CN110084398A - A kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data - Google Patents
A kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data Download PDFInfo
- Publication number
- CN110084398A CN110084398A CN201910198215.1A CN201910198215A CN110084398A CN 110084398 A CN110084398 A CN 110084398A CN 201910198215 A CN201910198215 A CN 201910198215A CN 110084398 A CN110084398 A CN 110084398A
- Authority
- CN
- China
- Prior art keywords
- industry
- enterprise
- electric power
- formula
- adapting detecting
- 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
- 238000000034 method Methods 0.000 title claims abstract description 44
- 230000005611 electricity Effects 0.000 claims abstract description 62
- 238000004519 manufacturing process Methods 0.000 claims abstract description 45
- 238000012546 transfer Methods 0.000 claims abstract description 8
- 238000004140 cleaning Methods 0.000 claims abstract description 6
- 230000012010 growth Effects 0.000 claims description 50
- 238000009826 distribution Methods 0.000 claims description 19
- 230000008859 change Effects 0.000 claims description 17
- 239000011159 matrix material Substances 0.000 claims description 10
- 238000012216 screening Methods 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 6
- 238000001744 unit root test Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000004458 analytical method Methods 0.000 abstract description 12
- 230000008901 benefit Effects 0.000 abstract description 4
- 230000002596 correlated effect Effects 0.000 abstract description 3
- 238000011160 research Methods 0.000 description 15
- 238000011161 development Methods 0.000 description 13
- 230000000875 corresponding effect Effects 0.000 description 8
- 230000000994 depressogenic effect Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 230000006399 behavior Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000005096 rolling process Methods 0.000 description 5
- 238000011144 upstream manufacturing Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000007704 transition Effects 0.000 description 4
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 3
- 230000014509 gene expression Effects 0.000 description 3
- 230000036541 health Effects 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 239000002131 composite material Substances 0.000 description 2
- 238000010219 correlation analysis Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 241000196324 Embryophyta Species 0.000 description 1
- 206010021703 Indifference Diseases 0.000 description 1
- 230000018199 S phase Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000000556 factor analysis Methods 0.000 description 1
- 230000008713 feedback mechanism Effects 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000008014 freezing Effects 0.000 description 1
- 238000007710 freezing Methods 0.000 description 1
- 238000000227 grinding Methods 0.000 description 1
- 230000007773 growth pattern Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000001556 precipitation Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 238000012549 training Methods 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Educational Administration (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of Industrial Cycle self-adapting detecting methods based on enterprise's electric power big data, comprising the following steps: step 1: carrying out cleaning arrangement for business electrical information;Step 2: building complex network model simultaneously chooses relevant industries;Step 3: separating different electricity consumption behavior states using hidden Markov model;Step 4: building Industrial Cycle index simultaneously carries out self-adapting detecting.Compared with prior art, the present invention has theoretical advanced, complex network model and hidden Markov model are organically combined, complex network model is by the analysis to global all variables and compares, improve the validity that correlated variables is chosen in hidden Markov model, feature dynamic associations and transfer characteristic between different industries, the result of complex network model is inputted into hidden Markov model, while the advantages that considering the influence of ambient weather, festivals or holidays factor, improving the accuracy to industry production boom condition predicting.
Description
Technical field
The present invention relates to electric power big data technical fields, more particularly, to a kind of industry scape based on enterprise's electric power big data
Gas self-adapting detecting method.
Background technique
Consumer confidence index is to reflect the quantitative target of every profession and trade operation conditions, for reflecting that the Economic Climate of industry changes shape
Condition.The movable development of Accurate Prediction Industrial Cycle exponent pair production and marcoeconomic regulation and control have great significance.In the world
The method of popular measurement Economic Climate situation is composite index number method (Composite Index), i.e., with a national industry
Growth level chooses some macrostatistics, is divided into leading indicators group, coincidence indicator group and lagging indicator as reference
Group, to construct Economic Climate Analysis index system, the turning point of analysis and prediction business cycle fluctuation and business cycle.But this
The reason of a little indexs are often monthly or season data, and very macroscopical, can not reflect Economic Climate change index behind, also without
Method reflects the variation of industry restructuring bring, less can be carried out and timely predicts in short term.
In general, GDP is the reflection most important index of national economic situation, and size often represents scape of prosperous economy
The degree of gas, prediction GDP aspect also there has been extensive research, and various prediction models are widely used, but using GDP as reflection
The index of one national prosperity degree is still more unilateral, because it can not capture influence prosperous an important factor for changing, such as
Health, individual freedom, safety guarantee etc..Many new indexs, the consumer confidence index issued such as London Legatum research institute, 2017
Economic freedom degree index, 2015 global Innovation Indexes, combine multiple dimension indexs to measure a national prosperous situation
And ranking, including economic growth, business environment, education, health, safety guarantee, personal welfare, natural environment etc., compensate for GDP
Deficiency." (2013) are reported in the city boom of the United Nations Human Settlements Programme " points out people-oriented city boom research,
Consider that many aspects such as infrastructure, justice, sustainability and quality of life analyze city boom, brings the whole world
Scholar's discusses warmly, and Habitat International periodical has a phase specially to discuss city boom, wherein plurality of articles
It is proposed new CPI index construction and measurement method and construct other city indexs of correlation, as city environmental quality health index,
Sustainability development index etc..
As the cost that machine learning and the rise of big data technology and high frequency bulk sample notebook data acquire reduces, many machines
The method of device study is applied in index construction.
In terms of the prosperous index of building reflection economic situation, more famous is " gram strong index ".Britain's " economics
People " magazine proposes a gram concept for strong index in December, 2010, comprehensive three indexs --- the whole province's volume of rail freigh, electricity consumption,
Bank loan, Lai Fanying real economy situation.The wherein significance level highest of electricity consumption speedup can be used to represent prosperous situation
And construct the new high frequency consumer confidence index that can reflect correlativity and the industrial structure between various industries.The user in Shanghai City at present
It is being gradually completing changing the outfit for intelligent electric meter, the data of the available day degree of intelligent electric meter even more high frequency, these data obtain
It is taken as day degree Economic Climate condition predicting and provides possibility.And since there are many factor for influencing mixed economy boom situation, each
The variation of electricity consumption speedup, the industry restructuring of microcosmic industry can all lead to the change of prosperous situation.It therefore can be anti-using electricity
It reflects the prosperous situation of industry and commerce production and faces a major issue, i.e., for each industry and commerce user, influence the variation of its electricity consumption
There are many factor, electricity consumption variation including weather, festivals or holidays, industrial characteristic, upstream and downstream industry etc..Therefore, using electricity consumption structure
The correlativity and the industrial structure that Economic Climate index depends between industry are built, and based on input_output relevance coefficient relationship between industry
Prediction technique is highly dependent on industry input_output relevance coefficient coefficient, due to time-lag effect is high, model error cannot be used for greatly it is short
Phase prediction.And emerging in large numbers for RF power data to study the real time correlation relationship between industry, this research is with Glan
Based on outstanding causal analysis method, industry and commerce complex network model is established, finds and excavate dominant, recessive phase between industry
Pass relationship constructs the high frequency consumer confidence index of new reflection macroeconomy situation, sets about from microcosmic industry, sufficiently excavated influence
The various different factors of Macro-economic Cycles, the influence for analyzing Industrial Structure Features and variation, are realized to macroeconomy status
Accurate description and short-term trend prediction will provide theoretical foundation for macro adjustments and controls and economic policy formulation, for investment and industry and commerce
Production development provides decision-making foundation.
Traditional industry research is based on input_output relevance coefficient relationship between industry, is highly dependent on industry input_output relevance coefficient system
Number, and industry chooses the experience that more is based on, while that there is also time-lag effects is high, model error is big, economic theory is assumed is strong, difficult
The problems such as to analyze comprehensively.
In high dimensional data, covariant or Correlative Influence Factors number are often very big, and sparsity principle thinks target
Variable, which only will receive a small number of correlated variables, to be influenced, and in view of this principle, has there is many variables for being directed to high dimensional data at present
The method of selection and screening feature, such as information criterion AIC, bayesian information criterion BIC, LASSO, Local Polynomial approximatioss
(forward, backward), principal component decomposition, factor analysis etc..These methods may can provide preferable recurrence and prediction result,
But all it is the analysis from data to data, the result of output can not be explained effectively.In actual analytic process,
Researchers can preset certain constraint conditions or important prediction becomes according to existing investigation or Heuristics
The reason of measuring, but still can not explaining the relationship and variables choice of internal system completely.
Complex network is to solve the model that complication system is whole and local relation and grows up, complex network model
The correlation and information flow process between the different elements of complicated composition system can be clearly portrayed, is grinding in recent years
Study carefully one of hot spot.Complex network is by indicating that the point of these elements is connected by the relationship (line segment with the arrow) between element
At network diagramming, a point to another point the side to be passed through item number be known as length, connect with a point other put
Number is known as the degree of the point, sums to obtain the intensity of the point to the weight on connected side.Complex Networks Theory is first by Watts etc.
(1998) proposed with Barabasi etc. [30], Duncan Watts etc. proposes " small-world network model " earliest, for describe from
Transformation of the regular network to random network, proposition " scales-free network " model such as Albert-LaszloBarabasi, degree distribution tool
There is power law form, for describing the network of many reality.Scales-free network is characterized in that a small number of nodes possess a large amount of connection, and
The only a small amount of connection of great deal of nodes, the status of each node in a network is different with effect, also different to the function of complex network.
Always there are many researchs to the property of network, generation, stability.And complex network is in the complication system of different reality in recent years
Application it is more extensive, such as electric system, public transit system, social networks, industrial system, urban energy distribution, neural network.
In addition to this there are also router or calculating in the stock research that much complex network is applied in financial system, computer system
The connection of machine, dissemination of news indicate different functional area or the research of neural cluster etc. in brain network.For between industry
Correlation research, Yao Can industry 108 months monthly electricity consumption data of medium 29 for analyzing five provinces of SOUTHERN CHINA,
The complex network model of industrial trade electricity consumption is established using granger-causality test method, finds industry electricity consumption relationship
It is typical scales-free network in power-law distribution.And the composition mechanism of industry energy-consuming is further studied, between discovery industry
Existing energy transferring structure and feedback mechanism.
Instantly, in the research of index construction and prediction model, the method based on machine learning is widely used.Certain machines
Device learning model can analyze the data of high frequency or big statistic and have good short-term forecast ability, but more for existing
There are the various states such as prosperous, depressed in the system of kind hidden state mode, such as consumer confidence index, and take under each state
From rule it is inconsistent, and each state occur time can not also observe, in some instances it may even be possible to various states superposition occur, be based on
This judgement uses hidden Markov model to be predicted to excavate the hidden state that trade power consumption amount is reflected.
Hidden Markov model is a kind of model of search time sequence, it is assumed that system there are multiple unobservable states,
It is Markov process between these states, and observation is the superposition of these states.Hidden Markov model is by original state
The distribution of probability, state-transition matrix, state is determined, provides the mathematics of its behind at first by L.E.Baum and its partner
Principle.Hidden Markov model (HMM) is frequently used to research complex dynamic systems at present, surveys in information retrieval, gene
The multiple fields successful applications such as sequence, speech recognition, natural language processing, accurate prediction.Hidden Markov model passes through many shapes
State can accurately describe problem, especially in the limited situation of data volume.Many researchs are dedicated to extending HMM model,
To provide better model and more accurately as a result, such as autoregression model, factorial HMM model, hierarchical model, Asymmetric Model.
State distribution is generalized to any Bayesian network, is desirably to obtain more accurate state space by asymmetric hidden Markov model.
Although asymmetric HMM learning model is more complicated, it is more flexible in practice, more preferable to the description of truthful data.
In conclusion complex network model and hidden Markov model have very big advantage in terms of consumer confidence index research,
The complex network between all trade power consumption amount growth rates, and benefit are established by granger-causality test and industry correlation analysis
It is given a mark with cyberrelationship to the importance of each industry in a network, and then obtains industry importance ranking and the sector
Relevant other industry and other industry give the influence degree etc. of the sector on the basis of complex network using network
Industrial structure and electricity consumption growth rate out constructs vertical hidden Markov model and is accurately predicted, last by state analysis
Obtain the production intensity consumer confidence index of industry.
Summary of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide one kind to be based on enterprise's electric power
The Industrial Cycle self-adapting detecting method of big data.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data, comprising the following steps:
Step 1: carrying out cleaning arrangement for business electrical information;
Step 2: building complex network model simultaneously chooses relevant industries;
Step 3: separating different electricity consumption behavior states using hidden Markov model;
Step 4: building Industrial Cycle index simultaneously carries out self-adapting detecting.
Further, the step 1 include it is following step by step:
Step 11: collecting business electrical amount data and be modified by interpolation method and sorted out with different industries;
Step 12: business electrical amount data being subjected to cleaning arrangement using symmetrical growth rate model.
Further, the symmetrical growth rate model in the step 12, describes formula are as follows:
In formula, dytFor symmetrical growth rate, L is lag operator, atFor residual sequence, φi,Φ,θiRespectively different is stagnant
Later period coefficient.
Further, the step 2 include it is following step by step:
Step 21: complex network model is constructed with unit root test, Granger CaFpngerusality test and Pearson correlation coefficient;
Step 22: by carrying out threshold value screening to each industry, weighting in-degree is calculated and various industries are arranged in sequence
Sequence is chosen.
Further, the receptance function of the hidden Markov model in the step 3 are as follows:
In formula, dyt(S=i) the electricity consumption growth rate for being state S=i, ci、αijAnd βikTo need the parameter taken into account, xj,t
For the variable of guide's industry, Outk,tFor out-degree strength weighted average value, εt(i) be mean value be 0, variance isNormal distribution.
Further, the transfer matrix of the hidden Markov model in the step 3, describes formula are as follows:
In formula, γijAnother shape probability of state, N are transferred to from a state for systemsFor total state number.
Further, the step 4 include it is following step by step:
Step 41: calculating and produce the related electric quantity change of prosperous situation in the industry;
Step 42: assigning different weights for the electric quantity change in the industry and construct industry production consumer confidence index;
Step 43: building major class consumer confidence index formula carries out self-adapting detecting to each major class industry.
Further, in the step 41 electric quantity change calculation formula are as follows:
In formula, uiIt (t) is the probability value at corresponding moment, dy'tFor total electricity consumption growth rate, dy "tFor electric quantity change.
Further, industry production consumer confidence index in the step 42, describes formula are as follows:
In formula, W (x) is assignment function, and PIT is the consumer confidence index in [a, b] time range.
Further, major class consumer confidence index formula in the step 43 are as follows:
In formula, PIT2For major class consumer confidence index, EleiFor the average electricity consumption in the class industry sample phase in i-th, Ele is
Average electricity consumption in the major class industry sample phase, PITiFor i-th of industry production consumer confidence index.
Compared with prior art, the invention has the following advantages that
(1) accuracy is high, and the present invention is organically combined complex network model and hidden Markov model, complex web
Network model is by the analysis to global all variables and compares, improve correlated variables in hidden Markov model choose it is effective
Property, dynamic associations and transfer characteristic between different industries are featured, the result of complex network model is inputted into hidden Ma Erke
Husband's model, while considering the influence of ambient weather, festivals or holidays factor, it improves to the accurate of industry production boom condition predicting
Property.
(2) theoretical advanced, complex network between the industry that the present invention constructs has excavated dynamic network of relation knot between industry
Structure is established hidden Markov model and is analyzed with the production boom situation to center industry, and this method is applied to own
Middle class industry, finally by the industry weight for combining complex network model output, and the tune based on hidden Markov model output
Electric power growth rate after whole, building production consumer confidence index, obtains the prosperous situation of Shanghai City various industries, to pass through to Shanghai City
Self-adapting detecting is played the role of in Ji operation.
Detailed description of the invention
Fig. 1 is overall flow schematic diagram of the invention;
Fig. 2 is the frequency statistics figure of the correlation coefficient value of the embodiment of the present invention, and wherein Fig. 2 (a) is chart of frequency distribution, Fig. 2
It (b) is density function distribution map;
Fig. 3 be the present embodiment treated industry is carried out between network relational network figure;
Upstream and downstream and mutually independent business connection figure of the Fig. 4 for the present embodiment;
Fig. 5 is the automobile component of the present embodiment and the optimal fitting result parameter estimated result figure of accessories manufacturing industry;
Fig. 6 is that the electricity of the present embodiment increases fitting and prediction waveform diagram;
Fig. 7 be the present embodiment in October, 2018 electricity growth rate and prediction waveform diagram;
Fig. 8 is that the day frequency of the present embodiment produces consumer confidence index figure;
Fig. 9 is the monthly production consumer confidence index figure of the present embodiment;
Figure 10 is to produce consumer confidence index figure in the season of the present embodiment.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiment is a part of the embodiments of the present invention, rather than whole embodiments.Based on this hair
Embodiment in bright, those of ordinary skill in the art's every other reality obtained without making creative work
Example is applied, all should belong to the scope of protection of the invention.
Embodiment
One, general description:
It is as shown in Figure 1 the overall flow schematic diagram of corresponding embodiment of the present invention, specifically includes the following steps:
(1) cleaning and arrangement of business electrical information
Using the means and method of big data analysis, the ammeter data of enterprise is screened, is cleaned, identifies abnormal number
According to, and interpolation, completion data are carried out using related algorithm to the missing of data.The middle class industry mark classified by industrial sectors of national economy
Standard summarizes the electric power data branch trade of original enterprise, and cleans again to the data after summarizing, to obtain structure
Change, analyzable high-frequency data.
(2) it constructs the complex network of different times and network is analyzed
Time series analysis is carried out using electric power growth rate of the related algorithm to industry, including the use of unit root test to sentence
It powers off whether power growth rate data are stationary sequence, differentiates between two sequences whether there is cause and effect using Granger CaFpngerusality test
Relationship differentiates the transmitting period etc. of Industry Effect with AIC information criterion, excavates the potential pass between industry by algorithm above
System identifies the upstream-downstream relationship between industry, constructs the complex network between industry on this basis.For dynamic industry structure, lead to
The window and rolling length for crossing setting certain length, the data chosen in the length of window construct current complex network model, will
Window translates corresponding rolling length backward, obtains the complex network model that data in new window construct a new phase, and repeating should
Process obtains the Research Dynamic Complex Networks between different times industry.By being analyzed network property the quality to judge network,
Such as whether meeting scales-free network feature etc., and is examined using QAP and compares the input-output table between the network and industry,
To ensure to construct the reasonability and reliability of network.
(3) it extracts network index and is predicted with hidden Markov state transition model
Under the conditions of ensuring that complex network is reliable and steady, by extracting network of relation index, such as out-degree, in-degree, phase
Close industry number, the variation of relevant industries number, the association between various industries etc., in conjunction with external data for example temperature, wind speed, precipitation,
The data such as air pressure, humidity, festivals or holidays are modeled using electricity consumption growth rate of the hidden Markov state transition model to industry,
By being trained to data and solved using EM algorithm, obtain state transition probability matrix, each time point state distribution with
And the indexs such as coefficient of each variable, it is predicted, is obtained using probability of the trained parameter to different industries future status
To the state distribution of following two months different industries.
(4) branch trade constructs the production intensity consumer confidence index based on operation power data and compares with output value speedup
In obtaining each on the basis of the hidden Markov state distribution probability of class industry, by electricity adjusted
Power increases the different weight of imparting and obtains day frequency index, and the index of day frequency is aggregated into the moon by the method using integral mean,
The monthly production intensity index of the sector is obtained, is aggregated into season using same method, obtains season production intensity index.With each row
Industry is averaged electricity consumption as weight, the production intensity index of middle class industry is weighted and averaged, the production for obtaining major class industry is strong
It spends index, and is compared with the output value speedup of the above enterprise of scale of statistics bureau statistics, due to the difference in Statistical Criteria, two
There are deviations between person, but electric power, as the important investment of one of production factors, the output value data compared to lag can more shift to an earlier date
It learns the production status of current industry, thus self-adapting detecting can be played the role of to the prosperous situation of industry production.
Two, data sources and model specification
The description of 1.1 data and pretreatment
The present embodiment freezes electricity and corresponding enterprise day using enterprise's ammeter more than 8100, Shanghai City scale
Segmented industry code, time span are on the October 31st, 1 day 1 of August in 2015, are amounted to 1034 days.It also obtains simultaneously same
The observation data of regional day frequency weather, including mean temperature, rainfall, wind speed, air pressure, humidity, due to the highest temperature and minimum gentle
In strong correlation, therefore chooses high temperature and analyzed.It is associated with, leads to further, since industry and commerce production exists with working day, festivals or holidays
Spending the lookup public holiday sets two dummy variables, respectively indicates day off, public holiday, wherein variable takes the legal vacation of 10 expressions
Day, 11 indicate that the weekend of non-public holidays, 00 indicate working day (there may be weekends to take off).
Power information matching is carried out by information such as title, addresses to more than the 8000 above enterprises of scale, is matched to
5900 enterprise's power informations amount to 8233 families number;To other 2200 enterprises, since enterprise name change, the electricity charge are paid
Information it is incomplete etc. reasons do not match associate power information.By data cleansing, exceptional value, the missing in every group of data are identified
Value, and take dependent interpolation method to obtain complete time series data, and enterprise's electricity of the same industry is summarized,
Obtain the time series data of the total electricity consumption of the sector.It will be more than five times of variances with mean bias during data screening
Value is regarded abnormal point deletion and is modified by interpolation method, and the electricity consumption data of complete 165 industries of data is finally obtained.
After business electrical amount is aggregated into industry, since trade power consumption amount absolute value is main and industrial nature, industry size
It is related, it can not directly reflect the prosperous situation of the production intensity an of industry.Symmetrical growth rate can reflect true electricity consumption
Growth pattern, with other growth rate mode indifferences, therefore, Economic Climate situation is calculated using symmetrical growth rate to be had centainly
Representative and reasonability.Since there are apparent 7 days periodicity for symmetrical growth rate, in order to remove the period, autoregression product is utilized
Moving average model arima (7,1,7) model is divided to be adjusted:
In formula, dytFor symmetrical growth rate, L is lag operator, atFor residual sequence, φi, Φ, θiRespectively different is stagnant
Later period coefficient.
According to the autocorrelation figure of residual sequence adjusted, 7 days periods were eliminated, and single order truncation, explanation
Time series data adjusted becomes weakly stationary sequence, meets the analysis requirement of prosperous status score.
Further, since electricity consumption growth rate is not only related with the development inside itself the industry, also and the industrial structure, on
Downstream industry development, external climate etc. are related, therefore, it is necessary on the basis of controlling these factors analyze electricity consumption growth rate, one
As for, bigger when electricity consumption growth rate is timing, and the sector production is prosperous a possibility that, electricity consumption growth rate is negative, industry production
Stagnant possibility is bigger.
1.2 choose relevant industries using complex network model
In order to take into account the factor of the industrial structure, industry upstream-downstream relationship, the present embodiment uses complex network model
Method analyzed.Association between industry and the cyberrelationship of evolution are described with complex network model, for dynamic
The state industrial structure sets the window and rolling length of certain length, and the data chosen in the length of window construct current complex web
Network model, and obtain data in new window by the way that data window to be translated to corresponding rolling length backward and construct answering for a new phase
Miscellaneous network model, the process of repetition obtain Research Dynamic Complex Networks result.In complex network model, directed complex networks model master
To include three component parts, be node in network, side, direction respectively.
1.2.1 network struction
Each node in network indicates an industry, the growth rate for freezing electricity average day of size of node industry
To indicate.For the prosperous status score of each industry, it is necessary to be one and be cut smoothly without the diurnal periodicity that works, seasonal periodicity
Time series data, according to the requirement of time series data weakly stationary, which must be that mean value is constant, the auto-correlation of arbitrary order
Coefficient goes to zero, and that is to say and gradually tends to a white noise sequence, therefore uses the flat of the method checking sequence of unit root test
Stability:
(1-pL)xt=εt
E [ε]=0, V (ε)=σ < ∞, Cov (ε, ε)=μ < ∞
Wherein L is lag operator, and ε t is a white noise sequence.If | ρ | 1 time series of < is stationary sequence.It utilizes
This method carries out unit root test to the time series data of all trade power consumption growth rates.
For the influence degree between various industries, we are measured using Pearson correlation coefficients.In general, Pearson's phase
Closing property can be used to calculate the similarity degree (single order is similar) of two sequences, and formula is as follows:
Wherein xt, ytThe time series data of respectively two industries, T are data length.If similarity degree rxy?
95% confidence level non-zero is then denoted as the weight (side) of line between two nodes.Even if due to being closed between industry without correlation
System, related coefficient is also likely to be the smaller value being not zero of an absolute value, is set it is therefore contemplated that related coefficient has to be larger than
Fixed threshold value just calculates that there are the associations between industry, it is possible thereby to which it is faint even without associated industry to exclude those relevances.It is logical
It crosses and correlation analysis two-by-two is carried out to all industries, the interrelated degree between all industries can be obtained.
After the correlativity of industry and industry has been determined, it is also necessary to which whether the driving relationship between determining industry two-by-two
In the presence of and influence transmit required for the time.We are judged between industry using Akaike information rule (AIC) is minimized
The issue that interactive behavior occurs, the i.e. prosperous situation of guide's industry change, which will by lag of how many phases
It is transmitted to impacted industry, formula is as follows:
Wherein, k is the number of regression variable, and n is the scale of sample,For the residuals squares of recurrence
With.
Examine driving relationship whether there is based on autoregressive Granger CaFpngerusality test model simultaneously, i.e., two-by-two industry it
Between driving relationship whether on statistical relationship at expression formula are as follows:
Wherein residual error μt, εtIt is the normal distribution that average value is 0, variance is constant, p, q are lag order, ytIndicate target
Industry, xtIndicate guide's industry, q minimizes information rule by AIC above and obtains here.Alternative hypothesis is factor betajIt is incomplete
It is zero, if it is assumed that setting up then xtIt is ytThe reason of changing is examined by F and is determined.
1.2.2 industry importance ranking and relevant industries are chosen
In a complex network, the variable of reflection network midpoint importance is the degree and intensity of point, is connect with a point
Numbers of other points be known as the degree of the point, the weight on the side of node connection is summed to obtain the intensity of the point, towards the point
Number of edges be in-degree, otherwise be out-degree.Out-degree after weighting can reflect the degree for the other industry that the node can influence, can
With the importance for being used to indicate the sector in a network and it is ranked up.
For most important industry, complex network result often simultaneously provide very high in-degree and weighting in-degree, i.e., with this
The other industry number of industry strong correlation is more, and the development of these industries more or less can all influence the development of center industry, but this
A little industries are also likely to be present very strong correlativity between each other.If all stronger industries of correlation all as shadow
Ring center trade power consumption because usually centering heart industry is predicted, then may cause Problems of Multiple Synteny, may make finally
Prediction result it is inaccurate or not significant.Therefore, it needs to remove on the basis of retaining guide's industry before being predicted more
Weight synteny, the method that the present embodiment uses be if two guide's industries simultaneously and center industry there are strong correlation relationships, and
There are strong correlation relationships between the two industries, then remove the lesser industry of degree associated with center industry.
The separation of different industries importance in a network and related coefficient power in order to obtain, to it is all less than 1 and
The correlation coefficient value of non-zero carries out frequency statistics, obtains accumulative perception distribution such as Fig. 2.Fig. 2 (a) is frequency disribution, most of
Related coefficient is smaller, and nearly all correlation coefficient value is less than 0.8.Fig. 2 (b) is density function distribution, and abscissa is related coefficient
Value, ordinate are its density, show that 90% confidence level non-zero correlation coefficient value is 0.4 in figure, therefore be set as two for 0.4
Industry whether the threshold value of strong correlation.
By carrying out the screening of threshold value to each industry, weighting in-degree calculates and sequence, and industry is pressed importance ranking, is led to
Complex network model is crossed, therefore can most clearly have found the society of the upstream and downstream incidence relation and inner working between industry
Unity structure, the various factors for influencing some center trade power consumption is found using this upstream-downstream relationship, i.e., property associated therewith is most strong
Mutually independent industry, and be used to electricity consumption growth predict.In terms of the selection of relative influence industry, by all
After the related coefficient sequence and multicollinearity analysis of relevant industries, available each industry is corresponding to influence what it developed
Mutually independent industry.
1.3 separate different electricity consumption behavior states using hidden Markov model
In general, the production of industry is influenced by three classes factor, the first kind is the influence of industry factor itself, including this
All kinds of factors such as industry requirement end, supply side.Second class is the development of the sector vertical industry, will be driven by industry conduction
The sector development.Third class is pure external factor, such as weather.The production prosperity degree of industry be by the sector itself with
And the internal factors such as vertical industry chain conduction determine that unrelated with external factor, therefore, the production prosperity degree of industry should be button
After external factor, purely as caused by the internal factors such as industry itself and industrial structure production is expanded or reduced as.
A kind of investment of the electricity consumption as enterprise's production factors, its variation can represent the expansion of enterprise production in a short time
It opens or reduces, it is since the variation of electricity consumption is influenced by three classes factor, i.e., not only related to the development of industry itself, also and first
State is related in a network for the electricity consumption variation of leading industry, guide's industry, in addition, also will receive external weather conditions, festivals or holidays
Etc. external factor influence, therefore, by combine complex network model, control industrial structure factor influence, and combine weather
Etc. pure external factor, establishes model and predicted.Assuming that center industry be affected by other factors be it is linear, herein
Under the conditions of establish hidden Markov model, the wherein receptance function of state S=i are as follows:
Wherein dytThe electricity consumption growth rate of expression center industry, xjIndicate the variable of guide's industry, including NrelaAssociated row
Industry number,The weighted average of electricity consumption growth rate,In-degree strength weighted average value,Out-degree intensity
Weighted average, wherein the weight of relevant industries is the related coefficient with target industry, weights and considers guide's industry in electricity consumption
Behavior to center industry conduction issue.When the industrial structure upstream-downstream relationship of the sector changes, complex network mould
The related coefficient for the various industries that type provides and relevant industry can also change, and the behavioral characteristics of the industrial structure are then at this time
It has been reflected in the index.OutK, tIndicate all external influences factor, including temperature, temperature square, rainfall, wind speed, air pressure,
Humidity, day off, weekend etc., n are external factor number.ci, αij, βijIt is all the parameter to be estimated.Hidden Markov model
State cannot observe directly, but can be by certain probability by observation vector sequence inspection, each observation vector
Density Distribution shows as various states, if εt(i) be mean value be 0, variance isNormal distribution, the item of the state is indicated with this
Part probability distribution, formula are as follows:
Hidden Markov model carries out predicting that it thinks that current value is therefore the superposition of each state value carries out predicting it
It is preceding it needs to be determined that transition probability between each state, if total status number is Ns, then transfer matrix are as follows:
Wherein γij=p (St=i, St+1=j) it is the probability that system is transferred to another state j from a state i, when being
Between it is relevant.By the hypothesis of Markov model be subsequent time stateful generation probability only it is related with the present situation,
It is unrelated with account of the history, then if it is known that the stateful probability and state-transition matrix of initial time, then can find out each
The state probability at moment.The sum of any time all state probabilities are 1, so total independent parameter includes that institute is stateful initial general
Rate, transfer matrix, each state linear regression coeffficient and variance, therefore total freedom degree are as follows:
1.4 building Industrial Cycle indexes are simultaneously predicted
After determining each element of hidden Markov model, by EM algorithm solve all parameter coefficients of HMM model,
Transfer matrix, state probability, then the electric power growth rate at lower a moment can be predicted as
WhereinIt is the stateful probability vector of t moment, Γ is transfer matrix,It is each state of system
Economic development growth rate vector.dy′t+1It is the sum of stateful economic growth rate weighting of subsequent time institute, for determining subsequent time
Electric power growth rate.
And purely develops caused electricity consumption growth rate from industry self-growth and relevant industries and increase equal to total electricity consumption
Rate deducts the influence of the external factors such as weather, festivals or holidays, therefore electric quantity change related with the prosperous situation of the sector production are as follows:
In formula, uiIt (t) is the probability value at corresponding moment, dy 'tFor total electricity consumption growth rate, dy "tFor electric quantity change.
For different conditions, the coefficient of external factor is significantly different.The production of growth rate, that is, the sector adjusted is prosperous
Situation, it is however generally that, a possibility that the sector production boom, is larger if the value is greater than zero, and the sector is raw if the value is less than zero
It is larger to produce depressed possibility, for prosperous and depressed, is divided with 100 for critical value, after to different adjustment
Electric power growth rate assigns different weight building industry production consumer confidence indexes, and defined formula is as follows:
In formula, W (x) is assignment function, and PIT is the consumer confidence index in [a, b] time range.A, b is time interval,
PIT indicates consumer confidence index in [a, b] time range, and size is between [0,200], the specific criteria for classifying are as follows: 175 with
Upper is " very prosperous " section, [175,125) it is " more prosperous " section, [125,100) it is " faint boom " section, 100 are
Prosperous critical point, (100,75) be " faint depressed " section, (75,50] be " relatively depressed " section, (50,25] be " compared with
It is depressed " section.(25,0) are very depressed.
For the prosperous situation of major class industry, using the average electricity consumption of middle class industry as weight, it is weighted and averaged to obtain
To the prosperous situation of major class industry, the consumer confidence index formula of major class industry are as follows:
In formula, PIT2For major class consumer confidence index, EleiFor the average electricity consumption in the class industry sample phase in i-th, Ele is
Average electricity consumption in the major class industry sample phase, PITiFor i-th of industry production consumer confidence index.
Three, actual results
2.1 based on complex network between the above enterprise's electric power data building industry of scale
To being analyzed two-by-two by the symmetrical growth rate sequence data of period modulation for all industries, phase relation is calculated
Number, Granger causality and lag issue, obtain the matrix between three industries.For dynamic network, window phase length is set
It was 1 year for 365 days, and rolling length is 30 days i.e. one month, since the state change of the industrial structure is slower, one month
Period can capture the dynamic law of the industrial structure, therefore obtain 39 phase networks altogether.Since the industry-specific data in portion start
Time is later, starts have part industry to be not included in the network of several phases, therefore network during consumer confidence index building
The value needs of middle variable are normalized according to node number.The Final Issue part industry obtained by very big flat filter figure
In-degree, weighting in-degree, out-degree, weighting out-degree by weighting out-degree descending arrangement, see the table below, in table first row industry code be state
The economic segmented industry code of the people indicates different industries.
The indexs of correlation such as complex network out-degree, in-degree
It is illustrated in figure 3 relational network figure carrying out treated industry to network:
By complex network model, the upstream and downstream incidence relation between industry and internal fortune can be most clearly had found
The community structure of work finds the various factors for influencing some center trade power consumption using this upstream-downstream relationship, i.e., associated therewith
Property strongest mutually independent industry, and be used to predict electricity consumption growth.In terms of the selection of relative influence industry, pass through
After the related coefficient sequence and multicollinearity analysis of all relevant industries, available each industry is corresponding to influence it
The mutually independent industry of development.
By auto parts and components and accessories manufacturing (industry code: for 367), upstream and downstream and mutually independent business connection
As shown in Figure 4.
2.2 production intensity consumer confidence index buildings
2.2.1 hidden Markov model is estimated by taking automobile component and accessories manufacturing industry as an example
Related to center trade power consumption and mutually independent some rows are obtained by the screening of complex network, in the present embodiment
Industry estimates it entirely due to industry development bring electricity consumption increases core industry in conjunction with meteorological data using hidden Markov model
Long situation.Here other factors, such as macroeconomy situation, price index are not accounted for, it can be considered that these factors
Be already contained in the relevant industry considered, do not need individually to be analyzed, if introduce these factors it is also possible to
Lead to multicollinearity.
Provided by the screening technique of above-mentioned complex network model influence automobile component and the behavior of accessories manufacturing trade power consumption and
Mutually independent industry is as shown in figure 4, indicate various industries with industry code, since the time of in August, 2015, it is assumed that system
In the presence of prosperous and depressed two states, the residual error of each state is normal distribution.
Relevant industries weighting in-degree, weighting out-degree and the relevant industries electricity consumption pair provided in conjunction with complex network model
Claim growth rate, the factors such as ambient weather and festivals or holidays, establish hidden Markov model to the Economic Climate situation of target industry into
Row decomposes.By taking automobile component and accessories manufacturing industry as an example, in fit procedure, since the data volume of weather is bigger, in order to
The readability of model prediction result, to all numerical value divided by 100, final fitting result is as shown in Figure 5 and Figure 6.Fig. 6 can be seen
Out, original electricity consumption growth rate data fluctuation is very big, the changing rule of fitting data and hidden Markov model can succeed.
In order to verify the reliability of model prediction, using 2015.8 --- 2018.9 data are pre- using model as training set
The electricity consumption in October, 2018 is surveyed, as a result as shown in Figure 7.
Model state analysis: by the two states of analysis setting, residual error passed through normal distribution-test, and pass through by
Electricity consumption growth rate weeds out the influence of external factor, it can clearly be seen that:
The case where electricity consumption growth rate of adjustment is negative by model is defined as state 1, and the electricity consumption growth rate of adjustment is positive
Situation is defined as state 2
2.2.2 index is constructed using hidden Markov model
Electricity consumption growth rate after being adjusted can reflect the current prosperous situation of the sector and will be adjusted using weighting function
Electricity consumption growth rate after whole is converted to 0 to 200 prosperous value, and Fig. 8 is from 2015.8 --- 2018.10 probability and future two
A month consumer confidence index.
By analyzing day degree consumer confidence index, it can be seen that boom is converted mutually with depression gaseity, within a certain period of time scape
Gaseity occupies leading, and next period can then switch depression gaseity.Since boom is economic fortune within the scope of certain time
Capable situation, therefore respectively using the moon, season as the period, the prosperous state of day degree is added up, aggregation can be obtained monthly to the moon
Consumer confidence index, aggregation is to obtaining season consumer confidence index, automobile component and accessories manufacturing industry result such as Fig. 9 and Figure 10 institute season
Show.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in various equivalent modifications or replace
It changes, these modifications or substitutions should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with right
It is required that protection scope subject to.
Claims (10)
1. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data, which comprises the following steps:
Step 1: carrying out cleaning arrangement for business electrical information;
Step 2: building complex network model simultaneously chooses relevant industries;
Step 3: separating different electricity consumption behavior states using hidden Markov model;
Step 4: building Industrial Cycle index simultaneously carries out self-adapting detecting.
2. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data according to claim 1, special
Sign is, the step 1 include it is following step by step:
Step 11: collecting business electrical amount data and be modified by interpolation method and sorted out with different industries;
Step 12: business electrical amount data being subjected to cleaning arrangement using symmetrical growth rate model.
3. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data according to claim 2, special
Sign is that the symmetrical growth rate model in the step 12 describes formula are as follows:
In formula, dytFor symmetrical growth rate, L is lag operator, atFor residual sequence, φi,Φ,θiRespectively different lag period systems
Number.
4. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data according to claim 1, special
Sign is, the step 2 include it is following step by step:
Step 21: complex network model is constructed with unit root test, Granger CaFpngerusality test and Pearson correlation coefficient;
Step 22: by carrying out threshold value screening to each industry, weighting in-degree is calculated and sequence is ranked up choosing to various industries
It takes.
5. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data according to claim 1, special
Sign is, the receptance function of the hidden Markov model in the step 3 are as follows:
In formula, dyt(S=i) the electricity consumption growth rate for being state S=i, ci、αijAnd βikTo need the parameter taken into account, xj,tFor elder generation
The variable of leading industry, Outk,tFor out-degree strength weighted average value, εt(i) be mean value be 0, variance isNormal distribution.
6. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data according to claim 1, special
Sign is that the transfer matrix of the hidden Markov model in the step 3 describes formula are as follows:
In formula, γijAnother shape probability of state, N are transferred to from a state for systemsFor total state number.
7. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data according to claim 1, special
Sign is, the step 4 include it is following step by step:
Step 41: calculating and produce the related electric quantity change of prosperous situation in the industry;
Step 42: assigning different weights for the electric quantity change in the industry and construct industry production consumer confidence index;
Step 43: building major class consumer confidence index formula carries out self-adapting detecting to each major class industry.
8. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data according to claim 7, special
Sign is, the calculation formula of electric quantity change in the step 41 are as follows:
In formula, uiIt (t) is the probability value at corresponding moment, dy'tFor total electricity consumption growth rate, dy "tFor electric quantity change.
9. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data according to claim 7, special
Sign is that industry production consumer confidence index, describes formula in the step 42 are as follows:
In formula, W (x) is assignment function, and PIT is the consumer confidence index in [a, b] time range.
10. a kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data according to claim 7, special
Sign is, major class consumer confidence index formula in the step 43 are as follows:
In formula, PIT2For major class consumer confidence index, EleiFor the average electricity consumption in the class industry sample phase in i-th, Ele is that this is big
Average electricity consumption in the class industry sample phase, PITiFor i-th of industry production consumer confidence index.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910198215.1A CN110084398A (en) | 2019-03-15 | 2019-03-15 | A kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910198215.1A CN110084398A (en) | 2019-03-15 | 2019-03-15 | A kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110084398A true CN110084398A (en) | 2019-08-02 |
Family
ID=67413195
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910198215.1A Pending CN110084398A (en) | 2019-03-15 | 2019-03-15 | A kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110084398A (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111582568A (en) * | 2020-04-28 | 2020-08-25 | 国网湖南省电力有限公司 | Electric power data-based enterprise rework prediction method during spring festival |
CN111680938A (en) * | 2020-08-13 | 2020-09-18 | 国网浙江省电力有限公司 | Power flow type big data based rework and production monitoring method and system and readable medium |
CN111737249A (en) * | 2020-08-24 | 2020-10-02 | 国网浙江省电力有限公司 | Abnormal data detection method and device based on Lasso algorithm |
CN111861206A (en) * | 2020-07-20 | 2020-10-30 | 国网上海市电力公司 | Industrial industry prosperity index obtaining method based on enterprise electric power big data |
CN112116265A (en) * | 2020-09-25 | 2020-12-22 | 国网上海市电力公司 | Industry landscape index construction method driven by electric power data |
CN112348281A (en) * | 2020-11-23 | 2021-02-09 | 国网北京市电力公司 | Power data processing method and device |
CN113098837A (en) * | 2021-02-19 | 2021-07-09 | 中国科学院信息工程研究所 | Industrial firewall state detection method and device, electronic equipment and storage medium |
TWI783826B (en) * | 2021-12-14 | 2022-11-11 | 國立中山大學 | Power system state analysis method |
-
2019
- 2019-03-15 CN CN201910198215.1A patent/CN110084398A/en active Pending
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111582568A (en) * | 2020-04-28 | 2020-08-25 | 国网湖南省电力有限公司 | Electric power data-based enterprise rework prediction method during spring festival |
CN111582568B (en) * | 2020-04-28 | 2023-12-19 | 国网湖南省电力有限公司 | Power data-based method for predicting reworking of enterprises in spring festival |
CN111861206A (en) * | 2020-07-20 | 2020-10-30 | 国网上海市电力公司 | Industrial industry prosperity index obtaining method based on enterprise electric power big data |
CN111680938A (en) * | 2020-08-13 | 2020-09-18 | 国网浙江省电力有限公司 | Power flow type big data based rework and production monitoring method and system and readable medium |
CN111737249A (en) * | 2020-08-24 | 2020-10-02 | 国网浙江省电力有限公司 | Abnormal data detection method and device based on Lasso algorithm |
CN112116265A (en) * | 2020-09-25 | 2020-12-22 | 国网上海市电力公司 | Industry landscape index construction method driven by electric power data |
CN112348281A (en) * | 2020-11-23 | 2021-02-09 | 国网北京市电力公司 | Power data processing method and device |
CN113098837A (en) * | 2021-02-19 | 2021-07-09 | 中国科学院信息工程研究所 | Industrial firewall state detection method and device, electronic equipment and storage medium |
CN113098837B (en) * | 2021-02-19 | 2022-08-23 | 中国科学院信息工程研究所 | Industrial firewall state detection method and device, electronic equipment and storage medium |
TWI783826B (en) * | 2021-12-14 | 2022-11-11 | 國立中山大學 | Power system state analysis method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110084398A (en) | A kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data | |
CN110046743B (en) | Public building energy consumption prediction method and system based on GA-ANN | |
Bai et al. | Compound autoregressive network for prediction of multivariate time series | |
CN109933953B (en) | Power distribution network switch state identification method based on probability map model | |
CN107644297B (en) | Energy-saving calculation and verification method for motor system | |
Sørensen et al. | Recent developments in multivariate wind and solar power forecasting | |
Zheng et al. | Interpretable building energy consumption forecasting using spectral clustering algorithm and temporal fusion transformers architecture | |
CN116227716A (en) | Multi-factor energy demand prediction method and system based on Stacking | |
Li et al. | GMM-HMM-based medium-and long-term multi-wind farm correlated power output time series generation method | |
CN114626640A (en) | Natural gas load prediction method and system based on characteristic engineering and LSTM neural network | |
Li et al. | Fault identification in power network based on deep reinforcement learning | |
Li et al. | An innovative combined model based on multi-objective optimization approach for forecasting short-term wind speed: A case study in China | |
Gao et al. | A multifactorial framework for short-term load forecasting system as well as the jinan’s case study | |
Gökçe et al. | Performance comparison of simple regression, random forest and XGBoost algorithms for forecasting electricity demand | |
CN105894138A (en) | Optimum weighted composite prediction method for shipment amount of manufacturing industry | |
Zhang et al. | Spatiotemporal prediction of O3 concentration based on the KNN-Prophet-LSTM model | |
Almqvist | A comparative study between algorithms for time series forecasting on customer prediction: An investigation into the performance of ARIMA, RNN, LSTM, TCN and HMM | |
Li et al. | A hybrid learning method for the data-driven design of linguistic dynamic systems | |
Irankhah et al. | A parallel CNN-BiGRU network for short-term load forecasting in demand-side management | |
López et al. | Standardization of short-term load forecasting models | |
CN115240871A (en) | Epidemic disease prediction method based on deep embedded clustering element learning | |
CN114971032A (en) | Electronic nose online gas concentration prediction method based on OS-ELM | |
Suranata et al. | Feature engineering and long short-term memory for energy use of appliances prediction | |
CN112765141A (en) | Continuous large-scale water quality missing data filling method based on transfer learning | |
Jilani et al. | A new quantile based fuzzy time series forecasting model |
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 |