CN114139794A - Regional public parking lot charging demand prediction method based on parking big data - Google Patents

Regional public parking lot charging demand prediction method based on parking big data Download PDF

Info

Publication number
CN114139794A
CN114139794A CN202111413408.8A CN202111413408A CN114139794A CN 114139794 A CN114139794 A CN 114139794A CN 202111413408 A CN202111413408 A CN 202111413408A CN 114139794 A CN114139794 A CN 114139794A
Authority
CN
China
Prior art keywords
commercial vehicle
demand prediction
charging demand
parking
prediction model
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
Application number
CN202111413408.8A
Other languages
Chinese (zh)
Inventor
汪春
张卫华
李理
赵世
祝凯
吴丛
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.)
Hefei University Of Technology Design Institute Group Co ltd
Original Assignee
Hefei University Of Technology Design Institute Group 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 Hefei University Of Technology Design Institute Group Co ltd filed Critical Hefei University Of Technology Design Institute Group Co ltd
Priority to CN202111413408.8A priority Critical patent/CN114139794A/en
Publication of CN114139794A publication Critical patent/CN114139794A/en
Pending legal-status Critical Current

Links

Images

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
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/14—Traffic control systems for road vehicles indicating individual free spaces in parking areas
    • G08G1/145—Traffic control systems for road vehicles indicating individual free spaces in parking areas where the indication depends on the parking areas
    • G08G1/148—Management of a network of parking areas

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Charge And Discharge Circuits For Batteries Or The Like (AREA)

Abstract

The invention discloses a regional public parking lot charging demand prediction method based on parking big data, belonging to the technical field of public utilities, and the specific method comprises the following steps: the method comprises the following steps: the method comprises the steps of obtaining parking, charging and operating data of a new energy vehicle, and dividing a research range; step two: establishing a unified data access platform, and acquiring original driving data and parking data of a new energy vehicle driving in a research range; step three: establishing a non-commercial vehicle charging demand prediction model and a commercial vehicle charging demand prediction model; step four: acquiring parking and charging data of new energy vehicles in a public parking lot, and predicting the charging demands of non-commercial vehicles and commercial vehicles through a non-commercial vehicle charging demand prediction model and a commercial vehicle charging demand prediction model; step five: and evaluating the prediction data of the charging demands of the non-commercial vehicle and the commercial vehicle, and relearning the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model according to the evaluation result.

Description

Regional public parking lot charging demand prediction method based on parking big data
Technical Field
The invention belongs to the technical field of public parking lot charging, and particularly relates to a regional public parking lot charging demand prediction method based on parking big data.
Background
Along with new energy vehicle constantly increases, the vehicle demand of charging increases fast, and the supporting construction in public parking area fills electric pile demand constantly increases, and public parking area is based on the basic requirement that parking position proportion configuration fills electric pile also gradually becomes each place. At present, relatively clear requirements are provided for the proportion of charging piles which are constructed in a residential community and a public parking lot in a matching way, and no relevant requirements are provided for the types of the charging piles; fill electric pile construction and be in the fast propulsion phase for fill electric pile quantity and increase fast. Through investigation and discovery, the condition that the utilization ratio is uneven exists in the electric pile of filling of public parking area construction, and the supporting construction of filling electric pile according to fixed proportion does not cooperate with the demand of charging of new energy vehicle, causes to charge difficult and the idle two aspect contradiction coexistence of resource.
Disclosure of Invention
In order to solve the problems existing in the scheme, the invention provides a regional public parking lot charging demand prediction method based on parking big data.
The purpose of the invention can be realized by the following technical scheme:
the regional public parking lot charging demand prediction method based on parking big data comprises the following specific steps:
the method comprises the following steps: the method comprises the steps of obtaining parking, charging and operating data of a new energy vehicle, and dividing a research range;
step two: establishing a unified data access platform, and acquiring original driving data and parking data of a new energy vehicle driving in a research range;
step three: establishing a non-commercial vehicle charging demand prediction model and a commercial vehicle charging demand prediction model;
step four: acquiring parking and charging data of new energy vehicles in a public parking lot, and predicting the charging demands of non-commercial vehicles and commercial vehicles through a non-commercial vehicle charging demand prediction model and a commercial vehicle charging demand prediction model;
step five: and evaluating the prediction data of the charging demands of the non-commercial vehicle and the commercial vehicle, and relearning the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model according to the evaluation result.
Further, the method for dividing the research scope comprises the following steps: the method comprises the steps of obtaining an urban plan, identifying urban main roads and natural barriers, dividing minimum units according to the urban main roads and the natural barriers, carrying out hot spot analysis and clustering on new energy vehicles on each minimum unit in an urban area according to parking, charging and running data of the new energy vehicles, merging the minimum units based on a clustering rule, and determining a research range boundary.
Further, the research range division method based on hierarchical clustering is as follows:
SA 1: inputting: sample set D to be clustered is x1,x2,...,xi,...,xn}, the number of clusters; x is the number ofiIs the feature vector of the ith minimum cell,
Figure BDA0003375125670000021
n is the total number of urban units;
Figure BDA0003375125670000022
the average number of stops for the ith minimum unit over the statistical period,
Figure BDA0003375125670000023
the average parking time of the vehicle in the ith minimum unit is taken as the average parking time;
SA 2: all sample points in the sample set are taken as an independent cluster, and the distance d (C) between every two clusters is calculatedi,Cj)，d(Ci,Cj) Satisfies the following conditions:
Figure BDA0003375125670000024
SA 3: finding two nearest clusters CpAnd Cq，
Figure BDA0003375125670000025
Judging cluster CpAnd CqWhether or not less than the maximum parking service radius RparkIf yes, merging cluster CpAnd CqAs a new cluster Cg；
SA 4: recalculating new cluster CgDistance from all other classes;
SA 5: and repeating SA2-SA4 until all the classes are finally combined into one class, namely the research range after clustering and combination.
Further, the concrete training steps for establishing the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model in the third step are as follows:
inputting: training set
Figure BDA0003375125670000031
Maximum iteration number M;
for non-commercial vehicle prediction models, xi＝[S,T,t,E,P]i，yiA charging demand for an ith non-commercial vehicle;
for the commercial vehicle prediction model, xi＝[S,E,A]i，yiA charging demand for an ith operating vehicle;
and (3) outputting: and (4) training the charging demand prediction models of the trained non-commercial vehicles and commercial vehicles.
Further, the method for outputting the trained non-commercial vehicle and commercial vehicle charging demand prediction model comprises the following steps:
SB 1: initializing network parameters: setting the number of nodes of an input layer as n, the number of nodes of a hidden layer as l and the number of nodes of an output layer as m; input layer to hidden layer weight ωijThe weight from hidden layer to output layer is omegajkThe bias of the input layer to the hidden layer is ajThe bias from the hidden layer to the output layer is bk(ii) a Learning rate is eta, and excitation function is g (x); wherein, the excitation function g (x) takes a Sigmoid function, and the calculation formula is as follows:
Figure BDA0003375125670000032
SB 2: computing network hidden layer output HjAnd calculating a formula:
Figure BDA0003375125670000033
SB 3: computing network output layer output OkAnd calculating a formula:
Figure BDA0003375125670000034
SB 4: calculating a network error term E, and calculating a formula:
Figure BDA0003375125670000035
SB 5: updating the weight omegaijAnd ωjkAnd calculating a formula:
Figure BDA0003375125670000036
SB 6: updating the bias of the input layer to the hidden layer ajAnd biasing of the hidden layer to the output layer bkAnd calculating a formula:
Figure BDA0003375125670000041
in the steps SB1-SB6, i, j and k are integers, i belongs to [1, n ], j belongs to [1, l ], and k belongs to [1, m ];
SB 7: judging whether the iteration times are larger than the maximum learning times M or not, and if so, stopping training; otherwise, SB1-SB6 are repeated until the iteration stop condition is satisfied.
Further, e in SB4k＝Yk-Ok，YkIs the desired output.
Further, the method for relearning the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model according to the evaluation result in the fifth step includes:
and constructing a post-evaluation index system, determining post-evaluation standards of all indexes, performing post-evaluation of demand prediction according to prediction data of the charging demands of the non-commercial vehicle and the commercial vehicle and actual data after implementation according to the prediction, re-calibrating parameters of the prediction model according to a post-evaluation result, integrating the parameters into self-learning data, and inputting the self-learning data into a corresponding non-commercial vehicle charging demand prediction model or a commercial vehicle charging demand prediction model for relearning.
Further, the post-stratification evaluation process comprises:
optimizing parameters of the prediction model by using a genetic algorithm, wherein the parameters are optimized by: connecting weight omega of input layer and hidden layerijHidden layer bias value ajThe hidden layer is connected with the output layer to form a weight omegajkOutput layer bias value bkThe parameter optimization and adjustment steps are as follows:
SC 1: population initialization: coding the initialized neural network parameter optimization set by using a real number coding method;
SC 2: calculating a fitness function: according to the expected output value and the actual output value of the neural network, defining an individual fitness value F, and calculating according to the following formula:
Figure BDA0003375125670000042
wherein n is the number of network output nodes;
Figure BDA0003375125670000043
the expected output value of the ith node of the neural network; y isiIs the actual output value of the ith node; k is a coefficient;
SC 3: and (3) carrying out selection operation: when selecting roulette, i.e. selection strategies based on fitness ratio, the probability of selection p for each individual iiComprises the following steps:
Figure BDA0003375125670000051
wherein, FiFitness value for individual i; k is a coefficient; n is the number of population individuals;
SC 4: and (3) executing a cross operation: performing crossover operation by using real number crossover method, the kth chromosome akAnd a firstl chromosomes alThe method of the crossover operation in the j dimension is as follows:
Figure BDA0003375125670000052
b is [0,1]]A random number in between;
SC 5: performing mutation operation: selecting the jth gene a of the ith individualijPerforming mutation operation by the following method:
Figure BDA0003375125670000053
wherein, amaxAnd aminAre respectively a gene aijUpper and lower bounds of (a); f (G) is a random number satisfying f (G) r (1-G/G)max)2r, G is the current iteration number, GmaxIs the maximum number of evolutions; r is [0,1]]A random number in between;
SC 6: judging whether the maximum iteration times is reached, if so, stopping iteration, and outputting the optimal weight and bias value of the neural network to obtain an optimal neural network prediction model; otherwise, execution continues at steps SC3-SC 5.
Compared with the prior art, the invention has the beneficial effects that: and constructing a neural network prediction model based on the parking and charging data of the public parking lot, the daily operation and real-time electric quantity use data of the new energy vehicle in the research range, so that the prediction result is more real and credible, and real-time verification can be performed by using the measured data. The invention can enable the construction scale, type and time sequence of the charging pile to be more in line with the actual requirements of the new energy vehicles, effectively solves the problems of idle charging pile resources, difficult charging and the like existing in different parking lots at the present stage, enables the charging pile arrangement of a public parking lot to achieve high-quality dynamic balance, and solves the worries of charging after the new energy vehicles are preserved and rapidly increased.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a flow chart of a non-commercial vehicle and commercial vehicle charging demand prediction model of the present invention.
Detailed Description
The technical solutions of the present invention will be described clearly and completely with reference to the following embodiments, and it should be understood that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In an embodiment, as shown in fig. 1 to 2, a method for predicting a charging demand of a regional public parking lot based on parking big data specifically includes:
the method comprises the following steps: the method comprises the steps of obtaining parking, charging and operating data of a new energy vehicle, and dividing a research range;
and acquiring an urban plan, and identifying urban trunks and natural barriers, wherein the natural barriers are railways, rivers and the like. The method comprises the steps of dividing minimum units according to urban trunks and natural barriers, carrying out hot spot analysis and clustering on new energy vehicles for each minimum unit in an urban area according to parking, charging and running data of the new energy vehicles, combining the minimum units based on a clustering rule, and determining a research range boundary under a constraint condition.
Step two: establishing a unified data access platform, developing a data interface, and acquiring the original driving data of the new energy vehicle driving in a research range from a new energy vehicle real-time monitoring platform of a relevant department or manufacturer; the original data comprises data such as real-time position data of the new energy vehicle, real-time residual electric quantity of the new energy vehicle, daily running track of the new energy vehicle and the like; the method comprises the steps that parking data are obtained from an intelligent parking platform or a parking management system of a public parking lot under study, wherein the parking data comprise new energy vehicle parking time periods, parking duration, parking times and other data;
step three: establishing a non-commercial vehicle charging demand prediction model and a commercial vehicle charging demand prediction model;
according to the charging characteristics of the new energy vehicles at the present stage, a charging demand prediction model is constructed according to two categories of non-operation new energy vehicles and operation new energy vehicles;
the required quantity and type of the charging piles are closely related to the quantity of new energy vehicles in a research range, daily parking data (parking place, parking duration, parking time and the like) of the new energy vehicles, residual electric quantity during parking, utilization rate of the charging piles in a public parking lot, real-time electricity price data and the like; according to the new energy vehicle big data and the parking big data in the research range, the collected related data are used as input data, a charging supply and demand model based on a neural network is built, training and testing are performed by utilizing measured data, so that the parking demand and the charging demand of the new energy vehicle in the research range are output, and the proportion of the charging pile and the recommended value of the type of the charging pile in the public parking range are recommended according to the parking scale and the parking turnover rate of the public parking lot in the research range, and the layout and the scale of the newly-built charging pile are recommended.
The public parking lot non-commercial vehicle charging demand prediction model mainly refers to the following parameters:
s: the parking quantity of the new energy vehicles in the parking lot;
t: the parking time length;
t: parking period (related to electricity price);
e: remaining electric quantity when parking;
p: average parking cost;
the public parking lot commercial vehicle charging demand prediction model mainly refers to the following parameters:
s: the number of vehicles operated by new energy at the low-peak electricity utilization time period within the parking service radius is increased;
e: average residual capacity in a power utilization low peak period;
a: average turnover rate of public parking lots;
and predicting the charging demands of the non-commercial vehicles and the commercial vehicles by adopting an artificial neural network algorithm based on the parking and charging data of the new energy vehicles in the public parking lot.
Step four: acquiring parking and charging data of new energy vehicles in a public parking lot, and predicting the charging demands of non-commercial vehicles and commercial vehicles through a non-commercial vehicle charging demand prediction model and a commercial vehicle charging demand prediction model; the parking and charging data of the new energy vehicle in the public parking lot comprises input data required by a non-commercial vehicle charging demand prediction model and a commercial vehicle charging demand prediction model, and can be seen from the above description;
step five: and evaluating the prediction data of the charging demands of the non-commercial vehicle and the commercial vehicle, and relearning the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model according to the evaluation result.
Expressing the specific clustering process by using a hierarchical clustering method; there are different inputs and outputs for the present invention.
The research range division based on hierarchical clustering comprises the following steps:
SA 1: inputting: sample set D to be clustered is x1,x2,...,xi,...,xn}, the number of clusters; wherein x isiIs the feature vector of the ith minimum cell,
Figure BDA0003375125670000081
and n is the total number of urban units.
Figure BDA0003375125670000082
The average number of stops for the ith minimum unit over the statistical period,
Figure BDA0003375125670000083
the average parking time of the vehicle in the ith minimum unit is taken as the average parking time; the smallest unit is an urban unit and is a parking lot unit surrounded by roads, rivers and the like;
SA 2: all sample points in the sample set are taken as an independent cluster, and the distance d (C) between every two clusters is calculatedi,Cj)，d(Ci,Cj) Satisfies the following conditions:
Figure BDA0003375125670000084
SA 3: finding two nearest clusters CpAnd Cq，
Figure BDA0003375125670000085
Judging cluster CpAnd CqWhether or not less than the maximum parking service radius RparkIf yes, merging cluster CpAnd CqAs a new cluster Cg(ii) a The maximum parking service radius is set by the service radius of the charging pile;
SA 4: recalculating new cluster CgDistance from all other classes; all other classes refer to other cluster clusters;
SA 5: and repeating SA2-SA4 until all the classes are finally combined into one class, namely the research range after clustering and combination.
And determining the boundary of the research range based on the maximum parking service radius, parking and charging requirements and other constraint conditions in the research range.
The concrete training steps for establishing the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model in the third step are as follows:
inputting: training set
Figure BDA0003375125670000091
Maximum iteration number M;
for non-commercial vehicle prediction models, xi＝[S,T,t,E,P]i，yiA charging demand for an ith non-commercial vehicle;
for the commercial vehicle prediction model, xi＝[S,E,A]i，yiA charging demand for an ith operating vehicle;
and (3) outputting: and (4) training the charging demand prediction models of the trained non-commercial vehicles and commercial vehicles.
SB 1: initializing network parameters: suppose the number of nodes of the input layer is n, the number of nodes of the hidden layer is l, and the number of nodes of the output layer is m. Input layer to hidden layer weight ωijIs hiddenLayer to output layer weight is ωjkThe bias of the input layer to the hidden layer is ajThe bias from the hidden layer to the output layer is bk. Learning rate is η and excitation function is g (x). Wherein, the excitation function g (x) is a Sigmoid function, and the calculation mode is as follows:
Figure BDA0003375125670000092
SB 2: computing network hidden layer output HjThe calculation method is as follows:
Figure BDA0003375125670000093
SB 3: computing network output layer output OkThe calculation method is as follows:
Figure BDA0003375125670000094
SB 4: and calculating a network error term E in the following way:
Figure BDA0003375125670000101
wherein e isk＝Yk-Ok，YkIs the desired output.
SB 5: updating the weight omegaijAnd ωjkThe calculation method is as follows:
Figure BDA0003375125670000102
SB 6: updating the bias of the input layer to the hidden layer ajAnd biasing of the hidden layer to the output layer bkThe calculation method is as follows:
Figure BDA0003375125670000103
in the steps SB1-SB6, i, j and k are integers, i belongs to [1, n ], j belongs to [1, l ], and k belongs to [1, m ];
SB 7: judging whether the iteration times are larger than the maximum learning times M or not, and if so, stopping training; otherwise, SB1-SB6 are repeated until the iteration stop condition is satisfied.
In the fifth step, the method for relearning the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model according to the evaluation result comprises the following steps:
establishing a post-evaluation index system, determining post-evaluation standards of all indexes, establishing a layered post-evaluation model, performing post-evaluation of demand prediction according to prediction data of the charging demands of the non-commercial vehicle and the commercial vehicle and actual data after implementation according to prediction, re-calibrating parameters of the prediction model according to a post-evaluation result, integrating the parameters into self-learning data, and inputting the self-learning data into a corresponding non-commercial vehicle charging demand prediction model or commercial vehicle charging demand prediction model for relearning.
The evaluation process after layering comprises the following steps:
optimizing parameters of the prediction model by using a genetic algorithm, wherein the parameters are optimized by: connecting weight omega of input layer and hidden layerijHidden layer bias value ajThe hidden layer is connected with the output layer to form a weight omegajkOutput layer bias value bkThe parameter optimization and adjustment steps are as follows:
SC 1: population initialization: coding the initialized neural network parameter optimization set by using a real number coding method;
SC 2: calculating a fitness function: according to the expected output value and the actual output value of the neural network, defining an individual fitness value F, and calculating according to the following formula:
Figure BDA0003375125670000111
wherein n is the number of network output nodes;
Figure BDA0003375125670000112
the expected output value of the ith node of the neural network; y isiIs the actual output value of the ith node; k is a coefficient.
SC 3: and (3) carrying out selection operation: the genetic algorithm selection operation includes a roulette method, a tournament method, and the like. When selecting roulette, i.e. selection strategies based on fitness ratio, the probability of selection p for each individual iiComprises the following steps:
Figure BDA0003375125670000113
wherein, FiFitness value for individual i; k is a coefficient; n is the number of population individuals.
SC 4: and (3) executing a cross operation: performing crossover operation by using real number crossover method, the kth chromosome akAnd the l-th chromosome alThe method of the crossover operation in the j dimension is as follows:
Figure BDA0003375125670000114
wherein b is a random number between [0,1 ].
SC 5: performing mutation operation: selecting the jth gene a of the ith individualijPerforming mutation operation by the following method:
Figure BDA0003375125670000115
wherein, amaxAnd aminAre respectively a gene aijUpper and lower bounds of (a); f (G) is a random number satisfying f (G) r (1-G/G)max)2r, G is the current iteration number, GmaxIs the maximum number of evolutions; r is [0,1]]Random number in between.
SC 6: judging whether the maximum iteration times is reached, if so, stopping iteration, and outputting the optimal weight and bias value of the neural network to obtain an optimal neural network prediction model; otherwise, execution continues at steps SC3-SC 5.
Although the present invention has been described in detail with reference to the preferred embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the spirit and scope of the present invention.

Claims (8)

1. The regional public parking lot charging demand prediction method based on parking big data is characterized by comprising the following specific steps:
the method comprises the following steps: the method comprises the steps of obtaining parking, charging and operating data of a new energy vehicle, and dividing a research range;
step two: establishing a unified data access platform, and acquiring original driving data and parking data of a new energy vehicle driving in a research range;
step three: establishing a non-commercial vehicle charging demand prediction model and a commercial vehicle charging demand prediction model;
step four: acquiring parking and charging data of new energy vehicles in a public parking lot, and predicting the charging demands of non-commercial vehicles and commercial vehicles through a non-commercial vehicle charging demand prediction model and a commercial vehicle charging demand prediction model;
step five: and evaluating the prediction data of the charging demands of the non-commercial vehicle and the commercial vehicle, and relearning the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model according to the evaluation result.
2. The parking big data-based regional public parking lot charging demand prediction method according to claim 1, wherein the method for dividing the research scope comprises the following steps: the method comprises the steps of obtaining an urban plan, identifying urban main roads and natural barriers, dividing minimum units according to the urban main roads and the natural barriers, carrying out hot spot analysis and clustering on new energy vehicles on each minimum unit in an urban area according to parking, charging and running data of the new energy vehicles, merging the minimum units based on a clustering rule, and determining a research range boundary.
3. The method for predicting the charging demand of the regional public parking lot based on the parking big data as claimed in claim 2, wherein the research range division method based on hierarchical clustering is as follows:
SA 1: inputting: sample set D to be clustered is x1,x2,...,xi,...,xn}, the number of clusters; x is the number ofiIs the feature vector of the ith minimum cell,
Figure FDA0003375125660000011
i∈[1,n]n is the total number of urban units;
Figure FDA0003375125660000012
the average number of stops for the ith minimum unit over the statistical period,
Figure FDA0003375125660000013
the average parking time of the vehicle in the ith minimum unit is taken as the average parking time;
SA 2: all sample points in the sample set are taken as an independent cluster, and the distance d (C) between every two clusters is calculatedi,Cj)，d(Ci,Cj) Satisfies the following conditions:
Figure FDA0003375125660000021
SA 3: finding two nearest clusters CpAnd Cq，
Figure FDA0003375125660000022
Judging cluster CpAnd CqWhether or not less than the maximum parking service radius RparkIf yes, merging cluster CpAnd CqAs a new cluster Cg；
SA 4: recalculating new cluster CgDistance from all other classes;
SA 5: and repeating SA2-SA4 until all the classes are finally combined into one class, namely the research range.
4. The parking big data-based regional public parking lot charging demand prediction method according to claim 1, wherein the specific training steps for establishing the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model in the third step are as follows:
inputting: training set
Figure FDA0003375125660000023
Maximum iteration number M;
for non-commercial vehicle prediction models, xi＝[S,T,t,E,P]i，yiA charging demand for an ith non-commercial vehicle;
for the commercial vehicle prediction model, xi＝[S,E,A]i，yiA charging demand for an ith operating vehicle;
and (3) outputting: and (4) training the charging demand prediction models of the trained non-commercial vehicles and commercial vehicles.
5. The parking big data-based regional public parking lot charging demand prediction method according to claim 4, wherein the method of outputting the trained non-commercial vehicle and commercial vehicle charging demand prediction model comprises:
SB 1: initializing network parameters: setting the number of nodes of an input layer as n, the number of nodes of a hidden layer as l and the number of nodes of an output layer as m; input layer to hidden layer weight ωijThe weight from hidden layer to output layer is omegajkThe bias of the input layer to the hidden layer is ajThe bias from the hidden layer to the output layer is bk(ii) a Learning rate is eta, and excitation function is g (x); wherein, the excitation function g (x) takes a Sigmoid function, and the calculation formula is as follows:
Figure FDA0003375125660000031
SB 2: computing network hidden layer output HjAnd calculating a formula:
Figure FDA0003375125660000032
SB 3: computing network output layer output OkAnd calculating a formula:
Figure FDA0003375125660000033
SB 4: calculating a network error term E, and calculating a formula:
Figure FDA0003375125660000034
SB 5: updating the weight omegaijAnd ωjkAnd calculating a formula:
Figure FDA0003375125660000035
SB 6: updating the bias of the input layer to the hidden layer ajAnd biasing of the hidden layer to the output layer bkAnd calculating a formula:
Figure FDA0003375125660000036
in the steps SB1-SB6, i, j and k are integers, i belongs to [1, n ], j belongs to [1, l ], and k belongs to [1, m ];
SB 7: judging whether the iteration times are larger than the maximum learning times M or not, and if so, stopping training; otherwise, SB1-SB6 are repeated until the iteration stop condition is satisfied.
6. The parking big data-based regional public parking lot charging demand prediction method according to claim 5, wherein e in SB4k＝Yk-Ok，YkIs the desired output.
7. The parking big data-based regional public parking lot charging demand prediction method according to claim 1, wherein the method for relearning the non-commercial vehicle charging demand prediction model and the commercial vehicle charging demand prediction model according to the evaluation result in the fifth step comprises the following steps:
and constructing a post-evaluation index system, determining post-evaluation standards of all indexes, performing post-evaluation of demand prediction according to prediction data of the charging demands of the non-commercial vehicle and the commercial vehicle and actual data after implementation according to the prediction, re-calibrating parameters of the prediction model according to a post-evaluation result, integrating the parameters into self-learning data, and inputting the self-learning data into a corresponding non-commercial vehicle charging demand prediction model or a commercial vehicle charging demand prediction model for relearning.
8. The method for predicting the charging demand of the regional public parking lot based on the parking big data as claimed in claim 7, wherein the evaluation process after layering comprises:
optimizing parameters of the prediction model by using a genetic algorithm, wherein the parameters are optimized by: connecting weight omega of input layer and hidden layerijHidden layer bias value ajThe hidden layer is connected with the output layer to form a weight omegajkOutput layer bias value bkThe parameter optimization and adjustment steps are as follows:
SC 1: population initialization: coding the initialized neural network parameter optimization set by using a real number coding method;
SC 2: calculating a fitness function: according to the expected output value and the actual output value of the neural network, defining an individual fitness value F, and calculating according to the following formula:
Figure FDA0003375125660000041
wherein n is the number of network output nodes;
Figure FDA0003375125660000042
the expected output value of the ith node of the neural network; y isiIs the actual output value of the ith node; k is a coefficient;
SC 3: and (3) carrying out selection operation: when selecting roulette, i.e. selection strategies based on fitness ratio, the probability of selection p for each individual iiComprises the following steps:
Figure FDA0003375125660000043
wherein, FiFitness value for individual i; k is a coefficient; n is the number of population individuals;
SC 4: and (3) executing a cross operation: performing crossover operation by using real number crossover method, the kth chromosome akAnd the l-th chromosome alThe method of the crossover operation in the j dimension is as follows:
Figure FDA0003375125660000044
is [0,1]]A random number in between;
SC 5: performing mutation operation: selecting the jth gene a of the ith individualijPerforming mutation operation by the following method:
Figure FDA0003375125660000051
wherein, amaxAnd aminAre respectively a gene aijUpper and lower bounds of (a); f (G) is a random number satisfying f (G) r (1-G/G)max)2r, G is the current iteration number, GmaxIs the maximum number of evolutions; r is [0,1]]A random number in between;
SC 6: judging whether the maximum iteration times is reached, if so, stopping iteration, and outputting the optimal weight and bias value of the neural network to obtain an optimal neural network prediction model; otherwise, execution continues at steps SC3-SC 5.
CN202111413408.8A 2021-11-25 2021-11-25 Regional public parking lot charging demand prediction method based on parking big data Pending CN114139794A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111413408.8A CN114139794A (en) 2021-11-25 2021-11-25 Regional public parking lot charging demand prediction method based on parking big data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111413408.8A CN114139794A (en) 2021-11-25 2021-11-25 Regional public parking lot charging demand prediction method based on parking big data

Publications (1)

Publication Number Publication Date
CN114139794A true CN114139794A (en) 2022-03-04

Family

ID=80391716

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111413408.8A Pending CN114139794A (en) 2021-11-25 2021-11-25 Regional public parking lot charging demand prediction method based on parking big data

Country Status (1)

Country Link
CN (1) CN114139794A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114882729A (en) * 2022-04-22 2022-08-09 超级视线科技有限公司 Parking management method and system
CN117877313B (en) * 2024-03-12 2024-05-31 浙江宇泛精密科技有限公司 Parking lot management method and device based on Internet of things perception

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114882729A (en) * 2022-04-22 2022-08-09 超级视线科技有限公司 Parking management method and system
CN114882729B (en) * 2022-04-22 2023-12-08 超级视线科技有限公司 Parking management method and system
CN117877313B (en) * 2024-03-12 2024-05-31 浙江宇泛精密科技有限公司 Parking lot management method and device based on Internet of things perception

Similar Documents

Publication Publication Date Title
CN109523051B (en) Electric automobile charging real-time optimization scheduling method
CN112467722A (en) Active power distribution network source-network-load-storage coordination planning method considering electric vehicle charging station
CN110189025B (en) Electric vehicle charging station planning scheme acquisition method considering different load increases
CN111311122A (en) Charging station construction method for smart city
CN110866636A (en) Microgrid planning method comprehensively considering electric vehicle charging station and distributed energy
Qiu et al. Multi-service provision for electric vehicles in power-transportation networks towards a low-carbon transition: A hierarchical and hybrid multi-agent reinforcement learning approach
CN115063184A (en) Electric vehicle charging demand modeling method, system, medium, equipment and terminal
CN115171382A (en) Method for deploying driveways special for automatic driving of urban road network
CN114139794A (en) Regional public parking lot charging demand prediction method based on parking big data
CN109978236B (en) Small hydropower station short-term power refined prediction method based on feature combination
CN115204477A (en) Bicycle flow prediction method of context awareness graph recursive network
Wang et al. Multi-level charging stations for electric vehicles by considering ancillary generating and storage units
Farhadi et al. Data-driven multi-objective optimization for electric vehicle charging infrastructure
CN110929950B (en) Electric automobile load prediction method and system
CN117035185A (en) Electric vehicle charging station layout optimization method and system based on dynamic charging demand
CN116961057A (en) Multi-period power distribution network fault recovery method considering electric automobile
CN116826710A (en) Peak clipping strategy recommendation method and device based on load prediction and storage medium
CN116362523A (en) Coordinated optimization method for site selection and operation strategy of power exchange station considering temperature adaptability
CN109615142A (en) A kind of wind farm wind velocity combination forecasting method based on wavelet analysis
CN115222233A (en) Electric vehicle charging pile site selection and volume fixing method
CN115603446A (en) Power distribution station area operation monitoring system based on convolution neural network and cloud edge synergistic effect
CN115034608A (en) Distribution network risk assessment method based on distribution network element and neural network
CN113807019A (en) MCMC wind power simulation method based on improved scene classification and coarse grain removal
CN117994986B (en) Traffic flow prediction optimization method based on intelligent optimization algorithm
CN113241785A (en) Electric automobile charge and discharge polymerization optimal scheduling method and device

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