CN109118020A - A kind of subway station energy consumption short term prediction method and its forecasting system - Google Patents
A kind of subway station energy consumption short term prediction method and its forecasting system Download PDFInfo
- Publication number
- CN109118020A CN109118020A CN201811038101.2A CN201811038101A CN109118020A CN 109118020 A CN109118020 A CN 109118020A CN 201811038101 A CN201811038101 A CN 201811038101A CN 109118020 A CN109118020 A CN 109118020A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- subway station
- aco
- data
- station
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000005265 energy consumption Methods 0.000 title claims abstract description 105
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013528 artificial neural network Methods 0.000 claims abstract description 32
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000004378 air conditioning Methods 0.000 claims abstract description 21
- 238000009423 ventilation Methods 0.000 claims abstract description 19
- 238000012360 testing method Methods 0.000 claims abstract description 15
- 238000010276 construction Methods 0.000 claims abstract description 11
- 102000008297 Nuclear Matrix-Associated Proteins Human genes 0.000 claims description 9
- 108010035916 Nuclear Matrix-Associated Proteins Proteins 0.000 claims description 9
- 210000000349 chromosome Anatomy 0.000 claims description 9
- 210000000299 nuclear matrix Anatomy 0.000 claims description 9
- 238000005286 illumination Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 8
- 241000257303 Hymenoptera Species 0.000 claims description 6
- 235000013305 food Nutrition 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 230000008775 paternal effect Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000008774 maternal effect Effects 0.000 claims description 2
- 230000035772 mutation Effects 0.000 claims description 2
- 210000002569 neuron Anatomy 0.000 claims description 2
- 230000007704 transition Effects 0.000 claims description 2
- 230000007547 defect Effects 0.000 abstract description 2
- 238000003062 neural network model Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 230000005611 electricity Effects 0.000 description 3
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003750 conditioning effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 229910052742 iron Inorganic materials 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000012795 verification 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
-
- G06Q50/40—
Abstract
The invention discloses a kind of subway station energy consumption short term prediction methods, include the following steps, determining influence factor related with subway station subitem energy consumption determines major influence factors and acquired in influence factor；Subway station subitem energy consumption data library is established, is normalized together with the collected major influence factors, the data after normalized is finally divided into training data and test data again；Construction G-ACO-BP Network Prediction Model；Using training data training G-ACO-BP model, the error of reality output and target output is calculated；Test data is input in the G-ACO-BP model after training and is tested, ventilation and air conditioning energy consumption, power energy consumption, the prediction result of lighting energy consumption are obtained.The invention has the advantages that: first is that the subitem short-term forecast of subway station energy consumption more acurrate can be effectively realized；Second is that brought defect when BP neural network training can be effectively prevented from, reach the intelligent optimizing purpose of neural network model.
Description
Technical field
The present invention relates to the technical fields of energy consumption short-term forecast, more particularly to one kind to be based on the subway station KPCA-G-ACO energy
Consume short term prediction method.
Background technique
Recently as the continuous propulsion of Chinese city urban rail transit construction process, domestic city rail traffic distance travelled
It increases rapidly, by December 31st, 2016, inland of China had 29 cities and has 127 cities built up and formally runed
City's rail line, total kilometrage reach 3838km.The fast development of various regions urban track traffic is preferably facilitating people's traffic
Also various regions city energy consumption, or even the main body as urban energy consumption are greatly increased while trip.Therefore, how effectively to control
System and reduce urban track traffic energy consumption, save operation cost be the emphasis paid close attention to the most of each city underground operation management person it
One.All based on subway, the main energy consumption of metro operation is electric energy consumption for various regions urban track traffic, these energy consumptions have and can be divided into
Vehicle operation energy consumption and station energy consumption two large divisions.It is analyzed according to current operating line energy consumption statistic data, vehicle operation energy consumption
Account for the 50%~60% of total electricity consumption；Station, base energy consumption account for the 40%~50% of total electricity consumption.It is dynamic in the energy consumption of station
Power, illumination, ventilation and air conditioning system account for 90% or more of station electricity consumption.
Summary of the invention
The purpose of this section is to summarize some aspects of the embodiment of the present invention and briefly introduce some preferable implementations
Example.It may do a little simplified or be omitted to avoid our department is made in this section and the description of the application and the title of the invention
Point, the purpose of abstract of description and denomination of invention it is fuzzy, and this simplification or omit and cannot be used for limiting the scope of the invention.
In view of above-mentioned existing problem, the present invention is proposed.
Therefore, it is an object of the present invention to provide one kind to be based on the subway station KPCA-G-ACO energy consumption short term prediction method, realizes
To in subway station power, illumination, ventilation and air conditioning this three sport energy consumption carry out subitem forecast analysis, to find out subway station
Energy-saving potential.
In order to solve the above technical problems, the invention provides the following technical scheme: a kind of subway station energy consumption short-term forecast side
Method includes the following steps determining influence factor related with subway station subitem energy consumption, main influence is determined in influence factor
Factor simultaneously acquires, and pre-processes to the major influence factors of acquisition；It is carried out according to subway station subitem energy consumption historical data
Comprehensive analysis is established subway station subitem energy consumption data library, is normalized together with the collected major influence factors
Data after normalized are finally divided into training data and test data again by processing；Construction G-ACO-BP neural network forecast
Model further includes carrying out optimal settings to the parameter of G-ACO-BP Network Prediction Model, G-ACO-BP Network Prediction Model
Parameter includes cross and variation probability, information heuristic factor and BP learning rate；Utilize training data training G-ACO-BP model, meter
Calculate the error of reality output and target output；Test data is input in the G-ACO-BP model after training and is tested, is obtained
To ventilation and air conditioning energy consumption, power energy consumption, the prediction result of lighting energy consumption.
A kind of preferred embodiment as energy consumption short term prediction method in subway station of the present invention, in which: described and ground
Subitem energy consumption related influence factor in iron car station includes the station volume of the flow of passengers (x1), daily 24 integral point moment (x2), festivals or holidays
(x3), in station by when mean temperature (x4), in station by when average relative humidity (x5), season (x6), weather characteristics value (x7)、
Station entrance-exit quantity (x8), start columns (x9), station average illumination (x10), the scale (x at station11) and station spacing
(x12) this 12 influence factors.
A kind of preferred embodiment as energy consumption short term prediction method in subway station of the present invention, in which: described in shadow
It determines that major influence factors are to determine major influence factors in 12 influence factors with KPCA in the factor of sound, further includes following step
Suddenly, for 12 selected predicted characteristics x=[x1,x2,…,x12], according to formula (1) to the influence factor of subway station energy consumption
It is standardized:
Wherein, xiIt (k) is k-th of sampled value of i-th of influence factor, xmax(i) and xminIt (i) is respectively i-th of influence
The maximum value and minimum value of all sampled points of factor, TiIt (k) is the target data after standardization；Nuclear moment is sought according to formula (2)
Battle array K, is realized initial data using Radial basis kernel function by data space map to feature space:
Utilize centralization nuclear matrix KCTo correct nuclear matrix K, correction formula are as follows:
KC=K-lNK-KlN+lNKlN(3)
Wherein, KCFor the matrix of N × N, each element is for 1/N；Calculating matrix KCCharacteristic value, corresponding feature
Vector is λ1,λ2,...,λ12, variance v1,v2,...,v12, by characteristic value descending sort, feature vector makees corresponding adjust
It is whole；Using Schmidt process, orthogonalization and unitization feature vector, obtained feature vector are a1,a2,...,
a12；Calculate the contribution rate of accumulative total r of feature1,r2,...,r12, p is required according to given contribution rate, if rt> p, then t before choosing
A principal component a1,...,at(t < 12) is major influence factors.
A kind of preferred embodiment as energy consumption short term prediction method in subway station of the present invention, in which: the construction
G-ACO-BP Network Prediction Model be any neural network containing n-layer input, n-layer output, wherein including input layer, implicit
Layer and output layer, and the characteristic of each node is Sigmoid type function.
A kind of preferred embodiment as energy consumption short term prediction method in subway station of the present invention, in which: the construction
G-ACO-BP Network Prediction Model is further comprising the steps of, and a. carries out " digitlization " coding, initialization population to the potential solution of problem；
B. it starts the cycle over, assesses fitness individual corresponding to every chromosome；C. higher in accordance with fitness, the bigger original of select probability
Then, select two individuals as paternal and maternal from population；D. the chromosome for selecting parent both sides, is intersected, and son is generated
Generation；E. it makes a variation to the chromosome of filial generation；F. c, d, step e, until generating several optimization solutions are repeated；G. entire net is initialized
Network enables time t=0, cycle-index Nc=0, ifFor maximum cycle, the information of each element in each set is enabled
AmountAndWhole ants are placed in the nest of ant；H. start all ants, for set
Ant k (k=1,2 ..., h) calculates state transition probability according to formula (4):
I. h step is repeated, until all ant colonies all arrive at food source；J. t, N are assigned t+mc+ 1 assigns Nc, recycle
The output valve and error of weight and threshold calculations neural network that each ant is selected, record current optimal solution.By m time
Unit, ant reach food source from its nest, and the information content on each path updates according to the following formula:
Wherein, ekThe one group of weight and threshold value that k-th of ant is selected are as the output of BP neural network weight and threshold value
Error is defined as follows shown in formula:
ek=| O-Oq| (8)
Wherein, O is the real output value of BP neural network, OqFor BP neural network desired output；K. using verifying sample
This tests to the generalization ability of trained neural network, if ant colony all converges to a paths, i.e. optimal path
Or cycle-indexThen circulation terminates, and exports corresponding calculated result；Otherwise h step is jumped to continue to grasp
Make；The weight obtained under optimal path under G-ACO algorithm and threshold value are updated to BP neural network normalization by the l. training of network
Learning sample after processing is trained and tests using the weight and threshold value that optimize under BP neural network.
A kind of preferred embodiment as energy consumption short term prediction method in subway station of the present invention, in which: to G-ACO-
The parameter that the parameter of BP Network Prediction Model carries out optimal settings includes, neuron number be configured to input layer, hidden layer,
Output layer is respectively 3,11,3；Training algorithm using G-ACO algorithm, Population Size 20, crossover probability 0.5, mutation probability 0.5,
Ant number 20, information heuristic factor 1, the residual factor 0.5, BP learning rate 0.01, the number of iterations 200 and allowable error 10-4。
Beneficial effects of the present invention: first is that major influence factors, Neng Gougeng can be determined in influence factor with KPCA
Accurately and effectively realize the subitem short-term forecast of subway station energy consumption；Second is that by the ant colony optimization algorithm fusion of science of heredity in BP mind
Through using the connection weight and threshold value between G-ACO training each layer of Optimized BP Neural Network, then passing through BP nerve again in network
Network further does error Reverse optimization to the weight and threshold value of G-ACO training optimization, can be effectively prevented from BP neural network
Brought defect, reaches the intelligent optimizing purpose of neural network model when training.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, required use in being described below to embodiment
Attached drawing be briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this
For the those of ordinary skill of field, without any creative labor, it can also be obtained according to these attached drawings other
Attached drawing.Wherein:
Fig. 1 is the flow chart of subway station energy consumption short term prediction method described in first embodiment of the invention；
Fig. 2 is the overall flow figure of optimum algorithm of multi-layer neural network described in third embodiment of the invention；
Fig. 3 is power energy consumption prediction result figure of the invention；
Fig. 4 is lighting energy consumption prediction result figure of the invention；
Fig. 5 is ventilation and air conditioning energy consumption prediction result figure of the invention；
Fig. 6 is power energy consumption error comparison diagram of the invention；
Fig. 7 is lighting energy consumption error comparison diagram of the invention；
Fig. 8 is ventilation and air conditioning energy consumption error comparison diagram of the invention.
Specific embodiment
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, right with reference to the accompanying drawings of the specification
A specific embodiment of the invention is described in detail, it is clear that and described embodiment is a part of the embodiments of the present invention, and
It is not all of embodiment.Based on the embodiments of the present invention, ordinary people in the field is without making creative work
Every other embodiment obtained, all should belong to the range of protection of the invention.
In the following description, numerous specific details are set forth in order to facilitate a full understanding of the present invention, but the present invention can be with
Implemented using other than the one described here other way, those skilled in the art can be without prejudice to intension of the present invention
In the case of do similar popularization, therefore the present invention is not limited by the specific embodiments disclosed below.
Secondly, " one embodiment " or " embodiment " referred to herein, which refers to, may be included at least one realization side of the invention
A particular feature, structure, or characteristic in formula." in one embodiment " that different places occur in the present specification not refers both to
The same embodiment, nor the individual or selective embodiment mutually exclusive with other embodiments.
Thirdly, combination schematic diagram of the present invention is described in detail, when describing the embodiments of the present invention, for purposes of illustration only,
Indicate that the sectional view of device architecture can disobey general proportion and make partial enlargement, and the schematic diagram is example, herein not
The scope of protection of the invention should be limited.In addition, the three-dimensional space of length, width and depth should be included in actual fabrication.
Simultaneously in the description of the present invention, it should be noted that the orientation of the instructions such as " upper and lower, inner and outer " in term
Or positional relationship is to be based on the orientation or positional relationship shown in the drawings, and is merely for convenience of description of the present invention and simplification of the description, and
It is not that the device of indication or suggestion meaning or element must have a particular orientation, be constructed and operated in a specific orientation, therefore
It is not considered as limiting the invention.In addition, term " first, second or third " is used for description purposes only, and cannot understand
For indication or suggestion relative importance.
In the present invention unless otherwise clearly defined and limited, term " installation is connected, connection " shall be understood in a broad sense, example
Such as: may be a fixed connection, be detachably connected or integral type connection；It equally can be mechanical connection, be electrically connected or be directly connected to,
Can also indirectly connected through an intermediary, the connection being also possible to inside two elements.For the ordinary skill people of this field
For member, the concrete meaning of above-mentioned term in the present invention can be understood with concrete condition.
Embodiment 1
It is proposed by the present invention a kind of based on the subway station KPCA-G-ACO energy consumption short term prediction method, tool such as the signal of Fig. 1
Body includes the following steps:
Step 1, it determines influence factor related with subway station subitem energy consumption, determines and lead in influence factor with KPCA
Influence factor is wanted, then acquires major influence factors, and the major influence factors of acquisition are pre-processed, it should be noted that
Rejecting abnormalities data are referred to the pretreatment of major influence factors, fill up missing data；
Step 2, comprehensive analysis is carried out according to subway station subitem energy consumption historical data, establishes subway station subitem energy consumption number
According to library, then it is normalized together with the influence factor of collected subway station subitem energy consumption, will finally locates in advance again
Data after reason are divided into training data and test data；
Step 3, construction G-ACO-BP Network Prediction Model optimizes the parameter of G-ACO-BP Network Prediction Model
Setting, the parameter of G-ACO-BP Network Prediction Model include cross and variation probability, information heuristic factor and BP learning rate, this step
The parameters such as model initial cross and variation probability, information heuristic factor and BP learning rate all need to initially set up in rapid,
And these parameters are got by formula or experience, later by being trained to model, obtain optimal weight and threshold
Value；
Step 4, using training data training G-ACO-BP model, the error of reality output and target output is calculated；
Step 5, test data is input in the G-ACO-BP model after training and is tested, obtain ventilation and air conditioning energy
Consumption, power energy consumption, the prediction result of lighting energy consumption.
Wherein, influence factor related with subway station subitem energy consumption includes station volume of the flow of passengers x in step 11, it is 24 daily
Integral point moment x2, festivals or holidays x3, in station by when mean temperature x4, in station by when average relative humidity x5, season x6, weather it is special
Value indicative x7, station entrance-exit quantity x8, start columns x9, station average illumination x10, station scale x11And station spacing x12
This 12 influence factors.
Embodiment 2
For determining major influence factors in 12 influence factors with KPCA in above-described embodiment step 1 in the present embodiment
Specific steps are as follows:
Step 1.2, for 12 selected predicted characteristics x=[x1,x2,…,x12], according to formula (1) to subway station
The impact factor of energy consumption is standardized, and standardization herein refers to the pretreatment in step 1 to major influence factors,
It refers to rejecting abnormalities data, fills up missing data:
Wherein, xiIt (k) is k-th of sampled value of i-th of influence factor, xmax(i) and xminIt (i) is respectively i-th of influence
The maximum value and minimum value of all sampled points of factor, TiIt (k) is the target data after standardization；
Step 1.2, nuclear matrix K is asked according to formula (2), is realized using Radial basis kernel function by initial data by data sky
Between be mapped to feature space:
Step 1.3, centralization nuclear matrix K is utilizedCTo correct nuclear matrix K, correction formula are as follows:
KC=K-lNK-KlN+lNKlN(3)
Wherein, KCFor the centralization matrix of N × N, it is nuclear matrix, l that each element, which is for 1/N, K,N∈RN×N,lN(i,j)
=1/N；
Step 1.4, calculating matrix KCCharacteristic value, corresponding feature vector be λ1,λ2,...,λ12, variance v1,
v2,...,v12, by characteristic value descending sort, feature vector makees corresponding adjustment；
Step 1.5, using Schmidt process, orthogonalization and unitization feature vector, obtained feature vector
For a1,a2,...,a12；
Step 1.6, the contribution rate of accumulative total r of feature is calculated1,r2,...,r12, p is required according to given contribution rate, if rt>
P, then t principal component a before choosing1,...,at(t < 12) is major influence factors.
Embodiment 3
For in step 3 in the present embodiment, the G-ACO-BP Network Prediction Model of construction is one containing n-layer input, n-layer
Any neural network of output, wherein including input layer, hidden layer and output layer, the characteristic of each node is Sigmoid type letter
Number, and in step 3 construction G-ACO-BP Network Prediction Model specific steps are as follows:
1. the potential solution of pair problem carries out " digitlization " coding, initialization population；
2. starting the cycle over, fitness individual corresponding to every chromosome is assessed；
3., select probability bigger principle higher in accordance with fitness selects two individuals as paternal and female from population
Side；
4. selecting the chromosome of parent both sides, intersected, generates filial generation；
5. the chromosome of pair filial generation makes a variation；
6. 3,4,5 steps are repeated, until generating several optimization solutions；
7. initializing whole network, time t=0, cycle-index N are enabledc=0, ifFor maximum cycle, enable each
The information content of each element in setAndWhole ants are placed in the nest of ant；
8. starting all ants, for setAnt k (k=1,2 ..., h) (4) calculating state transfer according to the following formula
Probability:
9. step 8 is repeated, until all ant colonies all arrive at food source；
10. a t+m assigns t, Nc+ 1 assigns Nc, the weight that recycles each ant to select and threshold calculations neural network
Output valve and error record current optimal solution.By m chronomere, ant is from its nest arrival food source, each path
Information content update according to the following formula:
Wherein, ekThe one group of weight and threshold value that k-th of ant is selected are as the output of BP neural network weight and threshold value
Error is defined as follows shown in formula:
ek=| O-Oq|(8)
Wherein, O is the real output value of BP neural network, OqFor BP neural network desired output.By can in formula (8)
See, error ekIt is worth smaller, the increase of corresponding information amount is more；
11. being tested using verifying sample to the generalization ability of trained neural network, if ant colony all restrains
To a paths, i.e. optimal path or cycle-indexThen circulation terminates, and exports corresponding calculated result；It is no
Step 8 is then jumped to continue to operate；
12. the weight obtained under optimal path under G-ACO algorithm and threshold value are updated to BP neural network by the training of network
Learning sample after normalized is trained and tests using the weight and threshold value that optimize under BP neural network.
By the training optimization process of the above G-ACO-BP neural network, referring to such as the overall flow figure of Fig. 2 optimization algorithm
Signal.And the present embodiment to the parameter of G-ACO-BP Network Prediction Model for carrying out the specific ginsengs of optimal settings in step 3
Number is referring to the following table 1.
(table 1)
Experimental verification is carried out based on the subway station KPCA-G-ACO energy consumption short term prediction method to proposed by the present invention below:
Experiment sample data source is the April 1 31 days to 2018 March in 2017 in No. 3 line stations of Nanjing subway
Ventilation and air conditioning energy consumption hourly, the subitem monitoring data of power energy consumption and lighting energy consumption of a year and a day day.Wherein in March, 2017
31 days to this 10 months 1 day 2 months 2018 7320 groups of data hourly are as training data, and 2 days to 2018 2 months 2018
1416 groups of data on April 1 are as test data.
The station volume of the flow of passengers that KPCA is extracted, daily 24 integral point moment, festivals or holidays, in station by when mean temperature and
Average illumination this 5 main affecting factors in station are joined with ventilation and air conditioning energy consumption, power energy consumption and lighting energy consumption together as input
Number establishes G-ACO-BP model prediction ventilation and air conditioning energy consumption on April 8,2 days to 2018 April in 2018, power energy consumption, illumination energy
Consumption.Based on same historical data, the data of subway energy consumption prediction model proposed by the present invention prediction and existing 2 kinds of energy consumptions are predicted
Model: it is compared based on GA-BP neural network with the data based on ACO model prediction.Choose one week energy consumption predicted value and reality
Actual value is compared this 3 kinds of models, compares as shown in Fig. 3, Fig. 4 and Fig. 5.As can be seen that G- from Fig. 3, Fig. 4 and Fig. 5
The predicted value effect of other two models of the predicted value curve ratio of ACO-BP model is more preferable, and predicted value curve is closer in reality
It is worth curve.
G-ACO-BP model, ACO model and GA-BP model predict power energy consumption, lighting energy consumption and ventilation and air conditioning energy consumption
Error compare as shown in Fig. 6, Fig. 7 and Fig. 8.From Fig. 6, Fig. 7 and Fig. 8 it can be seen that G-ACO-BP model, ACO model and
GA-BP model predictive error is all the fluctuation above and below 0 value, but G-ACO-BP model predictive error ratio ACO model and GA-BP
Model predictive error fluctuation is small, relatively steadily.Therefore Stability and veracity is higher in G-ACO-BP model training learning process, in advance
It is more excellent to survey performance.
Further to verify prediction result, using mean absolute error (MAE) and root-mean-square error (RMSE) as mould
The evaluation index of type performance carries out analysis comparison to test data respectively, obtains 3 models to power, illumination and ventilation and air conditioning
Energy consumption error comparison result is as shown in table 2.
The performance indicator comparison of table 2G-ACO-BP model and ACO model, the prediction of GA-BP model energy consumption
As can be seen from Table 2, G-ACO-BP model energy consumption prediction MAE and RMSE all than other two model predictions than
It is small, illustrate that G-ACO-BP prediction model proposed by the invention is more preferable to the effect of building energy consumption subitem prediction, improves building
The accuracy of energy consumption prediction, but G-ACO-BP model to the prediction error of ventilation and air conditioning energy consumption than power, lighting energy consumption it is pre-
It is big to survey error, because the energy consumption of ventilation and air conditioning system typically constitutes from the large percentage of total energy consumption, and ventilation and air conditioning system in subway
The difference of form will cause large effect to air conditioning energy consumption, while there are many impact factor for influencing ventilation and air conditioning energy consumption, this hair
It is bright not fully consider these factors when establishing model, cause the prediction error of ventilation and air conditioning energy consumption relatively large.
The invention further relates to a kind of forecasting systems of subway station energy consumption short term prediction method comprising a control is single
Member, the control unit implement foregoing subway station energy consumption short term prediction method, the energy consumption of the prediction subway station
Forecasting system include power-equipment, lighting apparatus, ventilation and air-conditioning equipment and control unit, wherein control unit being capable of root
According to temperature, brightness required for equipment or the setting value of power, the inside temperature of building is adjusted by braking to relevant device
Degree.
The invention further relates to the forecasting systems of the energy requirement of prediction subway station comprising allows the side for implementing to be described above
The computer of method.The system is advantageously associated with the heat supply of subway station and air-conditioning equipment, so as to be based upon reach station in
Desired comfort level carrys out calculated energy consumption to implement thermal conditioning.The system includes such as control unit, the control unit
Computer including the method for implementing to be described above.This method can be by storing software service on an information carrier come real
It applies.Finally, the energy consumption forecasting system for implementing the building for the method being described above can be equipped in subway station, for adjusting
It saves or manages its relevant device in a broad sense.
It should be noted that the above examples are only used to illustrate the technical scheme of the present invention and are not limiting, although referring to preferable
Embodiment describes the invention in detail, those skilled in the art should understand that, it can be to technology of the invention
Scheme is modified or replaced equivalently, and without departing from the spirit and scope of the technical solution of the present invention, should all be covered in this hair
In bright scope of the claims.
Claims (7)
1. a kind of subway station energy consumption short term prediction method, it is characterised in that: include the following steps,
It determines influence factor related with subway station subitem energy consumption, major influence factors are determined in influence factor and acquires,
And the major influence factors of acquisition are pre-processed；
Comprehensive analysis is carried out according to subway station energy consumption historical data of itemizing, subway station subitem energy consumption data library is established, and adopts
The major influence factors collected are normalized together, and the data after normalized are finally divided into training again
Data and test data；
Construction G-ACO-BP Network Prediction Model further includes carrying out optimizing to the parameter of G-ACO-BP Network Prediction Model setting
It sets, the parameter of G-ACO-BP Network Prediction Model includes cross and variation probability, information heuristic factor and BP learning rate；
Using training data training G-ACO-BP model, the error of reality output and target output is calculated；
By test data be input to training after G-ACO-BP model in test, obtain ventilation and air conditioning energy consumption, power energy consumption,
The prediction result of lighting energy consumption.
2. energy consumption short term prediction method in subway station as described in claim 1, it is characterised in that: described to itemize with subway station
The related influence factor of energy consumption includes the station volume of the flow of passengers (x1), daily 24 integral point moment (x2), festivals or holidays (x3), in station by
When mean temperature (x4), in station by when average relative humidity (x5), season (x6), weather characteristics value (x7), station entrance-exit number
Measure (x8), start columns (x9), station average illumination (x10), the scale (x at station11) and station spacing (x12) this 12 influences
Factor.
3. energy consumption short term prediction method in subway station as claimed in claim 1 or 2, it is characterised in that: described in influence factor
Middle determining major influence factors are to determine major influence factors in 12 influence factors with KPCA, further comprising the steps of,
For 12 selected predicted characteristics x=[x1,x2,…,x12], according to formula (1) to the influence factor of subway station energy consumption
It is standardized:
Wherein, xiIt (k) is k-th of sampled value of i-th of influence factor, xmax(i) and xminIt (i) is respectively i-th of influence factor
The maximum value and minimum value of all sampled points, TiIt (k) is the target data after standardization；
Nuclear matrix K is sought according to formula (2), is realized using Radial basis kernel function by initial data by data space map to feature
Space:
Utilize centralization nuclear matrix KCTo correct nuclear matrix K, correction formula are as follows:
KC=K-lNK-KlN+lNKlN (3)
Wherein, KCFor the matrix of N × N, each element is for 1/N；
Calculating matrix KCCharacteristic value, corresponding feature vector be λ1,λ2,...,λ12, variance v1,v2,...,v12, press
Characteristic value descending sort, feature vector make corresponding adjustment；
Using Schmidt process, orthogonalization and unitization feature vector, obtained feature vector are a1,a2,...,
a12；
Calculate the contribution rate of accumulative total r of feature1,r2,...,r12, p is required according to given contribution rate, if rt> p, then t before choosing
A principal component a1,...,at(t < 12) is major influence factors.
4. energy consumption short term prediction method in subway station as claimed in claim 3, it is characterised in that: the G-ACO-BP of the construction
Network Prediction Model is any neural network containing n-layer input, n-layer output, wherein including input layer, hidden layer and output
Layer, and the characteristic of each node is Sigmoid type function.
5. claim 1,2 or 4 it is any as described in subway station energy consumption short term prediction method, it is characterised in that: the construction
G-ACO-BP Network Prediction Model is further comprising the steps of,
A. " digitlization " coding, initialization population are carried out to the potential solution of problem；
B. it starts the cycle over, assesses fitness individual corresponding to every chromosome；
C. higher in accordance with fitness, the bigger principle of select probability selects two individuals as paternal and maternal from population；
D. the chromosome for selecting parent both sides, is intersected, and filial generation is generated；
E. it makes a variation to the chromosome of filial generation；
F. c, d, step e, until generating several optimization solutions are repeated；
G. whole network is initialized, time t=0, cycle-index N are enabledc=0, ifFor maximum cycle, each set is enabled
In each element information contentAndWhole ants are placed in the nest of ant；
H. start all ants, for setAnt k (k=1,2 ..., h) calculates state transition probability according to formula (4):
I. h step is repeated, until all ant colonies all arrive at food source；
J. t, N are assigned t+mc+ 1 assigns Nc, the output valve of the weight and threshold calculations neural network that recycle each ant to select
And error, record current optimal solution.By m chronomere, ant reaches food source from its nest, the information on each path
Amount updates according to the following formula:
Wherein, ekOne group of weight that k-th of ant is selected and threshold value as the output error of BP neural network weight and threshold value,
It is defined as follows shown in formula:
ek=| O-Oq| (8)
Wherein, O is the real output value of BP neural network, OqFor BP neural network desired output；
K. it is tested using verifying sample to the generalization ability of trained neural network, if ant colony all converges to one
Path, i.e. optimal path or cycle-indexThen circulation terminates, and exports corresponding calculated result；Otherwise it jumps
Continue to operate to h step；
The weight obtained under optimal path under G-ACO algorithm and threshold value are updated to BP neural network normalizing by the l. training of network
Learning sample after change processing, is trained and tests using the weight and threshold value that optimize under BP neural network.
6. energy consumption short term prediction method in subway station as claimed in claim 5, it is characterised in that: to G-ACO-BP neural network forecast
The parameter that the parameter of model carries out optimal settings includes,
Be configured to input layer, hidden layer, the output layer of neuron number are respectively 3,11,3；
Training algorithm uses G-ACO algorithm, Population Size 20, crossover probability 0.5, mutation probability 0.5, ant number 20, information
Heuristic factor 1, the residual factor 0.5, BP learning rate 0.01, the number of iterations 200 and allowable error 10-4。
7. a kind of forecasting system that prediction subway station energy consumption is short-term, it is characterised in that: the forecasting system includes control unit, institute
It states control unit and implements as short-term prediction technique such as the prediction subway station energy consumption of claim 1~2,4 or 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811038101.2A CN109118020A (en) | 2018-09-06 | 2018-09-06 | A kind of subway station energy consumption short term prediction method and its forecasting system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811038101.2A CN109118020A (en) | 2018-09-06 | 2018-09-06 | A kind of subway station energy consumption short term prediction method and its forecasting system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109118020A true CN109118020A (en) | 2019-01-01 |
Family
ID=64858820
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811038101.2A Pending CN109118020A (en) | 2018-09-06 | 2018-09-06 | A kind of subway station energy consumption short term prediction method and its forecasting system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109118020A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109840639A (en) * | 2019-03-05 | 2019-06-04 | 东北大学 | A kind of late time forecasting methods of high speed rail train operation |
| CN109973355A (en) * | 2019-03-29 | 2019-07-05 | 江苏极熵物联科技有限公司 | A kind of energy-saving method towards air compressor |
| CN110443425A (en) * | 2019-08-09 | 2019-11-12 | 长江慧控科技(武汉)有限公司 | Intelligent railway station electric energy energy consumption prediction technique based on Prophet |
| CN111144543A (en) * | 2019-12-30 | 2020-05-12 | 中国移动通信集团内蒙古有限公司 | Data center air conditioner tail end temperature control method, device and medium |
| CN112116070A (en) * | 2020-09-07 | 2020-12-22 | 北方工业大学 | Subway station environmental parameter monitoring method and device |
| CN112395311A (en) * | 2019-08-13 | 2021-02-23 | 阿里巴巴集团控股有限公司 | Method and device for predicting processing duration of request |
| CN112818286A (en) * | 2021-03-31 | 2021-05-18 | 中车青岛四方车辆研究所有限公司 | Method and system for calculating energy consumption of air conditioning system of metro vehicle in cooling season |
| CN114493928A (en) * | 2022-04-18 | 2022-05-13 | 成都四为电子信息股份有限公司 | Electric energy consumption subentry measurement model construction method based on railway passenger station energy consumption characteristics |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104484715A (en) * | 2014-11-28 | 2015-04-01 | 江苏大学 | Neural network and particle swarm optimization algorithm-based building energy consumption predicting method |
| US20170254658A1 (en) * | 2016-03-04 | 2017-09-07 | The Boeing Company | Fuel consumption predictions using associative memories |
| CN108320016A (en) * | 2018-03-08 | 2018-07-24 | 南京工业大学 | A kind of building energy consumption short term prediction method |
-
2018
- 2018-09-06 CN CN201811038101.2A patent/CN109118020A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104484715A (en) * | 2014-11-28 | 2015-04-01 | 江苏大学 | Neural network and particle swarm optimization algorithm-based building energy consumption predicting method |
| US20170254658A1 (en) * | 2016-03-04 | 2017-09-07 | The Boeing Company | Fuel consumption predictions using associative memories |
| CN108320016A (en) * | 2018-03-08 | 2018-07-24 | 南京工业大学 | A kind of building energy consumption short term prediction method |
Non-Patent Citations (1)
| Title |
|---|
| 陈旭: "基于神经网络的电力系统短期负荷预测研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109840639A (en) * | 2019-03-05 | 2019-06-04 | 东北大学 | A kind of late time forecasting methods of high speed rail train operation |
| CN109973355A (en) * | 2019-03-29 | 2019-07-05 | 江苏极熵物联科技有限公司 | A kind of energy-saving method towards air compressor |
| CN109973355B (en) * | 2019-03-29 | 2020-05-19 | 江苏极熵物联科技有限公司 | Energy saving and consumption reducing method for air compressor |
| CN110443425A (en) * | 2019-08-09 | 2019-11-12 | 长江慧控科技(武汉)有限公司 | Intelligent railway station electric energy energy consumption prediction technique based on Prophet |
| CN112395311A (en) * | 2019-08-13 | 2021-02-23 | 阿里巴巴集团控股有限公司 | Method and device for predicting processing duration of request |
| CN111144543A (en) * | 2019-12-30 | 2020-05-12 | 中国移动通信集团内蒙古有限公司 | Data center air conditioner tail end temperature control method, device and medium |
| CN112116070A (en) * | 2020-09-07 | 2020-12-22 | 北方工业大学 | Subway station environmental parameter monitoring method and device |
| CN112116070B (en) * | 2020-09-07 | 2024-04-05 | 北方工业大学 | Subway station environment parameter monitoring method and device |
| CN112818286A (en) * | 2021-03-31 | 2021-05-18 | 中车青岛四方车辆研究所有限公司 | Method and system for calculating energy consumption of air conditioning system of metro vehicle in cooling season |
| CN112818286B (en) * | 2021-03-31 | 2022-10-28 | 中车青岛四方车辆研究所有限公司 | Method and system for calculating energy consumption of air conditioning system of metro vehicle in cooling season |
| CN114493928A (en) * | 2022-04-18 | 2022-05-13 | 成都四为电子信息股份有限公司 | Electric energy consumption subentry measurement model construction method based on railway passenger station energy consumption characteristics |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109118020A (en) | A kind of subway station energy consumption short term prediction method and its forecasting system | |
| CN111091143B (en) | Distribution transformer weight overload early warning method based on deep belief network and K-means clustering | |
| CN108009674A (en) | Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks | |
| CN109886444A (en) | A kind of traffic passenger flow forecasting, device, equipment and storage medium in short-term | |
| CN111160628A (en) | Air pollutant concentration prediction method based on CNN and double-attention seq2seq | |
| CN109165764A (en) | A kind of line loss calculation method of genetic algorithm optimization BP neural network | |
| CN109146144A (en) | Based on the landslide disaster prediction technique for improving Bayesian network | |
| CN106651036A (en) | Air quality forecasting system | |
| CN109785629A (en) | A kind of short-term traffic flow forecast method | |
| CN106997669A (en) | A kind of method of the judgement traffic congestion origin cause of formation of feature based importance | |
| CN107704875A (en) | Based on the building load Forecasting Methodology and device for improving IHCMAC neutral nets | |
| CN106501721A (en) | A kind of lithium battery SOC estimation method based on biological evolution | |
| CN103679263A (en) | Thunder and lightning approach forecasting method based on particle swarm support vector machine | |
| CN109785618A (en) | Short-term traffic flow prediction method based on combinational logic | |
| CN108921339A (en) | Genetic Support Vector Machine photovoltaic power interval prediction method based on quantile estimate | |
| CN113762603B (en) | Photovoltaic base station short-term photovoltaic power prediction method based on improved sparrow algorithm optimization | |
| CN113762387B (en) | Multi-element load prediction method for data center station based on hybrid model prediction | |
| CN112149887A (en) | PM2.5 concentration prediction method based on data space-time characteristics | |
| CN110264112A (en) | Distribution network reliability analysis of Influential Factors method based on bidirectional weighting grey correlation | |
| CN109816204A (en) | A kind of safety of subway operation Tendency Prediction method based on GA-SVM | |
| CN106327357A (en) | Load identification method based on improved probabilistic neural network | |
| CN110363349A (en) | A kind of LSTM neural network hydrologic(al) prognosis method and system based on ASCS | |
| Song et al. | An indoor temperature prediction framework based on hierarchical attention gated recurrent unit model for energy efficient buildings | |
| CN109492748A (en) | A kind of Mid-long term load forecasting method for establishing model of the electric system based on convolutional neural networks | |
| CN110070228A (en) | BP neural network wind speed prediction method for neuron branch evolution |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190101 |
|
| WD01 | Invention patent application deemed withdrawn after publication |