms of electricity.
I.e., if the preliminary evaluation score difference is less than 150 points, it is marked as 0.
Step 3, respectively carrying out multi-layer perceptron processing on model index feature vectors of a plurality of enterprises in the training data set to obtain corresponding model index feature coding sets, and training an enterprise electricity consumption behavior analysis model based on the ordering network by adopting the model index feature coding sets and the supervision information;
the method specifically comprises the following steps:
step 3.1, respectively performing multi-layer perceptron processing on model index feature vectors a of a plurality of enterprises in the training data set to obtain a model index feature coding set b:
b＝mlp(a,w)
wherein w is the characteristic weight set of each index in the parameter a of the multilayer perceptron mlp;
the model index feature vector a of each enterprise correspondingly outputs a model index feature code set b. And b is the input of the enterprise electricity behavior analysis model, and finally the enterprise electricity behavior evaluation value s is obtained, and the loss function value C can be obtained from the s.
Step 3.1 is an automated process, and inputting a, i.e. automatically learning output b, is common knowledge in the art.
Step 3.1.1, constructing a plurality of neurons as input layers of a multi-layer perceptron mlp, and sequentially inputting model index feature vectors a of enterprises in a training data set into each neuron; the number of neurons is set by people, and the specific number is adjusted according to the effect of the model and is irrelevant to the number of enterprises in the training data set and the number of index features in the model index feature vector a.
Step 3.1.2, constructing a plurality of hidden layers, encoding an input feature vector, namely firstly multiplying the input feature vector by a matrix to obtain a new vector as an encoding result, wherein the matrix value is a part of the parameter weight of the multi-layer perceptron, the matrix value can be obtained through training and learning, each bit of the new vector corresponds to a neuron, and determining whether the current neuron outputs a result or not by using a Tanh function as an activation function for each neuron;
in order to reduce the calculation cost of feature coding, a Tanh function is used as an activation function, and the formula is shown as follows;
where x is the value of the vector bit corresponding to the neuron and e is a natural constant.
Step 3.1.3, constructing a plurality of neurons as an output layer of the multi-layer perceptron, and sequentially outputting characteristic coding results of the characteristic vectors, wherein the obtained characteristic coding results are a high-dimensional vector b which is used as a characteristic transformation result of the characteristic vectors of the power utilization behavior model indexes, and the formula is as follows:
b＝mlp(a,w)
wherein w is the parameter weight of the multi-layer perceptron mlp;
the coding means in the present invention: carrying out weight assignment on each index in the vector;
step 3.2, constructing a sorting network model, and constructing a loss function based on the sorting network model, the supervision information and the symmetry of the sorting relation;
1) The ordered network model is constructed to be expressed in terms of the following formula,
wherein:
P ij the probability that the ordering network model considers that the electricity utilization behavior of the enterprise i is better than that of the enterprise j is represented;
the power consumption behavior evaluation of the assumption enterprise i is better than that of the assumption enterprise j;
sigma is a super-parameter constant;
s i ＝s(b i ) The enterprise electricity consumption behavior analysis model is represented to give enterprise electricity consumption behavior evaluation values of enterprise i;
s j ＝s(b j ) And the enterprise electricity consumption behavior analysis model is used for giving enterprise j an enterprise electricity consumption behavior evaluation value.
2) Based on the ordering network model, the supervision information and the ordering relation symmetry, a loss function is constructed, specifically as follows:
the cross entropy loss function is constructed, expressed in the following formula,
simplifying the formula to obtain:
wherein:
C ij values representing the loss functions of enterprise i and enterprise j;
S ij belonging to [ -1,0,1]The relative good-bad relationship of the enterprise electricity consumption behavior between the enterprise i and the enterprise j obtained according to the supervision information is represented;
defining a pairing set I according to the symmetry of the ordering relation, namely the symmetry of the relatively good relation between the first enterprise and the second enterprise, wherein the first enterprise is required to have better electricity consumption behavior than the second enterprise;
the loss function C of set I is defined as follows:
wherein s is satisfied by i >s j 。
Step 3.3, constructing a random gradient descent optimization model based on the loss function so as to learn weights in the model index feature coding set;
a random gradient descent optimization model is constructed based on the loss function, expressed by the following formula,
wherein: w (w) k The kth weight of the feature weight set w in the model index feature code set b is represented;
η represents a learning rate;
λ i is a coefficient for enterprise i;
the value of the loss function C can be calculated by step 3.2 using the feature encoding result b.
And 3.4, constructing and training an enterprise electricity behavior analysis model by adopting a random gradient descent optimization method according to the model index feature coding set and the supervision information.
An enterprise electricity behavior analysis model is constructed based on a random gradient descent optimization model, and is expressed by the following formula,
s i ,s j ＝f(a i ,a j ；w) (4)
wherein: w is a characteristic weight set in the model index characteristic coding set b obtained in the step 3.3;
f(a i ,a j the method comprises the steps of carrying out a first treatment on the surface of the w) represents a model index feature vector a according to the electricity behavior of enterprise i and enterprise j i ,a j Outputting a function of the evaluation value;
s i ＝s(b i ) The enterprise electricity behavior analysis model is used for giving enterprise i an enterprise electricity behavior evaluation value, and the model is input as a vector b i The characteristic coding result of the enterprise i;
s j ＝s(b j ) The enterprise electricity behavior analysis model is used for giving enterprise j an enterprise electricity behavior evaluation value, and the model is input as a vector b j And (5) encoding a result for the characteristics of the enterprise j.
And 4, analyzing the electricity consumption behavior of the enterprise according to the enterprise electricity consumption behavior analysis model and the reference enterprise screened from the training data set.
In the invention, the interpretation is considered when the actual enterprise electricity behavior analysis is performed, the standard score of the percentile is used, and the whole real number range is output by using the enterprise electricity behavior analysis model and cannot be directly fitted to the standard score, so that the evaluation value of the output of the regular neural network reaches 0-100 through the step 4. In view of this, if an electricity consumption behavior evaluation score of 0 to 100 is to be obtained, a reference enterprise must be used, and this requirement cannot be satisfied only by comparing the electricity consumption behaviors of two enterprises.
In step 4, introducing a reference enterprise, and arranging the evaluation value output by the enterprise to be analyzed by the enterprise electricity behavior analysis model to be 0-100 evaluation score, wherein the evaluation score is specifically as follows:
step 4.1, obtaining an evaluation value of each enterprise in the training data set by adopting an enterprise electricity behavior analysis model, and training according to the evaluation valuePerforming ascending sorting on all enterprises in the training data set, wherein the first enterprise definition evaluation is divided into 0 score, the last enterprise definition evaluation is divided into 100 scores, and sampling 100 enterprises as reference enterprises b according to average intervals 1 -b 100 The model index feature set is A= { a 1 ,a 2 ,……a 100 And define its electricity behavior evaluation score S 1 -S 100 Sequentially {1,2, … …,100};
step 4.2, assuming that the model index feature vector of the enterprise to be evaluated is a;
firstly, sequentially combining a model index feature vector a of an enterprise to be evaluated with each reference enterprise b in a feature set A of the reference enterprise i Vector a of (2) i Forming a pair of enterprise electricity consumption behavior analysis models input into the step 3.4 to obtain corresponding electricity consumption behavior evaluation values s and s i Namely substituting the enterprise to be evaluated and each reference enterprise into the model to run once for 100 times to obtain s and s 1 -s 100 Based on this, a pair of adjacent reference enterprises b is found i And b i+1 So that s is at s i Sum s i+1 Obtaining an interval where the evaluation score of the enterprise to be evaluated is located;
obtaining weights according to the evaluation values, and obtaining a reference enterprise b i And b i+1 Evaluation score S of (2) i And S is i+1 Weighting to obtain the electricity consumption behavior evaluation score S of the enterprise to be evaluated:
S＝S i+1 *(s-s i )/(s i+1 -s i )+S i *(s i+1 -s)/(s i+1 -s i )。
assuming that i=1, the electricity consumption behavior evaluation value of the reference enterprise 1 is-20.0, the evaluation is 1, the evaluation value of the reference enterprise 2 is 10.0, the evaluation is 2, the evaluation value of the enterprise to be evaluated is-5.0, and the evaluation score of the enterprise to be evaluated is between 1 and 2;
the evaluation of the power consumption behavior of the enterprise to be evaluated is divided into a weighted combination of evaluation scores of a reference enterprise 1 and a reference enterprise 2, and the weight of the weighted combination is determined according to the evaluation value of the power consumption behavior of the enterprise, specifically: 2 (-5.0- (-20.0))/(10.0- (-20.0)) +1 (10.0- (-5.0))/(10.0- (-20.0))=1.5.
The invention relates to an enterprise electricity behavior analysis system based on a multi-layer perceptron and a sequencing network, which runs the enterprise electricity behavior analysis method based on the multi-layer perceptron and the sequencing network, and comprises the following steps:
the model index feature vector construction and index scoring module is used for constructing an electricity utilization feature index set to obtain model index feature vectors and scoring electricity utilization behaviors of the indexes;
the training data set construction module is used for constructing a training data set, and comprises a feature vector set formed by model index feature vectors of a plurality of enterprises and supervision information which is obtained based on the power consumption behavior scores of the indexes and represents the preliminary evaluation results of the power consumption behaviors of different enterprises;
the enterprise electricity behavior analysis model training module is used for respectively carrying out multi-layer perceptron processing on model index feature vectors of a plurality of enterprises in the training data set to obtain corresponding model index feature coding sets, and training an enterprise electricity behavior analysis model based on a sequencing network by adopting the model index feature coding sets and supervision information;
and the enterprise electricity consumption behavior analysis module is used for analyzing the electricity consumption behavior of the enterprise according to the enterprise electricity consumption behavior analysis model and the reference enterprise screened from the training data set.
The invention has the beneficial effects that compared with the prior art:
according to the method, the ordering method is adopted to convert the non-quantitative dimension of the electric power data into the qualitative problem, and the accuracy of marked data is obviously improved. Aiming at the problems that the multi-dimension of the electric power data is not well utilized and the weak supervision learning is difficult to realize, the patent adopts a multi-layer perceptron to process the electric power data so as to meet the problems; the invention adopts the multi-layer perceptron and the ordering network to construct the enterprise electricity behavior analysis model, realizes the automatic enterprise electricity behavior analysis, greatly reduces the cost compared with manual expert evaluation, has good expansibility, and can better provide paid electric power data products and services such as financial wind control, client recommendation, data sharing and the like for financial institutions.
The method extracts the enterprise electricity utilization characteristics through a data-based classification method, and converts the data requirements on the enterprise electricity utilization behavior from quantitative to qualitative, so that the expandability of the model is greatly improved; secondly, by utilizing the model index feature vector and the supervision information, an enterprise electricity behavior analysis model is constructed and trained by a multi-layer perceptron and sequencing network method, and compared with methods such as logistic regression, numerical regression, linear regression and the like, the method has the advantages that:
the invention is based on the enterprise electricity behavior analysis model of the multi-layer perceptron and the ordering network, the step 1 index grading only needs to help an expert to primarily evaluate the electricity behaviors among different enterprises to obtain the supervision information, the specific index grading is not used for model training, such as 400 minutes of a first enterprise and 250 minutes of a second enterprise, the index grading is not used for model training, the model training only uses the label information with 400 minutes more than 250 minutes and the model index feature vector (the model index feature vector comprises a growth rate u, a fluctuation rate v, a difference degree w and the like and does not comprise a score), so the index grading can not be very accurate, and the primary judgment is more accurate as long as the difference of the indexes is larger. The accurate quantitative electricity consumption behavior evaluation score of the given training sample is not needed, and only qualitative evaluation of the electricity consumption behavior of each pair of enterprises in the training data set is needed, so that the cost of data marking is greatly reduced.
The enterprise electricity consumption behavior analysis model can overcome the defect of strong subjectivity of the traditional expert experience model, can better utilize multi-dimensional large electric power data, ensures the accuracy of enterprise electricity consumption characteristic extraction, and overcomes the subjectivity and inaccuracy of the expert manual evaluation method.
While the applicant has described and illustrated the embodiments of the present invention in detail with reference to the drawings, it should be understood by those skilled in the art that the above embodiments are only preferred embodiments of the present invention, and the detailed description is only for the purpose of helping the reader to better understand the spirit of the present invention, and not to limit the scope of the present invention, but any improvements or modifications based on the spirit of the present invention should fall within the scope of the present invention.
Claims (4)
1. The enterprise electricity consumption behavior analysis method based on the multi-layer perceptron and the ordering network is characterized by comprising the following steps of:
the method comprises the following steps:
step 1, constructing a power utilization characteristic index set to obtain a model index characteristic vector, and grading power utilization behaviors of the indexes;
step 2, a training data set is constructed, wherein the training data set comprises a feature vector set formed by model index feature vectors of a plurality of enterprises and monitoring information which is obtained based on the power consumption behavior scores of the indexes and represents preliminary evaluation results of power consumption behaviors among different enterprises; the method specifically comprises the following steps:
step 2.1, constructing a feature vector set of the training data set: taking the model index feature vector of each enterprise as the feature vector of the training data set, and collecting the model index feature vectors of a plurality of enterprises to form a feature vector set;
step 2.2, constructing supervision information of the training data set: pairing and sampling are carried out on enterprises in the training data set, each pair of enterprises is endowed with a label representing the preliminary evaluation result of the power consumption behavior among the enterprises, and the labels representing the preliminary evaluation result of the power consumption behavior among different enterprises form supervision information;
the value mode of the label is as follows:
accumulating the scores of the power consumption behaviors of the enterprises obtained by the scores of the power consumption behaviors of the indexes in the step 1, and using the scores for primarily evaluating the power consumption behaviors of different enterprises;
comparing the primary evaluation scores of the first enterprise and the second enterprise in each pair of enterprises, and if the first enterprise is higher than the second enterprise by a set score, marking the label as 1 to indicate that the electricity utilization performance of the first enterprise is better; if the second enterprise is higher than the first enterprise by a set score, the label is marked as-1, which indicates that the second enterprise has better electricity utilization performance; otherwise, the tag is marked as 0, which indicates that the electricity utilization behaviors of the first enterprise and the second enterprise are the same;
step 3, respectively carrying out multi-layer perceptron processing on model index feature vectors of a plurality of enterprises in the training data set to obtain corresponding model index feature coding sets, and training an enterprise electricity consumption behavior analysis model based on the ordering network by adopting the model index feature coding sets and the supervision information; the method specifically comprises the following steps:
step 3.1, respectively performing multi-layer perceptron processing on model index feature vectors a of a plurality of enterprises in the training data set to obtain a model index feature coding set b:
b＝mlp(a,w)
wherein w is the characteristic weight set of each index in the parameter a of the multilayer perceptron mlp;
step 3.2, constructing a sorting network model, and constructing a loss function based on the sorting network model, the supervision information and the symmetry of the sorting relation;
the ordered network model is constructed to be expressed in terms of the following formula,
wherein:
P ij the probability that the ordering network model considers that the electricity utilization behavior of the enterprise i is better than that of the enterprise j is represented;
the power consumption behavior evaluation of the assumption enterprise i is better than that of the assumption enterprise j;
sigma is a super-parameter constant;
s i ＝s(b i ) The enterprise electricity consumption behavior analysis model is represented to give enterprise electricity consumption behavior evaluation values of enterprise i;
s j ＝s(b j ) The enterprise electricity consumption behavior analysis model is represented to give enterprise electricity consumption behavior evaluation values of enterprise j;
based on the ordering network model, the supervision information and the ordering relation symmetry, a loss function is constructed, specifically as follows:
the cross entropy loss function is constructed, expressed in the following formula,
and (3) formulating and simplifying the cross entropy loss function to obtain:
wherein:
C ij values representing the loss functions of enterprise i and enterprise j;
S ij belonging to [ -1,0,1]The relative good-bad relationship of the enterprise electricity consumption behavior between the enterprise i and the enterprise j obtained according to the supervision information is represented;
defining a pairing set I according to the symmetry of the ordering relation, namely the symmetry of the relatively good relation between the first enterprise and the second enterprise, wherein the first enterprise is required to have better electricity consumption behavior than the second enterprise;
the loss function C of set I is defined as follows:
wherein s is satisfied by i >s j ；
Step 3.3, constructing a random gradient descent optimization model based on the loss function so as to learn weights in the model index feature coding set;
a random gradient descent optimization model is constructed based on the loss function, expressed by the following formula,
wherein: w (w) k Representation model fingerThe kth weight of the feature weight set w in the mark feature code set b;
η represents a learning rate;
λ i is a coefficient for enterprise i;
step 3.4, constructing and training an enterprise electricity behavior analysis model by adopting a random gradient descent optimization method according to the model index feature code set and the supervision information;
an enterprise electricity behavior analysis model is constructed based on a random gradient descent optimization model, and is expressed by the following formula,
s i ,s j ＝f(a i ,a j ；w) (4)
wherein: w is a characteristic weight set in the model index characteristic coding set b obtained in the step 3.3;
f(a i ,a j the method comprises the steps of carrying out a first treatment on the surface of the w) represents a model index feature vector a according to the electricity behavior of enterprise i and enterprise j i ,a j Outputting a function of the evaluation value;
s i ＝s(b i ) The enterprise electricity consumption behavior analysis model is represented to give enterprise electricity consumption behavior evaluation values of enterprise i;
s j ＝s(b j ) The enterprise electricity consumption behavior analysis model is represented to give enterprise electricity consumption behavior evaluation values of enterprise j;
and 4, analyzing the electricity consumption behavior of the enterprise according to the enterprise electricity consumption behavior analysis model and the reference enterprise screened from the training data set.
2. The enterprise electricity consumption behavior analysis method based on the multi-layer perceptron and the sorting network of claim 1, wherein:
in the step 1, constructing a power utilization characteristic index set comprising power utilization load level indexes, power utilization standard indexes and power utilization interaction capacity indexes of an enterprise, and sequentially splicing all indexes in the power utilization characteristic index set to obtain a model index characteristic vector; and scoring the electricity consumption behavior of each index according to expert experience to construct supervision information.
3. The enterprise electricity consumption behavior analysis method based on the multi-layer perceptron and the sorting network of claim 1, wherein:
in step 4, introducing a reference enterprise, and arranging the evaluation value output by the enterprise to be analyzed by the enterprise electricity behavior analysis model to be 0-100 evaluation score, wherein the evaluation score is specifically as follows:
step 4.1, obtaining evaluation values of all enterprises in the training data set by adopting an enterprise electricity behavior analysis model, sorting all enterprises in the training data set in ascending order according to the evaluation values, wherein the first enterprise definition evaluation is divided into 0 points, the last enterprise definition evaluation is divided into 100 points, and sampling 100 enterprises as reference enterprises b according to average intervals 1 -b 100 The model index feature set is A= { a 1 ,a 2 ,……a 100 And define its electricity behavior evaluation score S 1 -S 100 Sequentially {1,2, … …,100};
step 4.2, assuming that the model index feature vector of the enterprise to be evaluated is a;
firstly, sequentially combining a model index feature vector a of an enterprise to be evaluated with each reference enterprise b in a feature set A of the reference enterprise i Vector a of (2) i A pair of power utilization behavior evaluation values s and s are obtained by inputting the power utilization behavior evaluation values into an enterprise power utilization behavior analysis model i Based on this, a pair of adjacent reference enterprises b is found i And b i+1 So that s is at s i Sum s i+1 Obtaining weights according to the evaluation values, and obtaining a reference enterprise b i And b i+1 Evaluation score S of (2) i And S is i+1 Weighting to obtain the electricity consumption behavior evaluation score S of the enterprise to be evaluated:
S＝S i+1 *(s-s i )/(s i+1 -s i )+S i *(s i+1 -s)/(s i+1 -s i )。
4. an enterprise electricity behavior analysis system based on a multi-layer perceptron and a sorting network, which operates the enterprise electricity behavior analysis method based on the multi-layer perceptron and the sorting network as set forth in any one of claims 1 to 3, characterized in that:
the system comprises:
the model index feature vector construction and index scoring module is used for constructing an electricity utilization feature index set to obtain model index feature vectors and scoring electricity utilization behaviors of the indexes;
the training data set construction module is used for constructing a training data set, and comprises a feature vector set formed by model index feature vectors of a plurality of enterprises and supervision information which is obtained based on the power consumption behavior scores of the indexes and represents the preliminary evaluation results of the power consumption behaviors of different enterprises;
the enterprise electricity behavior analysis model training module is used for respectively carrying out multi-layer perceptron processing on model index feature vectors of a plurality of enterprises in the training data set to obtain corresponding model index feature coding sets, and training an enterprise electricity behavior analysis model based on a sequencing network by adopting the model index feature coding sets and supervision information;
and the enterprise electricity consumption behavior analysis module is used for analyzing the electricity consumption behavior of the enterprise according to the enterprise electricity consumption behavior analysis model and the reference enterprise screened from the training data set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210737510.1A CN115310999B (en) | 2022-06-27 | 2022-06-27 | Enterprise electricity behavior analysis method and system based on multi-layer perceptron and sequencing network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210737510.1A CN115310999B (en) | 2022-06-27 | 2022-06-27 | Enterprise electricity behavior analysis method and system based on multi-layer perceptron and sequencing network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115310999A CN115310999A (en) | 2022-11-08 |
| CN115310999B true CN115310999B (en) | 2024-02-02 |
Family
ID=83856032
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210737510.1A Active CN115310999B (en) | 2022-06-27 | 2022-06-27 | Enterprise electricity behavior analysis method and system based on multi-layer perceptron and sequencing network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115310999B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020000399A1 (en) * | 2018-06-29 | 2020-01-02 | 东莞理工学院 | Multi-agent deep reinforcement learning proxy method based on intelligent grid |
| CN112633421A (en) * | 2021-03-09 | 2021-04-09 | 国网江西综合能源服务有限公司 | Method and device for detecting abnormal electricity utilization behavior of user |
| CN113988459A (en) * | 2021-11-11 | 2022-01-28 | 国网江苏省电力有限公司苏州供电分公司 | Small and medium-sized enterprise growth assessment method and system based on electric power marketing data |
| CN114004530A (en) * | 2021-11-11 | 2022-02-01 | 国网江苏省电力有限公司苏州供电分公司 | Enterprise power credit score modeling method and system based on sequencing support vector machine |
| CN114219606A (en) * | 2021-11-11 | 2022-03-22 | 国网四川省电力公司信息通信公司 | Power data-based power consumption enterprise credit evaluation method and system |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11250314B2 (en) * | 2017-10-27 | 2022-02-15 | Cognizant Technology Solutions U.S. Corporation | Beyond shared hierarchies: deep multitask learning through soft layer ordering |
| US11232344B2 (en) * | 2017-10-31 | 2022-01-25 | General Electric Company | Multi-task feature selection neural networks |
| US20190378017A1 (en) * | 2018-06-12 | 2019-12-12 | Sun-Yuan Kung | System and method for implementing a neural network |
| US11537954B2 (en) * | 2018-09-04 | 2022-12-27 | Beijing Didi Infinity Technology And Development Co., Ltd. | System and method for ride order dispatching and vehicle repositioning |
| US20210248450A1 (en) * | 2020-02-07 | 2021-08-12 | Google Llc | Sorting attention neural networks |
-
2022
- 2022-06-27 CN CN202210737510.1A patent/CN115310999B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020000399A1 (en) * | 2018-06-29 | 2020-01-02 | 东莞理工学院 | Multi-agent deep reinforcement learning proxy method based on intelligent grid |
| CN112633421A (en) * | 2021-03-09 | 2021-04-09 | 国网江西综合能源服务有限公司 | Method and device for detecting abnormal electricity utilization behavior of user |
| CN113988459A (en) * | 2021-11-11 | 2022-01-28 | 国网江苏省电力有限公司苏州供电分公司 | Small and medium-sized enterprise growth assessment method and system based on electric power marketing data |
| CN114004530A (en) * | 2021-11-11 | 2022-02-01 | 国网江苏省电力有限公司苏州供电分公司 | Enterprise power credit score modeling method and system based on sequencing support vector machine |
| CN114219606A (en) * | 2021-11-11 | 2022-03-22 | 国网四川省电力公司信息通信公司 | Power data-based power consumption enterprise credit evaluation method and system |
Non-Patent Citations (2)
| Title |
|---|
| 售电市场环境下基于数据驱动的用户用电行为分析;赵阳;胡诗尧;杨书强;冯成;王毅;;电力需求侧管理(第04期);第45-50页 * |
| 基于多模态知识感知注意力机制的问答方法;张莹莹;钱胜胜;方全;徐常胜;;计算机研究与发展(第05期);第1037-1045页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115310999A (en) | 2022-11-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111414942B (en) | Remote sensing image classification method based on active learning and convolutional neural network | |
| CN108921604B (en) | Advertisement click rate prediction method based on cost-sensitive classifier integration | |
| CN110807760B (en) | Tobacco leaf grading method and system | |
| CN109492748B (en) | Method for establishing medium-and-long-term load prediction model of power system based on convolutional neural network | |
| CN111079941B (en) | Credit information processing method, credit information processing system, terminal and storage medium | |
| CN111583031A (en) | Application scoring card model building method based on ensemble learning | |
| CN112925908A (en) | Attention-based text classification method and system for graph Attention network | |
| CN116109195B (en) | Performance evaluation method and system based on graph convolution neural network | |
| CN112613536A (en) | Near infrared spectrum diesel grade identification method based on SMOTE and deep learning | |
| CN115470962A (en) | LightGBM-based enterprise confidence loss risk prediction model construction method | |
| CN111681022A (en) | Network platform data resource value evaluation method | |
| CN117151870B (en) | Portrait behavior analysis method and system based on guest group | |
| CN114764682B (en) | Rice safety risk assessment method based on multi-machine learning algorithm fusion | |
| CN115795351A (en) | Elevator big data risk early warning method based on residual error network and 2D feature representation | |
| CN113591947A (en) | Power data clustering method and device based on power consumption behaviors and storage medium | |
| CN115310999B (en) | Enterprise electricity behavior analysis method and system based on multi-layer perceptron and sequencing network | |
| CN113361202A (en) | Distribution transformer heavy overload rapid early warning method based on characteristic load prediction | |
| CN112686498A (en) | Enterprise credit rating method based on deep convolutional network | |
| CN112163731A (en) | Special transformer user electric charge recovery risk identification method based on weighted random forest | |
| CN114004530B (en) | Enterprise electric power credit modeling method and system based on ordering support vector machine | |
| CN113837266B (en) | Software defect prediction method based on feature extraction and Stacking ensemble learning | |
| CN113205274A (en) | Quantitative ranking method for construction quality | |
| CN114091961A (en) | Power enterprise supplier evaluation method based on semi-supervised SVM | |
| CN113807671A (en) | Oil well supply and production matching degree quantitative evaluation method based on multi-source data | |
| CN113935413A (en) | Distribution network wave recording file waveform identification method based on convolutional neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |