CN110322063A - A kind of power consumption simulated prediction method and storage medium - Google Patents
A kind of power consumption simulated prediction method and storage medium Download PDFInfo
- Publication number
- CN110322063A CN110322063A CN201910569997.5A CN201910569997A CN110322063A CN 110322063 A CN110322063 A CN 110322063A CN 201910569997 A CN201910569997 A CN 201910569997A CN 110322063 A CN110322063 A CN 110322063A
- Authority
- CN
- China
- Prior art keywords
- power consumption
- time
- prediction
- data
- simulation
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 80
- 238000003860 storage Methods 0.000 title claims abstract description 10
- 238000004088 simulation Methods 0.000 claims abstract description 73
- 230000005611 electricity Effects 0.000 claims abstract description 55
- 238000013135 deep learning Methods 0.000 claims abstract description 21
- 238000009826 distribution Methods 0.000 claims abstract description 18
- 230000008569 process Effects 0.000 claims abstract description 13
- 238000004364 calculation method Methods 0.000 claims description 27
- 238000012360 testing method Methods 0.000 claims description 26
- 238000012549 training Methods 0.000 claims description 24
- 238000012545 processing Methods 0.000 claims description 21
- 238000013136 deep learning model Methods 0.000 claims description 18
- 238000010606 normalization Methods 0.000 claims description 7
- 238000004220 aggregation Methods 0.000 claims description 6
- 230000002776 aggregation Effects 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 238000007476 Maximum Likelihood Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 238000006116 polymerization reaction Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000006399 behavior Effects 0.000 abstract description 4
- 241001269238 Data Species 0.000 abstract 1
- 230000003203 everyday effect Effects 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000005265 energy consumption Methods 0.000 description 4
- 230000002354 daily effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 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
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Abstract
The present invention provides a kind of power consumption simulated prediction method and storage medium, is fitted by the way that the instantaneous electricity-consumption power of default history section is carried out Poisson twice, obtains the stability bandwidth distribution and the use single unit distributed number of predicted time section of predicted time section;The instantaneous electricity-consumption power of predicted time section is calculated with single unit distributed number for predicted time section is obtained further according to described stability bandwidth distribution and predicted time section, and is accurate to each minute, improves the precision of simulation and prediction of the invention.The present invention carries out daily average consumption power prediction also by the way of deep learning, not only there is instantaneous electricity-consumption power in the data that deep learning process uses, there are also meteorological datas and air quality data, since weather condition, air quality can all influence the behavior of power unit, and then influence the power consumption of electricity consumption entity, it eventually results in power consumption to change, therefore considers meteorological data and air quality data, can be further improved predictablity rate.
Description
Technical Field
The invention relates to the technical field of power utilization, in particular to a power consumption simulation prediction method and a storage medium.
Background
With the development of power supply technology and the diversification of power generation modes, electric energy gradually occupies a dominant position in various energy sources. As the world's largest energy consuming country, energy consumption in china is increasing with the improvement of industrial technology level and living standard of people, and energy management has become an important and urgent task for both supply side and consumption side of energy. According to the demand side response theory, the electricity consumption can be reduced by 4-15% by directly feeding back the detailed information of the real-time electricity consumption of the user, and the balance of the supply and demand of the electricity and the safety and stability of the electric power system are favorably maintained. In order to reasonably arrange energy production and management and solve the problem of uneven energy consumption in peak periods of electricity utilization, effective simulation and prediction of electric energy data become an urgent task.
Most of the existing energy prediction methods are based on judging the general trend of energy consumption, reliable prediction and simulation based on specific power consumption unit historical data are lacked, and the instantaneous fluctuation rate of power consumption is not taken into account, so that the judgment result precision is not accurate enough. In recent years, with the continuous development of big data and artificial intelligence technology, deep learning models are widely applied to the prediction field and are effectively practiced in the energy consumption prediction field. However, because the power consumption data of a specific power consumption unit also has the characteristics of high noise and instability, effective simulation of power consumption is still a technical problem to be solved.
Therefore, it is urgently needed to provide a power consumption simulation prediction method and system, which can solve the problems of inaccurate prediction result and the like by using the instantaneous fluctuation rate in a simulation model.
Disclosure of Invention
The invention aims to provide a power consumption power simulation prediction method and a storage medium, which can effectively solve the problems of large time range and the like of the traditional prediction by combining the fluctuation rate and the prediction of the number of user units to further obtain the instantaneous power consumption power prediction of a certain time period or time point in a certain time range in the future; and the meteorological data and the air quality data are introduced into the prediction model through deep learning, so that the prediction accuracy can be improved.
In order to solve the above technical problem, the present invention provides a power consumption simulation prediction method, which includes the following steps: the method comprises the steps of establishing a database, wherein electricity utilization data of at least one electricity utilization entity is input into the database, and each electricity utilization entity comprises at least one electricity utilization unit; the electricity utilization data of each electricity utilization entity comprises instantaneous electricity consumption power of a preset historical time period of the electricity utilization entity and the number of electricity utilization units in the electricity utilization entity, wherein the preset historical time period comprises a plurality of first time sequences; a first sample collection step, wherein first samples are collected from the database, and each first sample comprises instantaneous power consumption of a first time sequence and the number of power consumption units; a first simulation step, namely constructing a first simulation model by using the first sample; and outputting the average instantaneous power consumption per day of a second time series, wherein the second time series is later than the preset historical time period.
Furthermore, the electricity utilization data of each electricity utilization entity further comprises air quality data and air temperature data, and the preset historical time period further comprises a plurality of third time sequences; the power consumption simulation prediction method further comprises the following steps: a second sample collection step of collecting a second sample from the database, wherein the second sample comprises a third time series instantaneous power consumption, air quality data and air temperature data; a second sample classification step, wherein the second sample is randomly classified into a second training sample and a second testing sample; a second prediction model construction step, wherein more than two second training samples are used for training and constructing a prediction model; and a second prediction step of inputting the second test sample into the second prediction model to obtain the power consumption of a fourth time series, wherein the fourth time series is included in the third time series.
Further, the time interval of the instantaneous power consumption in the preset historical time period is 0.5-1.5 minutes; and/or the time range of the first time sequence is 1-28 days; and/or the time range of the third time series is 28 days to 84 days.
Further, after the database establishing step, a preset historical time period data missing value processing step is further included, and the method specifically includes the following steps: inquiring missing time points in the historical time period, wherein the database does not have instantaneous power consumption data corresponding to the missing time points; and an interpolation step of calculating instantaneous power consumption data corresponding to the missing time point and interpolating the instantaneous power consumption data corresponding to the missing time point into the database.
Further, in the interpolation step, according to the instantaneous power consumption data corresponding to the time points before and after the missing time point and the instantaneous power consumption data corresponding to the missing time point in the other first time sequence, the instantaneous power consumption data corresponding to the missing time point is calculated and interpolated by adopting a mean interpolation method, a similar mean interpolation method, a maximum likelihood estimation method or a multiple interpolation method.
Further, in the first simulation step, the method includes: a data preprocessing step, namely performing normalization processing on the instantaneous power consumption of the first time sequence and calculating the power consumption ratio of the power consumption per minute to the whole first time sequence to obtain the instantaneous power consumption ratio X of the first time sequencejThe calculation formula comprises:wherein p isijIs the instantaneous power consumed in the j minute on the ith day in the first time series range; a first Poisson simulation step, by applying a first X to each XjPerforming Poisson simulation to obtain the power consumption fluctuation rate sigma (sigma) of the second time series1,σ2,...,σn) Wherein n x t 1440, said first time interval being comprised in said first time series; a second poisson simulation step of performing simulation calculation on the speed of the electricity utilization unit entering and exiting the electricity utilization entity by using a poisson process to obtain the number Num of the electricity utilization units in the second time sequence (Num ═ Num)1,Num2,...,Num1440) (ii) a A second time-series consumed electric power calculation step of calculating consumed electric power by using the second time-seriesThe instantaneous power consumption calculation of the second time sequence is carried out according to the number of the internal power consumption units and the instantaneous power consumption fluctuation rate of the second time sequence to obtain Pow, and the calculation formula comprises:wherein the power utilization unit comprises a first power utilization unit D1~N(b1,c1) Element and second electrical unit D1~N(b2,c2) And delta is the number of first power utilization units in the first time sequence, the second power utilization units in the second time sequence with low fluctuation rate are included in the second power utilization units in the time sequence, and the instantaneous power consumption Pow of the second time sequence obeys Gaussian distribution.
Further, in the second poisson simulation step, the method specifically includes the following steps: a first data aggregation step of XjPolymerizing at 10-50 time intervals to obtain the instantaneous power consumption ratio of the fifth time sequenceThe polymerization mode is that X in the 10-50 time intervals is calculatedjAverage value of (d); calculating average power consumption of power consumption unit by XjCalculating to obtain the average instantaneous power consumption ratio of the power utilization unit, wherein the calculation formula is as follows:wherein m is the number of power consumption units, XmaxIs the maximum value of the power consumption ratio of the first time series, XminIs the minimum value of the power consumption ratio of the first time sequence; calculating the speed of the electricity utilization unit entering and exiting the electricity utilization entity, namely calculating the electric power consumption of the electricity utilization unit in unit time, wherein the calculation formula comprises the following steps:poisson fitting procedure, according to parameter lambda2And fitting a Poisson process to obtain the number of the power utilization units in the second time sequence.
Further, the method can be used for preparing a novel materialIn the first poisson simulation step, the method specifically includes the following steps: a first time-series division step of dividing XjPerforming a first time interval for cutting, wherein the first time interval comprises 10-50 time intervals; an output step of outputting X in each first time intervaljFitting the poisson process to obtain a poisson fitting value poi of a second time interval; a power consumption fluctuation rate calculation step of calculating a power consumption fluctuation rate by using X in said first time intervaljAnd the fluctuation rate of each time interval is obtained by the calculation of the poi, and the calculation formula comprises the following steps:
further, the fluctuation rate of the first electrical unit is less than the fluctuation rate of the second electrical unit; in the second time-series consumed electric power calculating step, Δ, c1、c2Obtained by solving the following equation:wherein,sampling data of a first power unit in Num and corresponding power consumption fluctuation rate of the second time sequence;the data of the second electrical unit in Num is sampled and the corresponding power consumption fluctuation rate of the second time series is obtained.
Further, in the second sample collection step, the method further includes: a second data aggregation step of obtaining an average instantaneous power consumption per day of the third time series by averaging the instantaneous power consumption of the third time series at time intervals of 1 day; and a normalization processing step of performing normalization processing on the average power consumption per day of the third time series to obtain the second sample.
Further, in the second prediction model constructing step, the method includes: second prediction modelingAnd learning, wherein the second prediction model carries out deep learning according to a first formula, and the first formula comprises the following steps:wherein x isiIs the ith training sample in the training set, n is the number of training samples in the training set, L () is the mean square loss function, λ is the regularization coefficient, J (w) is the regularization term, w is the weight parameter, y is the weight parameteri(vi) simulation prediction data for the power consumed by the user entity in the fourth time series, f (x) for the ith test sample in the test set, vi) A current second prediction model; a step of calculating a first prediction error, wherein a second formula is adopted to determine the first prediction error of the current deep learning model, and the second formula comprises the following steps:wherein M is the first prediction error, n is the number of power load prediction data of the user entity in the fourth time series, Xt'predicting data for the user's t power consumption per day in said fourth time series, XtThe real value of the power consumed each day in the tth second test sample in the test set; if the first prediction error is lower than a preset first error lower limit, the prediction capability meets the preset requirement; and if the prediction capability does not meet the preset requirement, optimizing the current second prediction model until a preset first error lower limit is met.
The present invention also relates to a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, is adapted to carry out any of the above-mentioned methods for power consumption simulation prediction.
The invention has the beneficial effects that: the invention provides a power consumption power simulation prediction method and a storage medium, wherein the fluctuation rate distribution of a prediction time period and the quantity distribution of power utilization units of the prediction time period are obtained by performing Poisson fitting on instantaneous power consumption power of a preset history period for two times; and calculating the instantaneous power consumption of the prediction time period according to the fluctuation rate distribution of the obtained prediction time period and the power consumption unit number distribution of the prediction time period, and the accuracy is accurate to each minute, so that the prediction accuracy of the method is improved. The invention also adopts a deep learning mode to predict the power consumption every day, the deep learning is used as an artificial intelligence method, compared with the traditional time sequence method and a machine learning method, the prediction accuracy is higher, the data adopted in the deep learning process not only has instantaneous power consumption, but also has meteorological data and air quality data, and because weather conditions and air quality can influence the behavior of the power consumption unit, the power consumption of the power consumption entity is further influenced, and finally the power consumption is changed, the meteorological data and the air quality data are considered, and the prediction accuracy can be further improved.
Drawings
The invention is further described below with reference to the figures and examples.
FIG. 1 is a flow chart of a simulation prediction method for power consumption according to the present invention;
FIG. 2 is a flowchart of the default historical time period missing data processing steps provided in the present invention;
FIG. 3 is a flow chart of a first simulation step provided by the present invention;
FIG. 4 is a flowchart of a first Poisson simulation step provided by the present invention;
FIG. 5 is a flowchart of a second Poisson simulation step provided by the present invention;
FIG. 6 is a flow chart of a second sample collection step provided by the present invention;
FIG. 7 is a flowchart of a second predictive model building step provided by the present invention;
FIG. 8 is a functional block diagram of a power consumption simulation prediction system according to the present invention;
FIG. 9 is a functional block diagram of a default historical time period data missing value processing unit according to the present invention;
FIG. 10 is a functional block diagram of a second predictive model building unit provided by the present invention;
a power consumption simulation prediction system 100;
a database 110; a data processing system 200; a preset history time period data missing value processing unit 21;
a first sample collection unit 22; a first simulation unit 23; a second sample acquisition unit 24;
a second sample classification unit 25; a second prediction model construction unit 26; a second prediction unit 27;
a query unit 211; an interpolation unit 212; an initialization unit 261;
a deep learning unit 262; a capability measuring unit 263; a load prediction unit 264.
Detailed Description
The following description of the embodiments refers to the accompanying drawings for illustrating the specific embodiments of the invention which may be wet. The names of the elements, such as the first, the second, etc., mentioned in the present invention are only used for distinguishing different elements and can be better expressed. In the drawings, like parts are designated by like reference numerals and adjacent or similar parts are designated by like reference numerals.
Embodiments of the present invention will be described in detail herein with reference to the accompanying drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. These embodiments are provided to explain the practical application of the invention and to enable others skilled in the art to understand the invention for various embodiments and with various modifications as are suited to the particular use contemplated.
As shown in FIG. 1, the present invention provides a computer software method for implementing a power consumption simulation prediction method by a computer, comprising the following steps S1-S8.
S1), a database establishing step, wherein electricity utilization data of at least one electricity utilization entity is recorded into a database, and each electricity utilization entity comprises at least one electricity utilization unit; the electricity utilization data of each electricity utilization entity comprises instantaneous electricity consumption power of a preset historical time period of the electricity utilization entity, the number of electricity utilization units in the electricity utilization entity, air quality data and air temperature data, wherein the preset historical time period comprises a plurality of first time sequences and a plurality of third time sequences.
The time interval of the instantaneous power consumption in the preset historical time period is 0.5-1.5 minutes, and the optimal time interval is 1 minute; the time range of the first time sequence is 1-28 days.
S2) a preset historical time period data missing value processing step for interpolating instantaneous power consumption data of the missing preset historical period; as shown in fig. 2, the method specifically includes the following steps S21 to S22.
S21), inquiring missing time points in the historical time period, wherein the database does not have instantaneous power consumption data corresponding to the missing time points.
S22), an interpolation step of calculating instantaneous power consumption data corresponding to the missing time point and interpolating the instantaneous power consumption data corresponding to the missing time point into the database.
In the interpolation step, according to the instantaneous power consumption data corresponding to the time points before and after the missing time point and the instantaneous power consumption data corresponding to the missing time point in other first time sequences, the instantaneous power consumption data corresponding to the missing time point is calculated and interpolated by adopting a mean interpolation method, a similar mean interpolation method, a maximum likelihood estimation method or a multiple interpolation method.
S3), a first sample collection step of collecting first samples from the database, each first sample comprising a first time series of instantaneous power consumption and a number of power consumption units.
S4), a first simulation step, namely constructing a first simulation model by using the first sample; and outputting instantaneous power consumption of a second time series, wherein the second time series is later than the preset historical time period.
As shown in fig. 3, the first simulation step specifically includes the following steps S41 to S44.
S41) data preprocessing step, namely normalizing the instantaneous power consumption of the first time sequence and calculating the ratio of the power consumption per minute to the power consumption of the whole first time sequence to obtain the instantaneous power consumption ratio X of the first time sequencejThe calculation formula comprises:
wherein p isijIs the instantaneous consumed power of the j minute on the ith day in the range of the first time sequence.
The first time series is preferably every minute of the day, i.e. the first time series has 1440 data.
S42) a first Poisson simulation step, by for each XjPerforming Poisson simulation to obtain the power consumption fluctuation rate sigma (sigma) of the second time series1,σ2,...,σn) Wherein n x t 1440, said first time interval is comprised in said first time series.
As shown in fig. 4, the first poisson simulation step specifically includes the following steps S421 to S423.
S421) a first time-series segmentation step of dividing XjPerforming a first time interval for division, wherein the first time interval comprises 10-50 time intervals, and preferably 15 time intervals, namely 15 minutes;
s422) an output step of X in each first time intervalj(X1~X15、X16~X30..) fitting the poisson process to obtain poisson fitting values poi (poi) of a first time interval1~poi15、poi16~poi30、...);
S423) a power consumption fluctuation ratio calculation step by using X in a plurality of first time intervalsjAnd the fluctuation rate of each time interval is obtained by the calculation of the poi, and the calculation formula comprises the following steps:
where k is the length of the first time interval.
Preferably, a time of day according to said first time series, then σi=(σ1,σ2,...,σ96)。
S43) a second Poisson simulation step, wherein the electricity utilization unit enters and exits the electricity utilization unitThe speed of the entity is simulated and calculated by using a poisson process to obtain the number Num of the power utilization units in the second time sequence (Num)1,Num2,...,Num1440)。
As shown in fig. 5, the second poisson simulation step specifically includes the following steps S431 to S434.
S431) first data aggregation step for XjPolymerizing at 10-50 time intervals to obtain the instantaneous power consumption ratio of the fifth time sequenceThe polymerization mode is that X in the 10-50 time intervals is calculatedjAverage value of (d); the fifth time sequence is 15 time intervals, i.e. 15 minutes, then
S432) average power consumption calculation step of power consumption unit by XjCalculating to obtain the average instantaneous power consumption ratio of the power utilization unit, wherein the calculation formula is as follows:wherein m is the number of power consumption units, XmaxIs the maximum value of the power consumption ratio of the first time series, XminIs the minimum value of the power consumption ratio of the first time series.
S433) calculating the speed of the electricity utilization unit entering and exiting the electricity utilization entity, namely calculating the electric power consumed by the electricity utilization unit in unit time, wherein the calculation formula comprises the following steps:from step S431, i ∈ [1,96 ]]。
S434) Poisson fitting step according to parameter lambda2Fitting a poisson process to obtain the number of the power utilization units in the second time sequence; each lambda at the time of output2Including 15 time intervals, i.e., Num of outputs is 1440 time intervals.
S44) A second time series power consumption calculation step of calculating instantaneous power consumption of a second time series by using the number of power consumption units in the second time series and the instantaneous power consumption fluctuation rate of the second time series to obtain Pow, wherein the calculation formula comprises:
wherein the power utilization unit comprises a first power utilization unit D1~N(b1,c1) Element and second electrical unit D1~N(b2,c2) The fluctuation rate of the first electric unit is smaller than that of the second electric unit; the whole predicted number of the power utilization units can be regarded asA gaussian distribution of (a).
And delta is the number of the first power utilization units in the first time sequence, the second power utilization units in the second time sequence with low fluctuation rate are included in the second power utilization units in the time sequence, and the instantaneous power consumption Pow of the second time sequence obeys Gaussian distribution.
In the second time-series consumed electric power calculating step, Δ, c1、c2Obtained by solving the following equation:
wherein,sampling data of a first power unit in Num and corresponding power consumption fluctuation rate of the second time sequence;the data of the second electrical unit in Num is sampled and the corresponding power consumption fluctuation rate of the second time series is obtained.
S5), a second sample collection step, wherein data are collected from the database and processed to obtain a second sample, and the second sample comprises third time series instantaneous power consumption, air quality data and air temperature data.
As shown in fig. 6, the second sample collection step includes the following steps S51 to S52.
S51) a second data aggregation step of obtaining an average power consumption power per day of the third time series by averaging the instantaneous power consumption powers of the third time series with 1 day as a time interval.
S52), a normalization processing step of normalizing the average power consumption per day of the third time series to obtain the second sample.
S6), a second sample classification step, wherein the second sample is randomly classified into a second training sample and a second testing sample.
S7), a second prediction model construction step, wherein the prediction model is trained and constructed by more than two second training samples.
S8) a second prediction step, wherein the second test sample is input into the second prediction model to obtain the daily average power consumption of a fourth time series, and the fourth time series is later than the preset historical time period.
As shown in fig. 7, the second prediction model building step includes the following steps S81 to S82.
S81) a second prediction model learning step of deep learning according to a first formula including:
wherein x isiIs the ith training sample in the training set, n is the number of training samples in the training set, L () is the mean square loss function, λ is the regularization coefficient, J (w) is the regularization term, w is the weight parameter, y is the weight parameteriFor the ith test sample in the test set, v is power consumption prediction data for the user for each day of the fourth time series, f (x)i) A current second prediction model;
s82), determining a first prediction error of the current deep learning model using a second formula, the second formula comprising:
wherein M is the first prediction error, n is the number of power load prediction data of the user entity in the fourth time series, Xt' average power consumption prediction data per day for user in said fourth time series, XtIs the true value of the power consumed per day in the tth second test sample in the test set.
If the first prediction error is lower than a preset first error lower limit, the prediction capability meets the preset requirement; and if the prediction capability does not meet the preset requirement, optimizing the current second prediction model until a preset first error lower limit is met.
The invention provides a power consumption power simulation prediction method, which comprises the steps of carrying out Poisson fitting twice on instantaneous power consumption power of a preset historical period to obtain fluctuation rate distribution of a prediction time period and power consumption unit quantity distribution of the prediction time period; and calculating the instantaneous power consumption of the prediction time period according to the fluctuation rate distribution of the obtained prediction time period and the power consumption unit number distribution of the prediction time period, wherein the instantaneous power consumption is accurate to each minute, so that the prediction accuracy of the method is improved.
The invention also adopts a deep learning mode to predict the power consumption every day, the deep learning is used as an artificial intelligence method, compared with the traditional time sequence method and a machine learning method, the prediction accuracy is higher, the data adopted in the deep learning process not only has the average power consumption every day, but also has meteorological data and air quality data, and because the weather condition and the air quality can influence the behavior of the power consumption unit, the power consumption of the power consumption entity is further influenced, and the power consumption is finally changed, the meteorological data and the air quality data are considered, and the prediction accuracy can be improved.
The invention also provides an electronic device, comprising a memory and a processor, wherein the memory is used for storing the executable program codes; the processor executes the program corresponding to the executable program code by reading the executable program code to perform the steps of the power consumption prediction method.
As shown in fig. 8, the electronic device includes a power consumption simulation prediction system 100, which includes a database 110 and a data processing system 200, where the database 110 is stored in the memory, and the data processing system 200 is the processor described above.
The database 110 includes power consumption data of at least one power consumption entity, the power consumption data of each power consumption entity includes instantaneous power consumption of the power consumption entity in a preset historical time period, the number of power consumption units in the power consumption entity, air quality data and air temperature data, and the preset historical time period includes a plurality of first time sequences and a plurality of third time sequences.
The data processing system 200 includes: the system comprises a preset historical time period data missing value processing unit 21, a first sample acquisition unit 22, a first simulation unit 23, a second sample acquisition unit 24, a second sample classification unit 25, a second prediction model construction unit 26 and a second prediction unit 27.
As shown in fig. 9, the preset historical time period data missing value processing unit 21 specifically includes an inquiring unit 211 and an interpolating unit 212.
The querying unit 211 queries the missing time point in the historical time period, and the database 110 does not have the instantaneous power consumption data corresponding to the missing time point.
The interpolation unit 212 is configured to calculate the instantaneous power consumption data corresponding to the missing time point, and interpolate the instantaneous power consumption data corresponding to the missing time point to the database 110.
The first sample acquiring unit 22 acquires first samples from the database 110, each of the first samples including a first time series of instantaneous power consumption and a number of power consuming units.
The first simulation unit 23 constructs a first simulation model by using the first sample; and outputting instantaneous power consumption of a second time series, wherein the second time series is later than the preset historical time period.
The second sample collection unit 24 collects a second sample from the database 110, the second sample including a third time series of instantaneous power consumption, air quality data, and air temperature data.
The second sample classification unit 25 randomly classifies the second samples into two classes, a second training sample and a second testing sample.
The second prediction model construction unit 26 trains and constructs a prediction model using two or more second training samples.
The second prediction unit 27 inputs the second test sample into the second prediction model to obtain the power consumption of a fourth time series, where the fourth time series is later than the preset historical time period.
As shown in fig. 10, the second prediction model building unit 26 is configured to determine a deep learning model for predicting the instantaneous power consumption of the sixth time series per day, and the second prediction model building unit 26 includes an initialization unit 261, a deep learning unit 262, a capacity measurement unit 263, and a load prediction unit 264.
The initialization unit 261 is configured to perform initialization setting on a network structure and model parameters of the deep learning model;
the deep learning unit 262 is configured to input the second training sample into a current deep learning model, and perform deep learning through the current deep learning model to obtain the daily instantaneous power consumption data of the user entity in the fourth time sequence.
The capability measuring unit 263 is configured to measure the prediction capability of the current deep learning model according to the daily average power consumption measurement data of the user entity in the fourth time interval and the second test sample, and if the prediction capability does not meet a predetermined requirement, adjust the network structure and/or the model parameters of the current deep learning model, and return the adjusted network structure and/or model parameters to the deep learning unit 262; and if not, taking the current deep learning model as the deep learning model for carrying out power consumption simulation prediction, and transferring to the load prediction module.
The load prediction unit 264 is configured to input the second test sample into the deep learning model for performing the power consumption simulation prediction, so as to obtain an average power consumption per day of the user in a sixth time series.
The characteristic extraction unit is used for extracting characteristics influencing the prediction result of the deep learning model for performing power consumption simulation prediction and influence degree values corresponding to the characteristics.
The electronic device provided by the invention is provided with a power consumption power simulation prediction system 100, the first simulation unit 23 outputs the instantaneous power consumption power per minute every day in the future according to the data in the historical stage, and the output time precision is small. And the second prediction model initializes the model structure and/or the model parameters to obtain an initial deep learning model, then the initial deep learning model carries out deep learning by using the training set to obtain a prediction result, the prediction capability of the model is determined according to the prediction result and the test set, if the prediction capability does not meet the requirement, the model structure and/or the parameters are adjusted until the prediction capability meets the requirement to obtain a final deep learning model, and the final deep learning model carries out testing by using the test set to obtain a desired test result.
The present invention also provides a computer-readable storage medium having a computer program stored thereon, which when executed in a processor of the electronic device, is capable of implementing the power consumption simulation prediction method.
The invention provides a power consumption power simulation prediction method and a storage medium, wherein the fluctuation rate distribution of a prediction time period and the quantity distribution of power utilization units of the prediction time period are obtained by performing Poisson fitting on instantaneous power consumption power of a preset history period for two times; and then, calculating the instantaneous power consumption of the prediction time period according to the fluctuation rate distribution of the obtained prediction time period and the power consumption unit number distribution of the prediction time period, and accurately obtaining the instantaneous power consumption of the prediction time period in each minute, thereby improving the prediction accuracy of the invention.
The invention also adopts a deep learning mode to predict the power consumption every day, the deep learning is used as an artificial intelligence method, compared with the traditional time sequence method and a machine learning method, the prediction accuracy is higher, the data adopted in the deep learning process not only has instantaneous power consumption, but also has meteorological data and air quality data, and because weather conditions and air quality can influence the behavior of the power consumption unit, the power consumption of the power consumption entity is further influenced, and finally the power consumption is changed, the meteorological data and the air quality data are considered, and the prediction accuracy can be further improved.
It should be noted that many variations and modifications of the embodiments of the present invention fully described are possible and are not to be considered as limited to the specific examples of the above embodiments. The above examples are intended to be illustrative of the invention and are not intended to be limiting. In conclusion, the scope of the present invention should include those changes or substitutions and modifications which are obvious to those of ordinary skill in the art.
Claims (12)
1. A power consumption simulation prediction method is characterized by comprising the following steps:
the method comprises the steps of establishing a database, wherein electricity utilization data of at least one electricity utilization entity is input into the database, and each electricity utilization entity comprises at least one electricity utilization unit; the electricity utilization data of each electricity utilization entity comprises instantaneous electricity consumption power of a preset historical time period of the electricity utilization entity and the number of electricity utilization units in the electricity utilization entity, wherein the preset historical time period comprises a plurality of first time sequences;
a first sample collection step, wherein first samples are collected from the database, and each first sample comprises instantaneous power consumption of a first time sequence and the number of power consumption units; and
a first simulation step, constructing a first simulation model by using the first sample; and outputting instantaneous power consumption of a second time series, wherein the second time series is later than the preset historical time period.
2. The consumed power simulation prediction method according to claim 1,
the electricity consumption data of each electricity consumption entity also comprises air quality data and air temperature data,
the preset historical time period further comprises a plurality of third time sequences;
the power consumption simulation prediction method further comprises the following steps:
a second sample collection step of collecting a second sample from the database, wherein the second sample comprises a third time series instantaneous power consumption, air quality data and air temperature data;
a second sample classification step, wherein the second sample is randomly classified into a second training sample and a second testing sample;
a second prediction model construction step, wherein more than two second training samples are used for training and constructing a prediction model;
and a second prediction step of inputting the second test sample into the second prediction model to obtain the average power consumption per day of a fourth time series, wherein the fourth time series is included in the third time series.
3. The consumed power simulation prediction method according to claim 1,
the time interval of the instantaneous power consumption in the preset historical time period is 0.5-1.5 minutes; and/or the presence of a gas in the gas,
the time range of the first time sequence is 1-28 days.
4. The consumed power simulation prediction method according to claim 1,
after the database establishing step, a preset historical time period data missing value processing step is further included, and the method specifically includes the following steps:
inquiring missing time points in the historical time period, wherein the database does not have instantaneous power consumption data corresponding to the missing time points; and
and an interpolation step of calculating the instantaneous power consumption data corresponding to the missing time point and interpolating the instantaneous power consumption data corresponding to the missing time point into the database.
5. The consumed power simulation prediction method according to claim 4,
in the step of the interpolation, the interpolation is performed,
instantaneous power consumption data according to time points before and after the missing time point, an
Instantaneous power consumption data corresponding to the missing time point in the other first time series,
and calculating and interpolating instantaneous power consumption data corresponding to the missing time point by adopting a mean interpolation method, a similar mean interpolation method, a maximum likelihood estimation method or a multiple interpolation method.
6. The consumed power simulation prediction method according to claim 1,
in the first simulation step, the method includes:
a data preprocessing step, namely performing normalization processing on the instantaneous power consumption of the first time sequence and calculating the power consumption ratio of the power consumption per minute to the whole first time sequence to obtain the instantaneous power consumption ratio X of the first time sequencejThe calculation formula comprises:
wherein ,pijIs the instantaneous power consumed in the j minute on the ith day in the first time series range;
a first Poisson simulation step, by applying a first X to each XjPerforming Poisson simulation processing to obtain power consumption fluctuation ratio sigma of the second time series1,σ2,...,σn) Wherein n x t 1440, said first time interval being comprised in said first time series;
a second Poisson simulation step of the power consumptionThe speed of the unit entering and exiting the electricity utilization entity is subjected to simulation processing by using a Poisson process, and the number Num (Num) of the electricity utilization units in the second time sequence is obtained1,Num2,...,Num1440);
A second time series power consumption calculation step of calculating instantaneous power consumption of a second time series by using the number of power consumption units in the second time series and the instantaneous power consumption fluctuation rate of the second time series to obtain Pow, wherein the calculation formula comprises:
wherein the power utilization unit comprises a first power utilization unit D1~N(b1,c1) Element and second electrical unit D1~N(b2,c2) Delta is the number of first power units in the first time sequence,
the second time-series low fluctuation rate electricity-using unit is included in the second time-series electricity-using unit, and the instantaneous power consumption Pow of the second time-series follows a gaussian distribution.
7. The consumed power simulation prediction method according to claim 6,
in the second poisson simulation step, the method specifically includes the following steps:
a first data aggregation step of XjPolymerizing at 10-50 time intervals to obtain the instantaneous power consumption ratio of the fifth time sequenceThe polymerization mode is that X in the 10-50 time intervals is calculatedjAverage value of (d);
calculating average power consumption of power consumption unit by XjCalculating to obtain the average instantaneous power consumption ratio of the power utilization unit, wherein the calculation formula is as follows:wherein m is the number of power consumption units, XmaxIs the maximum value of the power consumption ratio of the first time series, XminIs the minimum value of the power consumption ratio of the first time sequence;
calculating the speed of the electricity utilization unit entering and exiting the electricity utilization entity, namely calculating how many electricity utilization units consume the electric power in unit time, wherein the calculation formula comprises the following steps:
a poisson fitting step, according to poisson parameter lambda2And fitting a Poisson process to obtain the number of the power utilization units in the second time sequence.
8. The consumed power simulation prediction method according to claim 6,
in the first poisson simulation step, the method specifically includes the following steps:
a first time-series division step of dividing XjDividing the length of a first time interval, wherein the first time interval comprises 10-50 time intervals;
output step of λ1I.e. X in each first time intervaljCarrying out Poisson process fitting to obtain a Poisson fitting value poi of a first time interval;
a power consumption fluctuation rate calculation step of calculating a power consumption fluctuation rate by using X in a plurality of first time intervalsjAnd the fluctuation rate of each time interval is obtained by the calculation of the poi, and the calculation formula comprises the following steps:
wherein k is the length of the second time interval.
9. The consumed power simulation prediction method according to claim 6,
the fluctuation rate of the first electric unit is smaller than that of the second electric unit;
in the second time-series consumed electric power calculating step, Δ, c1、c2Obtained by solving the following equation:
wherein ,sampling data of a first power unit in Num and corresponding power consumption fluctuation rate of the second time sequence;the data of the second electrical unit in Num is sampled and the corresponding power consumption fluctuation rate of the second time series is obtained.
10. The consumed power simulation prediction method according to claim 2,
in the second sample collection step, the method further comprises:
a second data aggregation step of obtaining an average instantaneous power consumption per day of the third time series by averaging the instantaneous power consumption of the third time series at time intervals of 1 day;
and a normalization processing step of performing normalization processing on the average power consumption per day of the third time series to obtain the second sample.
11. The consumed power simulation prediction method according to claim 1,
in the second prediction model constructing step, the method includes:
a second prediction model learning step of performing deep learning according to a first formula, the first formula including:
wherein ,xiIs the ith training sample in the training set, n is the number of training samples in the training set, L () is the mean square loss function, λ is the regularization coefficient, J (w) is the regularization term, w is the weight parameter, y is the weight parameteri(vi) simulation prediction data for the power consumed by the user entity in the fourth time series, f (x) for the ith test sample in the test set, vi) A current second prediction model;
a step of calculating a first prediction error, wherein a second formula is adopted to determine the first prediction error of the current deep learning model, and the second formula comprises the following steps:
wherein M is the first prediction error, n is the number of power load prediction data of the user entity in the fourth time series, X'tPredicting data, X, for the t-th power consumption per day of the user in said fourth time seriestThe real value of the average power consumption per day in the tth second test sample in the test set is obtained;
if the first prediction error is lower than a preset first error lower limit, the prediction capability meets the preset requirement; and if the prediction capability does not meet the preset requirement, optimizing the current second prediction model until a preset first error lower limit is met.
12. A storage medium having a computer program stored thereon, wherein the method of any one of claims 1-11 is implemented when the computer program is executed by a processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910569997.5A CN110322063B (en) | 2019-06-27 | 2019-06-27 | Power consumption simulation prediction method and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910569997.5A CN110322063B (en) | 2019-06-27 | 2019-06-27 | Power consumption simulation prediction method and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110322063A true CN110322063A (en) | 2019-10-11 |
CN110322063B CN110322063B (en) | 2023-05-30 |
Family
ID=68121293
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910569997.5A Active CN110322063B (en) | 2019-06-27 | 2019-06-27 | Power consumption simulation prediction method and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110322063B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111126780A (en) * | 2019-10-31 | 2020-05-08 | 内蒙古电力(集团)有限责任公司包头供电局 | Non-invasive load monitoring method and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2014110679A (en) * | 2012-11-30 | 2014-06-12 | Mitsubishi Electric Corp | Power consumption estimation system, power consumption estimation method, and program |
CN107239852A (en) * | 2017-05-05 | 2017-10-10 | 南京邮电大学 | A kind of electric quantity consumption Forecasting Methodology based on deep learning |
CN107769235A (en) * | 2017-09-29 | 2018-03-06 | 国网上海市电力公司 | A kind of microgrid energy management method based on hybrid energy-storing and electric automobile |
CN108416695A (en) * | 2018-02-24 | 2018-08-17 | 合肥工业大学 | Electric load probability density prediction technique based on deep learning and system, medium |
-
2019
- 2019-06-27 CN CN201910569997.5A patent/CN110322063B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2014110679A (en) * | 2012-11-30 | 2014-06-12 | Mitsubishi Electric Corp | Power consumption estimation system, power consumption estimation method, and program |
CN107239852A (en) * | 2017-05-05 | 2017-10-10 | 南京邮电大学 | A kind of electric quantity consumption Forecasting Methodology based on deep learning |
CN107769235A (en) * | 2017-09-29 | 2018-03-06 | 国网上海市电力公司 | A kind of microgrid energy management method based on hybrid energy-storing and electric automobile |
CN108416695A (en) * | 2018-02-24 | 2018-08-17 | 合肥工业大学 | Electric load probability density prediction technique based on deep learning and system, medium |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111126780A (en) * | 2019-10-31 | 2020-05-08 | 内蒙古电力(集团)有限责任公司包头供电局 | Non-invasive load monitoring method and storage medium |
CN111126780B (en) * | 2019-10-31 | 2023-04-07 | 内蒙古电力(集团)有限责任公司包头供电局 | Non-invasive load monitoring method and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110322063B (en) | 2023-05-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108280552B (en) | Power load prediction method and system based on deep learning and storage medium | |
CN106600037B (en) | Multi-parameter auxiliary load prediction method based on principal component analysis | |
CN112946484B (en) | SOC estimation method, system, terminal equipment and readable storage medium based on BP neural network | |
CN102999791A (en) | Power load forecasting method based on customer segmentation in power industry | |
CN112990500B (en) | Transformer area line loss analysis method and system based on improved weighted gray correlation analysis | |
CN114004296A (en) | Method and system for reversely extracting monitoring points based on power load characteristics | |
CN113868953B (en) | Multi-unit operation optimization method, device and system in industrial system and storage medium | |
CN113344288A (en) | Method and device for predicting water level of cascade hydropower station group and computer readable storage medium | |
CN116227637A (en) | Active power distribution network oriented refined load prediction method and system | |
CN110212592B (en) | Thermal power generating unit load regulation maximum rate estimation method and system based on piecewise linear expression | |
CN114119273A (en) | Park comprehensive energy system non-invasive load decomposition method and system | |
CN116148753A (en) | Intelligent electric energy meter operation error monitoring system | |
CN115545333A (en) | Method for predicting load curve of multi-load daily-type power distribution network | |
CN106295877B (en) | Method for predicting electric energy consumption of smart power grid | |
CN109523077B (en) | Wind power prediction method | |
CN110322063A (en) | A kind of power consumption simulated prediction method and storage medium | |
CN110414734B (en) | Method for forecasting and evaluating wind resource utilization rate | |
CN116470491A (en) | Photovoltaic power probability prediction method and system based on copula function | |
CN116561569A (en) | Industrial power load identification method based on EO feature selection and AdaBoost algorithm | |
CN115130764A (en) | Power distribution network situation prediction method and system based on state evaluation | |
CN112801388B (en) | Power load prediction method and system based on nonlinear time series algorithm | |
CN114971090A (en) | Electric heating load prediction method, system, equipment and medium | |
CN110648248A (en) | Control method, device and equipment for power station | |
CN112836920A (en) | Coal electric unit energy efficiency state evaluation method and device and coal electric unit system | |
CN115114983B (en) | Method for acquiring and analyzing electric quantity data based on big data equipment and computer system |
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 | ||
TR01 | Transfer of patent right |
Effective date of registration: 20231130 Address after: 201108 3 / F and 4 / F, building 2, 598 Guanghua Road, Minhang District, Shanghai Patentee after: Shanghai Jientropy Data Technology Co.,Ltd. Address before: 201108 room D62, 1st floor, building 6, 4299 Jindu Road, Minhang District, Shanghai Patentee before: SHANGHAI MAXTROPY DATA TECHNOLOGY Co.,Ltd. |
|
TR01 | Transfer of patent right |