CN110807508A - Bus peak load prediction method considering complex meteorological influence - Google Patents

Bus peak load prediction method considering complex meteorological influence Download PDF

Info

Publication number
CN110807508A
CN110807508A CN201911056853.6A CN201911056853A CN110807508A CN 110807508 A CN110807508 A CN 110807508A CN 201911056853 A CN201911056853 A CN 201911056853A CN 110807508 A CN110807508 A CN 110807508A
Authority
CN
China
Prior art keywords
bus
peak load
prediction
value
optimal
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
Application number
CN201911056853.6A
Other languages
Chinese (zh)
Other versions
CN110807508B (en
Inventor
朱赫炎
张明理
于长永
蒋理
刘靖波
徐维懋
潘霄
宋坤
卢天琪
邬桐
南哲
梁毅
黄南天
贺庆奎
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Corp of China SGCC
Northeast Electric Power University
Economic and Technological Research Institute of State Grid Liaoning Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
Northeast Dianli University
Economic and Technological Research Institute of State Grid Liaoning Electric Power Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, Northeast Dianli University, Economic and Technological Research Institute of State Grid Liaoning Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN201911056853.6A priority Critical patent/CN110807508B/en
Publication of CN110807508A publication Critical patent/CN110807508A/en
Application granted granted Critical
Publication of CN110807508B publication Critical patent/CN110807508B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Abstract

A bus peak load prediction method considering complex meteorological influence belongs to the technical field of bus peak load prediction. The method takes the characteristic importance result of the condition mutual information on the characteristic to be selected in the original characteristic set as the basis, combines IPSO-ELM as a predictor, carries out forward characteristic selection, determines the optimal characteristic set of the bus peak load prediction, reduces the influence of characteristic redundancy on the prediction precision during the bus peak load prediction, effectively improves the prediction precision of different buses by respectively constructing an optimal prediction model aiming at different buses, introduces an improved particle swarm optimization extreme learning machine to be combined with a linear method, carries out the peak load prediction under different scenes, and meets the prediction requirement under a small sample or no sample scene.

Description

Bus peak load prediction method considering complex meteorological influence
Technical Field
The invention belongs to the technical field of bus peak load prediction, and particularly relates to a bus peak load prediction method considering complex meteorological influence.
Background
The problems of limited historical data, severe fluctuation, non-linearity and randomness of the peak load of the bus, low prediction accuracy and difficult prediction become problems to be solved urgently, and how to improve the prediction accuracy of the peak load of the bus becomes a problem to be solved urgently. The prediction of the peak load of the bus is an important basis for guaranteeing reliable and stable operation of a power system, and the method for analyzing and researching the prediction accuracy of the peak load of the bus has very important significance.
At present, many researches are carried out on bus load prediction, the bus load prediction is optimized respectively according to bus load characteristics, however, the influences of various factors such as natural weather and society on the bus load are not fully analyzed, when numerous factors are considered, feature selection is not carried out, the difference of different bus influencing factors is not considered, the targeted feature selection is not carried out on the different bus load influencing factors, and a targeted bus load prediction model is not established on the basis. The existing research related to peak load prediction is carried out on many influencing factors and prediction methods for the peak load of the urban power grid, and although the existing research improves the peak load prediction precision to a certain extent, the existing research does not carry out targeted analysis on small sample problems with limited historical data in peak load prediction.
Therefore, there is a need in the art for a new solution to solve this problem.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the bus peak load prediction method considering the complex meteorological influence is provided for solving the technical problem that in the prior art, no targeted analysis is carried out on small sample problems of severe peak load fluctuation and limited historical data in peak load prediction.
The bus peak load prediction method considering the complex meteorological influence comprises the following steps which are sequentially carried out,
the method comprises the steps that firstly, based on the influence of natural weather and social factors on the prediction precision of the peak load of the bus bar, an original feature set for predicting the peak load of the bus bar is constructed according to the natural weather and social factors, and the correlation between each feature in the original feature set and the peak load of the bus bar is analyzed through Conditional Mutual Information (CMI), so that feature importance degree sequencing is obtained;
step two, optimizing the input weight and the threshold value of the Extreme Learning Machine by improving a Particle Swarm algorithm to obtain an Improved Particle Swarm optimized Extreme Learning Machine (IPSO-ELM), then carrying out targeted forward feature selection by taking the prediction precision of the IPSO-ELM as a decision variable according to the feature importance ranking result obtained in the step one to obtain an optimal feature subset of bus peak load prediction, and retraining the IPSO-ELM according to the obtained optimal feature subset to obtain an optimal bus peak load prediction model;
step three, substituting test set data in the historical data within a set time period into the optimal bus peak load prediction model obtained in the step two to obtain a predicted value of the bus peak load;
respectively counting the peak load of the bus under the extremely high temperature condition and the peak load of the bus under the extremely low temperature condition through a linear model, and obtaining that the peak load of the bus has linear relations with the extremely high temperature and the extremely low temperature according to a statistical result;
and fifthly, performing linear fitting on the extreme high temperature and the extreme low temperature and the peak load of the bus by using a least square method, and respectively obtaining the peak load prediction linear models of the bus 1 and the bus 2 under the extreme temperature condition as follows:
under the extreme high temperature condition, the linear model for predicting the peak load of the bus 1 is as follows:
Figure BDA0002256772960000021
under the extreme low temperature condition, the linear model for predicting the peak load of the bus 1 is as follows:
Figure BDA0002256772960000022
under the extreme high temperature condition, the linear model for predicting the peak load of the bus 2 is as follows:
Figure BDA0002256772960000023
under the extreme low temperature condition, the linear model for predicting the peak load of the bus 2 is as follows:
in the formula:
Figure BDA0002256772960000025
peak load of the bus 1 under extreme high temperature conditions;peak load of the bus 2 under extreme high temperature conditions;
Figure BDA0002256772960000027
peak load of the bus 1 under extreme low temperature conditions;
Figure BDA0002256772960000028
peak load of the bus 2 under extreme low temperature conditions; t is the extreme temperature.
The natural weather includes longitude, latitude, temperature, air pressure, humidity, wind direction and wind speed.
The social factors include date and holidays.
The prediction precision in the second step is the average absolute percentage Error (MAPE) between the predicted value and the actual value, and the calculation formula of the MAPE is as follows:
Figure BDA0002256772960000031
in the formula, YiThe measured value of the peak load of the bus is taken as the measured value;
Figure BDA0002256772960000032
predicting a bus peak load value; and N is the number of prediction samples.
The specific method for obtaining the IPSO-ELM by optimizing the input right and the threshold of the extreme learning machine through the improved particle swarm optimization in the step two comprises the following steps:
1) constructing an ELM model for predicting the peak load of the bus bar based on the original characteristic set, and randomly generating an input weight omega of the ELM and a hidden layer unit bias threshold b;
2) determining data of an original feature set required by input bus peak load prediction, and setting iteration times;
3) predicting original characteristic set data aiming at the peak load of the bus, and carrying out normalization processing;
4) obtaining a fitness value according to an average absolute percentage error value of bus peak load prediction of the ELM, and obtaining a current individual fitness value and a group optimal fitness value;
5) updating the particle speed and the particle position according to a calculation formula of a nonlinear dynamic inertia weight coefficient and a learning factor calculation formula on the basis of a Particle Swarm Optimization (PSO);
6) calculating and obtaining a current particle fitness value, comparing the current particle fitness value with the historical optimal individual particle fitness value, and updating the optimal particle solution if the current particle fitness value is more optimal; otherwise, maintaining the individual optimal fitness value;
7) if the fitness value of the current particle is better than the optimal solution of the population, updating the optimal solution of the population; otherwise, maintaining the optimal solution of the group unchanged;
8) and (3) returning to the step (3) until the set iteration times are reached, substituting the optimal solution input weight omega and the hidden layer unit bias threshold b into the ELM, and constructing a bus peak load prediction model.
The expression of the nonlinear dynamic inertia weight coefficient is as follows:
Figure BDA0002256772960000041
in the formula: w is the inertial weight, wminIs the minimum value of the inertial weight, wmaxIs the maximum value of the inertial weight; f is the particle fitness; f. ofavgIs the average fitness; f. ofminIs the minimum fitness.
The expression of the learning factor is as follows:
Figure BDA0002256772960000042
in the formula: c. C1And c2As a learning factor, c1sIs c1Initial value of c2sIs c2Initial value of c1cIs c1End value of c2cIs c2A final value of (d); iter is the number of current iterations; itermaxIs the number of total iterations.
The number of iterations is 200.
Through the design scheme, the invention can bring the following beneficial effects:
the method takes the characteristic importance result of the conditional mutual information on the characteristic to be selected in the original characteristic set as the basis, combines IPSO-ELM as a predictor, carries out forward characteristic selection, determines the optimal characteristic set of bus peak load prediction, reduces the influence of characteristic redundancy on prediction precision during bus peak load prediction, effectively improves the prediction precision of different buses by respectively constructing optimal prediction models for different buses, and introduces the combination of IPSO-ELM and a linear method to carry out peak load prediction under different scenes, thereby meeting the prediction requirement under small sample or no sample scenes.
Drawings
The invention is further described with reference to the following figures and detailed description:
FIG. 1 is a scatter diagram and a fitting graph of the mean value of the peak load of the busbar 1 under extreme high temperature conditions in an embodiment of the method for predicting the peak load of the busbar taking into account the influence of complex weather.
FIG. 2 is a scatter diagram and a fitting graph of the mean of the peak load of the busbar 2 under extreme high temperature conditions in an embodiment of the method for predicting the peak load of the busbar taking into account the influence of complex weather.
FIG. 3 is a scatter diagram and a fitting graph of the bus 1 peak load average under extreme low temperature conditions in an embodiment of the bus peak load prediction method taking into account the effects of complex weather.
FIG. 4 is a scatter diagram and a fitting graph of the bus 2 peak load average under extreme low temperature conditions in an embodiment of the bus peak load prediction method taking into account the effects of complex weather.
FIG. 5 is an analysis diagram of the importance of the peak load characteristics of the bus 1 in the embodiment of the bus peak load prediction method considering the influence of complex weather.
FIG. 6 is an analysis diagram of the importance of the peak load characteristics of the bus 2 in the embodiment of the bus peak load prediction method considering the influence of complex weather.
FIG. 7 is a diagram of selecting an optimal feature set by combining mutual condition information with different predictors in an embodiment of a bus peak load prediction method considering complex weather influence according to the present invention.
FIG. 8 is a comparison graph of bus peak load predictions for a directed model and a unified model in an embodiment of the present invention accounting for complex weather effects.
FIG. 9 is a comparison graph of peak load bus prediction for three methods in an embodiment of the present invention accounting for the effects of complex weather.
Detailed Description
As shown in the figure, the bus peak load prediction method considering the influence of complex weather comprises the following steps:
1. bus peak load prediction original feature set construction
(1) Raw feature set for bus peak load prediction
The peak load of the bus can fluctuate due to various factors such as natural weather and society. Through analysis of complex meteorological factors and relevant literature research, the constructed original feature set is shown in table 1.
TABLE 1 bus Peak load prediction primitive feature set
Note:
1)FTindicates the predicted daily temperature peak, FTaveRepresents the average value of the temperature on the day of prediction; fT(max,d-1)Indicates the temperature peak of the day before the predicted day, FT(ave,d-1)Represents the average temperature of the day before the predicted day; fAIndicating the predicted timePressure of (F)HHumidity F representing predicted timeWWind direction F indicating predicted timeW1Wind speed, F, at the moment of predictionA、FH、FWAnd FW1The equal are all meteorological features;
2)FG1representing the longitude, F, of the busbar to be predictedG2Representing the latitude of the bus to be predicted; fT、FTave、FT(max,d-1)、FT(ave,d-1)、FA、FH、FW、FW1、FG1And FG2The equal are all natural meteorological features;
3)FD1to FD7The day of the week; fJ1Marking the working day; fJ2Is a non-working day; fH1Representing holidays; fH2Indicating a normal day; fD1To FD7、FJ1、FJ2、FH1And FH2Are all social factors;
4)FL(max,d-1)represents the peak load on the bus, F, one day before the predicted dayL(max,d-2)Representing the bus peak load of two days before the forecast day, and so on; fL(t-15)Representing the bus load 15 minutes before the peak load of the day before the day to be measured, FL(t-30)The bus load 30 minutes before the bus peak load of the day before the day to be measured is shown, and so on.
(2) Conditional mutual information
In the prediction of the peak load of the bus, D is set as an original characteristic set containing various factors such as natural weather, society and the like; q is the actually measured bus peak load value; the Z set is the selected feature. The mutual information between D and Q is:
Figure BDA0002256772960000061
in formula (1), F (D; Q) represents mutual information between D and Q, P (D) is a marginal density function of D, P (Q) is a marginal density function of Q, and P (D, Q) is a joint probability density of D and Q.
Under the condition of the known Z set, the condition mutual information of the set D and the actually measured bus peak load value Q is as follows:
Figure BDA0002256772960000062
in formula (2), F (D, Q | Z) represents conditional mutual information between D and Q under Z condition, and P (D | Z) is a probability density function of D, Q under Z condition; p (qz) is the probability density function of Q under Z condition; p (d, q | Z) is the joint probability density function of D, Q under Z; p (d, q, z) is a joint probability density function of D, Q, Z.
The bus peak load influence factors are numerous, if all influence factors are considered, information redundancy is caused, and the prediction accuracy of the bus peak load is low. In order to improve the prediction accuracy of the peak load of the bus bar, the relevance between each influence factor and different peak load of the bus bar is analyzed through condition mutual information, and the feature importance degree sequence in the original feature set is obtained.
2. Bus peak load prediction model construction
(2.1) extreme learning machine principle based on improved particle swarm optimization
A neural network and other predictors are trained by limited bus peak load data, and an ideal bus peak load prediction model is difficult to obtain. Therefore, a bus peak load predictor is built by applying an extreme learning mechanism suitable for small sample training. In order to avoid the influence of improper parameter selection on the prediction effect, an improved particle swarm optimization is applied to optimize the input weight and the threshold of the extreme learning machine and a bus peak load predictor so as to further improve the prediction precision.
(2.1.1) extreme learning machine
Is provided with N samples
Figure BDA0002256772960000071
Wherein the input data is xi＝[xi1,xi2,…,xin]T∈RnTarget output value of ti＝[ti1,ti2,…,tim]T∈Rm. Then the ELM extreme learning machine network model of the single hidden layer neural network with L number of hidden layer nodes can be expressed as
In the formula ojRepresenting a network output value; g represents an activation function; omegaiβ as input weightiIs the output weight; biIs the bias of the ith hidden layer unit; x is the number ofjDenotes xiThe data of (1).
Without error, the activation function approaches an arbitrary number N of samples indefinitely, i.e.
Figure BDA0002256772960000073
According to the formula (4), a
Wherein g represents an activation function; omegaiβ as input weightiIs the output weight; biIs the bias of the ith hidden layer unit; t is tjRepresenting the network output value when the activation function can approach any N samples with zero error;
the matrix form of N equations in the formula (5) is
Hβ＝T (6)
Wherein the content of the first and second substances,
Figure BDA0002256772960000075
Figure BDA0002256772960000081
h denotes the hidden layer node output, T is the desired output, β denotes the weight matrix between the output layer and the hidden layer.
By obtainingAnd
Figure BDA0002256772960000083
realizes the ELM training of the single hidden layer, so that
In the formula (I), the compound is shown in the specification,
Figure BDA0002256772960000085
representation βiAn optimal solution;
Figure BDA0002256772960000086
represents omegaiThe optimal solution of (2);
Figure BDA0002256772960000087
denotes biThe optimal solution of (2); h represents hidden layer node output; t is the desired output;
the minimum loss function equivalent to the formula (9) is
Figure BDA0002256772960000088
Wherein E represents a minimum loss function; g represents an activation function; omegaiβ as input weightiIs the output weight; biIs the offset of the ith hidden layer unit.
In the ELM, once the input weight omega and the bias threshold b of the hidden layer unit are randomly determined, the only hidden layer node output H can be obtained. Thus, the ELM structure is determined.
(2.2) improved particle swarm optimization based on extreme learning machine parameter optimization of improved particle swarm (2.2.1)
In order to optimize the problem that the particle swarm algorithm is easy to fluctuate around the global optimal solution, the traditional particle swarm is improved through inertial weight, and the expression is as follows:
in formula (11): w represents the weight of the inertia, and,wminis the minimum value of the inertial weight, wmaxIs the maximum value of the inertial weight; f is the particle fitness; f. ofavgIs the average fitness; f. ofminIs the minimum fitness.
In order to enable the particle swarm optimization algorithm to quickly determine the global optimal solution, the traditional particle swarm is dynamically adjusted according to the formula (12), namely
Figure BDA0002256772960000091
In (12): c. C1And c2Are all learning factors, c1sIs c1Initial value of c2sIs c2Initial value of c1cIs c1A final value of (d); c. C2cIs c2A final value of (d); iter is the number of current iterations; itermaxIs the number of total iterations.
(2.2.2) ELM parameter optimization based on improved particle swarm
Taking the optimal process of constructing the bus peak load prediction model by the original feature set as an example, the IPSO optimization ELM process is shown as follows:
1) constructing a busbar peak load prediction ELM based on an original characteristic set, and randomly generating an input weight omega and a hidden layer unit bias threshold b of the ELM;
2) determining data of an original feature set required by input bus load peak value prediction, and setting iteration times;
3) carrying out normalization processing on data of the bus peak load prediction original feature set;
4) obtaining a fitness value according to the average absolute percentage error of the bus peak load prediction of the ELM, and determining the optimal fitness value of the current individual and group;
5) updating the particle speed and position based on the conventional PSO according to the formulas (11) and (12);
6) firstly, calculating a current particle fitness value, then comparing the current particle fitness value with a historical optimal value, and if the current particle fitness value is more optimal, updating a particle optimal solution; otherwise, maintaining the individual optimal fitness value;
7) if the fitness value of the current particle is better than the optimal solution of the population, updating the optimal solution of the population; otherwise, maintaining the optimal solution of the group unchanged;
8) and if the iteration times are not reached, returning to 3), otherwise, substituting the optimal solution omega and b into the ELM to construct a bus peak load prediction model.
In the postamble feature selection link, the method is adopted for constructing the targeted optimal ELM predictor for different dimensional feature sets.
3. Bus peak load prediction model without training sample
Predictors such as ELM are suitable for small sample prediction, but construction of the predictors still depends on historical samples. If extreme temperatures (extremely high temperatures and extremely low temperatures) that have not occurred in the history occur at the predicted day, the predicted effect cannot be guaranteed. In order to improve the prediction precision of the bus peak load of the untrained sample, the linear model is introduced to independently predict the bus peak load of the untrained sample in extremely high and low temperature scenes.
Example (b):
and carrying out statistical analysis on peak loads of the plurality of buses under extreme meteorological conditions of extreme high temperature and extreme low temperature in 2018 of a certain city in northeast China. In this embodiment, the extreme high temperature is set to be the highest daily temperature higher than 30 degrees, and the extreme low temperature is set to be the lowest daily temperature lower than-20 degrees. According to the statistical result, the peak load of the anemarrhena line and the highest temperature have an obvious linear relation, so that the least square method is used for carrying out linear fitting on the peak load of the extreme temperature and the bus.
Fig. 1 to 4 respectively show a scatter diagram of the average value of the peak load of the bus and a corresponding fitting curve under extreme meteorological conditions of extreme high temperature and extreme low temperature of the bus 1 and the bus 2 in 2018 of a certain northeast city. As can be seen from fig. 1 to 4 and the linear model of the peak load prediction of the two bus bars under extreme meteorological conditions, the extreme meteorological conditions have different effects on the peak load of different bus bars. Therefore, in extreme meteorological conditions, for different buses, targeted modeling analysis is required, and the prediction accuracy of the peak load of the bus is improved.
The linear model of the peak load prediction of the bus 1 and the bus 2 under the extreme meteorological condition, which is obtained by linear fitting of the extreme temperature and the peak load of the bus by using the least square method, is as follows:
under the extreme high temperature condition, the linear model for predicting the peak load of the bus 1 is as follows:
Figure BDA0002256772960000101
under the extreme low temperature condition, the linear model for predicting the peak load of the bus 1 is as follows:
Figure BDA0002256772960000102
and obtaining a bus 2 linear prediction model by adopting the same method.
Under the extreme high temperature condition, the linear model for predicting the peak load of the bus 2 is as follows:
Figure BDA0002256772960000103
under the extreme low temperature condition, the linear model for predicting the peak load of the bus 2 is as follows:
Figure BDA0002256772960000104
in the formula:and
Figure BDA0002256772960000106
peak loads of the bus 1 and the bus 2 under an extremely high temperature condition are respectively obtained;
Figure BDA0002256772960000107
and
Figure BDA0002256772960000108
peak loads of the bus 1 and the bus 2 under an extreme low temperature condition are respectively obtained; t is the extreme temperature.
In the research, 2018 peak load and meteorological information data of a certain northeast city bus in the year of northeast are applied, wherein data in months 1, 4, 8 and 10 are used as a verification set, data in month 7 are used as a test set, and the rest 7 months are used as training sets, so that targeted feature selection is carried out on different buses. To prove the advancement of the new method, experiments were compared with IPSO-ELM and BP neural network (BPNN). The model prediction effect is evaluated by Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE), and the index calculation method is as follows:
Figure BDA0002256772960000112
in formulae (17) and (18) YiThe measured value of the peak load of the bus is taken as the measured value;predicting a bus peak load value; and N is the number of prediction samples.
(1) Selecting analysis based on condition mutual information characteristic
Fig. 5 and 6 are graphs for analyzing the feature importance of peak loads of different bus bars, and it can be seen from the graphs that there is a difference in correlation between the peak loads and the features of different bus bars. FIG. 7 shows the number of features and the corresponding MAPE value included in the optimal feature set after the CMI selects the optimal feature set of the bus 1 by respectively combining IPSO-ELM, BPNN and ELM; table 2 gives the optimal feature set containing features. As can be seen from fig. 7, in the prediction of the peak load of the bus bar, the prediction error is different depending on the selected feature set. When IPSO-ELM, BPNN and ELM are used as predictors respectively, and the characteristic dimension of the characteristic subset is 25, 28 and 34 respectively, the error of the bus peak load prediction is minimum. And as can be seen from fig. 7, according to the optimal feature set, when BPNN, ELM, and IPSO-ELM are respectively used as predictors, MAPE values predicted by the bus 1 peak load are respectively 4.54%, 3.75%, and 3.04%, and among the three predictors, the MAPE value predicted by the IPSO-ELM bus peak load is the smallest, which indicates the advantage of high IPSO-ELM prediction accuracy. Similarly, the optimal feature subset for bus 2 may be determined.
TABLE 2 optimal feature set
Figure BDA0002256772960000114
Figure BDA0002256772960000121
(2) Analysis of bus peak load prediction results
In order to prove the advantage of high precision of the proposed bus peak load prediction method, peak load prediction results of a bus 1 and a bus 2 in northeast certain city in 7 months in 2018 are listed. FIG. 8 shows the bus peak load prediction results of two bus lines for the targeted modeling and the prediction model construction using the original feature set; table 3 corresponds to the bus peak load prediction error in fig. 8. Under the optimal model, the predicted MAPE of the peak load of the two buses is 3.04% and 2.98% respectively; when modeling the original feature set, the MAPE of the two bus peak loads is 3.89% and 4.01%, respectively. Through comparison, targeted feature selection is carried out on different buses, a bus peak load prediction model is established in a targeted mode according to a feature selection result, and prediction accuracy is high.
Fig. 9 shows the prediction results of the peak load of the busbar in 7 months for the busbar 1 in northeast city in 2018 when IPSO-ELM, ELM and BPNN are respectively used as the busbar peak load predictor, and table 4 corresponds to the prediction errors of the three busbar peak load prediction methods in fig. 9. As can be seen from Table 4, the MAPE predicted for the peak load of the busbar using BPNN, ELM and IPSO-ELM is 4.23%, 3.93% and 3.04%, respectively, and the MAPE predicted for the peak load of the IPSO-ELM busbar is the smallest. Therefore, the method provided by the invention has high prediction precision for the problem of severe fluctuation of the peak load of the bus caused by complex meteorological factors.
TABLE 3 bus Peak load prediction error
Figure BDA0002256772960000122
TABLE 4 Peak load prediction error for three methods
Figure BDA0002256772960000123
In order to further verify the prediction effectiveness of the new method under the scene of no training sample with extremely high temperature and extremely low temperature, the highest and lowest temperature days in 2018 years in a certain city in northeast are removed from the training samples, the highest and lowest temperature days are used as days to be predicted, and historical data before the temperature days are used as training data. The peak loads of the different buses on the day are predicted by the improved particle swarm optimization extreme learning machine and the linear model respectively, and the results are shown in table 5. As can be seen from Table 5, the linear model has higher accuracy when predicting the bus peak load of the non-historical temperature sample, and the new method is proved to be capable of effectively avoiding the peak load prediction error caused by the lack of historical data under the extreme temperature of the non-training sample. The new method has better applicability.
TABLE 5 extreme weather day bus peak load prediction
Figure BDA0002256772960000131
In conclusion, the bus peak load prediction feature selection is carried out on the basis of the conditional mutual information value, the influence of feature redundancy on prediction precision in bus peak load prediction is reduced, the optimal prediction models are respectively constructed for different buses, the prediction precision of different buses is effectively improved, the improved particle swarm optimization extreme learning machine is introduced to be combined with a linear method, peak load prediction is carried out in different scenes, and the prediction requirement in small-sample or no-sample scenes is met.

Claims (8)

1. The bus peak load prediction method considering the complex meteorological influence is characterized by comprising the following steps: comprises the following steps which are sequentially carried out,
the method comprises the steps that firstly, based on the influence of natural weather and social factors on the bus peak load prediction precision, an original characteristic set for bus peak load prediction is constructed according to the natural weather and social factors, and the correlation between each characteristic in the original characteristic set and the bus peak load is analyzed through Conditional Mutual Information (CMI), so that characteristic importance degree ranking is obtained;
secondly, optimizing the input weight and the threshold value of the Extreme Learning Machine by improving a Particle Swarm algorithm to obtain an Improved Particle Swarm optimized Extreme Learning Machine (IPSO-ELM), then carrying out targeted forward feature selection by taking the IPSO-ELM prediction precision as a decision variable according to the feature importance ranking result obtained in the first step to obtain an optimal feature subset of bus peak load prediction, and retraining the IPSO-ELM according to the obtained optimal feature subset to obtain an optimal bus peak load prediction model;
step three, substituting test set data in the historical data within a set time period into the optimal bus peak load prediction model obtained in the step two to obtain a predicted value of the bus peak load;
respectively counting the peak load of the bus under the extremely high temperature condition and the peak load of the bus under the extremely low temperature condition through a linear model, and obtaining that the peak load of the bus has linear relations with the extremely high temperature and the extremely low temperature according to a statistical result;
and fifthly, performing linear fitting on the extreme high temperature and the extreme low temperature and the peak load of the bus by using a least square method, and respectively obtaining the peak load prediction linear models of the bus 1 and the bus 2 under the extreme temperature condition as follows:
under the extreme high temperature condition, the linear model for predicting the peak load of the bus 1 is as follows:
under the extreme low temperature condition, the linear model for predicting the peak load of the bus 1 is as follows:
under the extreme high temperature condition, the linear model for predicting the peak load of the bus 2 is as follows:
Figure FDA0002256772950000013
under the extreme low temperature condition, the linear model for predicting the peak load of the bus 2 is as follows:
Figure FDA0002256772950000021
in the formula:
Figure FDA0002256772950000022
peak load of the bus 1 under extreme high temperature conditions;
Figure FDA0002256772950000023
peak load of the bus 2 under extreme high temperature conditions;
Figure FDA0002256772950000024
peak load of the bus 1 under extreme low temperature conditions;
Figure FDA0002256772950000025
peak load of the bus 2 under extreme low temperature conditions; t is the extreme temperature.
2. The method of claim 1, wherein the method comprises: the natural weather includes longitude, latitude, temperature, air pressure, humidity, wind direction and wind speed.
3. The method of claim 1, wherein the method comprises: the social factors include date and holidays.
4. The method of claim 1, wherein the method comprises: the prediction precision in the second step is the average Absolute percentage error (MAPE) between the predicted value and the actual value, and the calculation formula of the MAPE is as follows:
Figure FDA0002256772950000026
in the formula, YiThe measured value of the peak load of the bus is taken as the measured value;
Figure FDA0002256772950000027
predicting a bus peak load value; and N is the number of prediction samples.
5. The method of claim 1, wherein the method comprises: the specific method for obtaining the IPSO-ELM by optimizing the input right and the threshold of the extreme learning machine through the improved particle swarm optimization in the step two comprises the following steps:
1) constructing an ELM model for predicting the peak load of the bus bar based on the original characteristic set, and randomly generating an input weight omega of the ELM and a hidden layer unit bias threshold b;
2) determining data of an original feature set required by input bus peak load prediction, and setting iteration times;
3) predicting original characteristic set data aiming at the peak load of the bus, and carrying out normalization processing;
4) obtaining a fitness value according to an average absolute percentage error value of bus peak load prediction of the ELM, and obtaining a current individual fitness value and a group optimal fitness value;
5) updating the particle speed and the particle position according to a calculation formula of a nonlinear dynamic inertia weight coefficient and a learning factor calculation formula on the basis of a Particle Swarm Optimization (PSO);
6) calculating and obtaining a current particle fitness value, comparing the current particle fitness value with the historical optimal individual particle fitness value, and updating the optimal particle solution if the current particle fitness value is more optimal; otherwise, maintaining the individual optimal fitness value;
7) if the fitness value of the current particle is better than the optimal solution of the population, updating the optimal solution of the population; otherwise, maintaining the optimal solution of the group unchanged;
8) and (3) returning to the step (3) until the set iteration times are reached, substituting the optimal solution input weight omega and the hidden layer unit bias threshold b into the ELM, and constructing a bus peak load prediction model.
6. The method of claim 5, wherein the method comprises: the expression of the nonlinear dynamic inertia weight coefficient is as follows:
Figure FDA0002256772950000031
in the formula: w is the inertial weight, wminIs the minimum value of the inertial weight, wmaxIs the maximum value of the inertial weight; f is the particle fitness; f. ofavgIs the average fitness; f. ofminIs the minimum fitness.
7. The method of claim 5, wherein the method comprises: the expression of the learning factor is as follows:
Figure FDA0002256772950000032
in the formula: c. C1And c2As a learning factor, c1sIs c1Initial value of c2sIs c2Initial value of c1cIs c1End value of c2cIs c2A final value of (d); iter is the number of current iterations; itermaxIs the number of total iterations.
8. The method of claim 5, wherein the method comprises: the number of iterations is 200.
CN201911056853.6A 2019-10-31 2019-10-31 Bus peak load prediction method considering complex weather influence Active CN110807508B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911056853.6A CN110807508B (en) 2019-10-31 2019-10-31 Bus peak load prediction method considering complex weather influence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911056853.6A CN110807508B (en) 2019-10-31 2019-10-31 Bus peak load prediction method considering complex weather influence

Publications (2)

Publication Number Publication Date
CN110807508A true CN110807508A (en) 2020-02-18
CN110807508B CN110807508B (en) 2023-06-09

Family

ID=69489951

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911056853.6A Active CN110807508B (en) 2019-10-31 2019-10-31 Bus peak load prediction method considering complex weather influence

Country Status (1)

Country Link
CN (1) CN110807508B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112308169A (en) * 2020-11-10 2021-02-02 浙江大学 Effluent quality prediction method based on improved online sequence extreme learning machine
CN112348235A (en) * 2020-10-21 2021-02-09 国网安徽省电力有限公司阜阳供电公司 Wind-solar bus load self-adaptive prediction method and device and computer equipment
CN116432821A (en) * 2023-03-08 2023-07-14 河北省气象服务中心(河北省气象影视中心) Meteorological condition-based next-day power load peak prediction method

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060030972A1 (en) * 2003-06-27 2006-02-09 Robert Schlueter Voltage collapse diagnostic and ATC system
CN102799953A (en) * 2012-07-16 2012-11-28 安徽省电力公司池州供电公司 Bus load prediction method based on stacked generalization training strategy
CN106485262A (en) * 2016-09-09 2017-03-08 国网山西省电力公司晋城供电公司 A kind of bus load Forecasting Methodology
CN106855957A (en) * 2015-12-09 2017-06-16 四川大学 Factory's bus load prediction based on similar day and least square method supporting vector machine
CN107403239A (en) * 2017-07-25 2017-11-28 南京工程学院 A kind of parameters analysis method for being used for control device in power system
CN107480815A (en) * 2017-08-03 2017-12-15 国网河北省电力公司保定供电分公司 A kind of power system taiwan area load forecasting method

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060030972A1 (en) * 2003-06-27 2006-02-09 Robert Schlueter Voltage collapse diagnostic and ATC system
CN102799953A (en) * 2012-07-16 2012-11-28 安徽省电力公司池州供电公司 Bus load prediction method based on stacked generalization training strategy
CN106855957A (en) * 2015-12-09 2017-06-16 四川大学 Factory's bus load prediction based on similar day and least square method supporting vector machine
CN106485262A (en) * 2016-09-09 2017-03-08 国网山西省电力公司晋城供电公司 A kind of bus load Forecasting Methodology
WO2018045642A1 (en) * 2016-09-09 2018-03-15 国网山西省电力公司晋城供电公司 A bus bar load forecasting method
CN107403239A (en) * 2017-07-25 2017-11-28 南京工程学院 A kind of parameters analysis method for being used for control device in power system
CN107480815A (en) * 2017-08-03 2017-12-15 国网河北省电力公司保定供电分公司 A kind of power system taiwan area load forecasting method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
熊图: "基于特征排序与深度学习的母线负荷预测方法" *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112348235A (en) * 2020-10-21 2021-02-09 国网安徽省电力有限公司阜阳供电公司 Wind-solar bus load self-adaptive prediction method and device and computer equipment
CN112308169A (en) * 2020-11-10 2021-02-02 浙江大学 Effluent quality prediction method based on improved online sequence extreme learning machine
CN112308169B (en) * 2020-11-10 2022-05-03 浙江大学 Effluent quality prediction method based on improved online sequence extreme learning machine
CN116432821A (en) * 2023-03-08 2023-07-14 河北省气象服务中心(河北省气象影视中心) Meteorological condition-based next-day power load peak prediction method
CN116432821B (en) * 2023-03-08 2023-09-26 河北省气象服务中心(河北省气象影视中心) Meteorological condition-based next-day power load peak prediction method

Also Published As

Publication number Publication date
CN110807508B (en) 2023-06-09

Similar Documents

Publication Publication Date Title
Li et al. A hybrid model based on synchronous optimisation for multi-step short-term wind speed forecasting
CN108876054B (en) Short-term power load prediction method based on improved genetic algorithm optimization extreme learning machine
CN111260136A (en) Building short-term load prediction method based on ARIMA-LSTM combined model
CN108009674A (en) Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks
CN110705743B (en) New energy consumption electric quantity prediction method based on long-term and short-term memory neural network
Lin et al. High-precision forecast using grey models
CN110807508A (en) Bus peak load prediction method considering complex meteorological influence
CN110059867B (en) Wind speed prediction method combining SWLSTM and GPR
CN111861013B (en) Power load prediction method and device
CN112381673B (en) Park electricity utilization information analysis method and device based on digital twin
Li et al. A k-nearest neighbor locally weighted regression method for short-term traffic flow forecasting
CN115600640A (en) Power load prediction method based on decomposition network
CN113449919B (en) Power consumption prediction method and system based on feature and trend perception
CN114418234A (en) Power battery manufacturing capacity online prediction method based on reinforcement learning
Usman et al. Data analytics for short term price and load forecasting in smart grids using enhanced recurrent neural network
CN113435595A (en) Two-stage optimization method for extreme learning machine network parameters based on natural evolution strategy
CN112307672A (en) BP neural network short-term wind power prediction method based on cuckoo algorithm optimization
CN116151469A (en) Model for forecasting air quality
CN114234392B (en) Air conditioner load fine prediction method based on improved PSO-LSTM
CN115481788A (en) Load prediction method and system for phase change energy storage system
CN114784795A (en) Wind power prediction method and device, electronic equipment and storage medium
CN115310355A (en) Multi-energy coupling-considered multi-load prediction method and system for comprehensive energy system
CN115081323A (en) Method for solving multi-objective constrained optimization problem and storage medium thereof
CN113762591A (en) Short-term electric quantity prediction method and system based on GRU and multi-core SVM counterstudy
Wu et al. Short-Term Electric Load Forecasting Model based on PSO-BP

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant