Disclosure of Invention
Aiming at the problem that the charging capacity is insufficient due to the fact that the charging load of an existing charging station is not matched with the charging load of an electric vehicle, the invention provides a power distribution network double-layer scheduling method based on dynamic matching of the electric vehicle and a charging facility. And the upper-layer EV charging time scheduling model determines the charging time of the electric automobile according to the charging requirement. And coordinating the centralized charging station and the distributed shared charging pile by the lower-layer EV charging position scheduling model to determine the charging position of the EV. The method and the device consider that each private charging pile aggregator in the sharing model mutually compete in the game based on the sharing quotation and the sharing cost, and determine the optimal sharing scheme of the charging piles when the game balance is achieved. And the EV double-layer space-time scheduling model is solved by adopting a dynamic self-adaptive feedback quantum particle swarm algorithm, so that the convergence of the solving process is better.
The technical scheme adopted by the invention is as follows:
the double-layer power distribution network scheduling method based on dynamic matching of the electric automobile and the charging facility is characterized by comprising the following steps of:
step 1, establishing an EV charging time scheduling model in a power distribution network based on EV battery charge state and charging demand trip data and with the goals of minimum load variance and minimum economic cost of the power distribution network as targets;
step 2, determining the capacity of a private charging pile to be shared in the power distribution network according to the EV charging time scheduling model, and establishing a private charging pile sharing model;
and 3, cooperating the centralized charging stations in each area and the shared private charging piles to charge the EVs, establishing an EV charging space scheduling model by taking the optimized distribution network voltage deviation extreme value as a target according to the EV charging time scheduling model, and determining the specific charging position and charging time of each EV.
The invention further comprises the following preferred embodiments:
in step 1, establishing an EV charging time scheduling model in the power distribution network includes:
the method comprises the steps of scheduling EV charging time based on the charge state of an EV battery, charging requirements and basic load data of a power distribution network, selecting the power distribution network with the minimum load variance and the minimum economic cost as scheduling targets, and reducing the load variance of the power distribution network and reducing the network loss and the load peak-valley difference by transferring EV loads of the power distribution network in each time period so as to determine the initial charging time of each EV when the load variance of the power distribution network is minimum.
In step 1, the intra-distribution-network EV charging time scheduling model is as follows:
in the formula (f)
lv、f
costF is the load variance of the power distribution network, the economic cost and the comprehensive objective function respectively; t denotes a scheduling period,
T sum1,2, … T represents that the scheduling cycle is divided into T periods; f. of
lvRepresenting the load variance, L, after an EV load has been connected to the distribution network
sum(t) represents the total load after the EV is accessed to the power distribution network in the t-th period; l is a radical of an alcohol
b(t) represents a base load of a t-th period; l is
c(t) a total EV load corresponding to a t-th period; l is a radical of an alcohol
avRepresents the average load over the entire scheduling period; p
eg(t) represents the exchange power of the distribution network and the external network, p
eg(t) represents the price of electricity purchased from the external power grid; p is a radical of
bFor the price of energy-storing batteries for distribution networks, Delta E
k(t) is the charge and discharge power of the energy storage battery, N
cFor the nominal number of cycles of the energy storage cell,
is the energy storage battery capacity; p
deg(t) is the power generated by the diesel engine, P
degNRated power for the firewood machine, a
1、a
2Is a polynomial first coefficient and second coefficient, p
gIs monovalent for diesel oil, c
eThe cost coefficient of the diesel engine power generation environment; f. of
lv,min、f
lv,maxRepresenting the minimum value and the maximum value of the load variance of the power distribution network; f. of
cost,min、 f
cost,maxRepresenting the minimum value and the maximum value of the economic cost of the power distribution network; omega
1、ω
2The weight coefficients of the first target and the second target are respectively. Wherein the content of the first and second substances,the energy storage battery does not comprise an Electric Vehicle (EV) battery.
The polynomial first coefficient a1The values are as follows: a is more than or equal to 0.15L/(kW.h)1≤0.21L/(kW.h);
Polynomial second coefficient a2The values of (A) are as follows: a is more than or equal to 0.22L/(kW.h)2≤0.28L/(kW.h);
First target weight coefficient omega1Weight coefficient omega of the second target2Should be a number between (0,1), and ω1>ω2。
The EV charging demand trip data is actively uploaded by an EV owner or is obtained by prediction based on user characteristic data.
Performing EV travel data prediction based on the user characteristic data, specifically comprising the following contents:
acquiring individual characteristic data and historical trip data of an electric vehicle in a power distribution network, and performing data preprocessing operation;
extracting and analyzing EV user characteristics by using a deep convolutional neural network;
extracting individual features of the EV user through convolution pooling operation, and establishing a mapping relation between the individual features of the EV user and travel features;
training the deep neural network by an error back propagation method;
and importing the EV user characteristic data into a trained deep convolutional neural network, mapping and outputting corresponding EV travel data, determining single EV load distribution according to the EV travel data, and accumulating to obtain total EV load distribution in the whole power distribution network.
The individual characteristic data of the electric automobile comprises the age, the gender, the position of a vehicle home region and the number of family members of an EV user; the travel data includes a stop start time, a driving range, and a stop duration after the EV reaches the destination.
The data preprocessing comprises processing of abnormal data and data standardization processing;
the method for processing the abnormal data comprising the defect data and the redundant data is a direct elimination method, and an extreme difference standardization method is adopted to convert the individual characteristic data of the electric automobile in the power distribution network and each type of characteristic data in the historical trip data set into data of a dimensionless unit:
in the formula, data (nu, D) represents the normalized nu sample D-th class feature data, and D belongs to {1, 2.. D }, namely the common D class feature data; data*(nu, d) represents the nu sample class d feature data before normalization.
After data standardization processing, the numerical value of each EV user characteristic data is between [0 and 1], but the EV user characteristic data are still numerous in variety, a data class with extremely small correlation with EV travel data is eliminated through correlation analysis, and the definition of a correlation coefficient and intermediate variables are as follows:
wherein Corrc (d)1,d2) Representing a class d of data1And class d of data2The correlation coefficient of (a); cov (d)1,d2) Representative data class d1And class d of data2The covariance of (a); var (d) is the standard deviation of class d data; average (d) is the average value of d types of data; corrc (d)1,d2) Is at [ -1,1 [)]When the correlation is larger than 0, the correlation is positive, when the correlation is smaller than 0, the correlation is negativeA value of 0 indicates complete irrelevance, and an absolute value of 1 indicates complete correlation.
In step 1, establishing an EV charging time scheduling model in a power distribution network, specifically including:
1.1 initializing each EV initial charging time as a population particle position;
1.2 calculating the load variance of the power distribution network according to the initialized EV initial charging time;
1.3, judging whether the EV initial charging time, the ending charging time, the driving mileage, the charging state and the EV load access total power meet EV charging time constraint, EV charging behavior continuity constraint, driving mileage constraint, EV charging state constraint and total power constraint of a power distribution network allowing EV load access, if not, returning to 1.1 to reinitialize each EV initial charging time, and if so, entering 1.4;
1.4 calculating the initial charging time of each EV when the load variance of the power distribution network is minimum, acquiring the individual optimal position and the group optimal position in the particle swarm, entering 1.5 when the lower convergence condition is not met, or entering 1.6:
position of kth generation particle population in formula
Wherein
Representing the position of the extreme value of the population of the kth-dimensional vector, | | | | | | non-calculation
2Represents the norm of L2 and epsilon represents the set convergence error.
1.5, after updating all particle positions and the initial charging time of each EV, continuing to execute 1.2-1.4;
and 1.6, finishing iteration, and obtaining EV load time distribution according to the initial charging time of each EV when the load variance of the power distribution network is minimum.
In 1.5, the particle position is updated according to the following equation:
in the formula (I), the compound is shown in the specification,
respectively corresponding to the h-dimension position of the (k + 1) -th generation e-th particle;
representing attractors of the h dimension of the kth generation of the e particle, each particle in the population iteration converges to the respective attractor
A random number represented on (0, 1);
representing the characteristic length of the h-dimension potential well of the kth generation e-particle;
the position of the extreme value of the body of the e-th particle in the h-th vector of the k generation is shown,
wherein, in the formula for updating the particle position, when
When the positive polarity is larger than 0.5, the positive polarity is taken out, otherwise, the negative polarity is taken out.
Further preferably, in 1.5, the particle position is updated according to the following formula:
in the formula (I), the compound is shown in the specification,
respectively corresponding to the h-dimension position of the (k + 1) -th generation e-th particle;
representing attractors of the h dimension of the kth generation of the e particle, each particle in the population iteration converges to the respective attractor
Representing adaptive feedback coefficients
The initial generation value of (a) is,
to describe the indirect variable of the distance variation factor of the particle,
the coordinates of the central point of the optimal position of all the particle individuals are obtained,
a random number represented on (0, 1);
representing the individual extreme position of the e-th particle in the h-th vector of the k generation,
the position of the group extreme value of the h-dimension vector of the kth generation is shown,
indirect variable SV
e upper,kDistance variation factor
Respectively calculated according to the following formula:
wherein, the first and the second end of the pipe are connected with each other,
represents the position of the kth generation of the e-th particle, | | | | | non-woven phosphor
2Represents the norm of L2, and the value range of mpu is (0, 0.01)]。
Further, in step 2, the private charging pile sharing model is preferably a private charging pile sharing model based on a non-cooperative game, the main body of the game is private charging pile aggregators at each position, game content, namely a power grid company provides shared offers for the aggregators at each position, the offers change along with changes of time, position and supply and demand relation of the vehicle piles, each aggregator develops the game by taking own shared income as a maximum target, and a game equilibrium solution is taken as a sharing scheme of each aggregator.
Further, in step 3, the EV charging space scheduling model determines the charging positions of all EVs by using the minimum voltage deviation extreme value of each key node after the EV load is connected to the power distribution network in each period as a scheduling target, and the expression f of the charging positionsvdIs represented as follows:
in the formula, Vdmax(t) represents the maximum voltage offset for the t-th period after EV access; vd(j, T) represents the voltage deviation of a key node j of the power distribution network in the T-th period after the EV is accessed, and T is the total period number in the scheduling time; j is as large as JsumJ represents a total of J grid node locations.
Further, in step 3, establishing an EV charging space scheduling model specifically includes:
3.1 determining EV loads of all charging positions in the current time period, and taking the EV loads as initialized population particle positions in the t-th time period;
3.2 calculating voltage deviation and voltage deviation extreme values of all nodes according to the particle positions, namely EV loads of all charging positions in the current time period;
3.3, judging whether the access EV load of each time interval of each region and the charging load of a charging station, the charging load of each node access shared charging pile of each time interval, and the distance from the charging pile of each node to the charging station of other regions meet the maximum charging power constraint of each time interval allowed by the access of the EV load of each region, the capacity constraint of the charging station and the shared private charging pile, the utilization rate constraint of the shared charging pile, the equation constraint of the EV charging load of each region and the scheduling range constraint of a region centralized charging station, if not, returning to 3.1 to reinitialize the EV load of each charging position of the current time interval, and if so, entering 3.4;
3.4 obtaining the individual optimal position and the group optimal position, comparing and recording the EV load of each charging position when the voltage deviation is minimum, and entering step 3.5 when the following convergence condition is not met, or entering step 3.6;
extreme position of kth generation particle population in formula
Wherein
Representing the position of the extreme value of the population of the kth-dimensional vector, | | | | | | non-calculation
2Represents the norm L2, and ε is the convergence error.
3.5 adding 1 to the iteration times, regenerating the particle position, and repeating for 3.1-3.4;
3.6, determining the EV load at each position when the voltage deviation extreme value in the T-th period is minimum, obtaining the spatial distribution of the EV load in the T-th period, and if the period T is less than the total period number T in the set scheduling time, returning T to be T +1 by 3.1;
3.7 determining the EV load spatial distribution in each time period according to the average voltage deviation extreme value in all the time periods.
Further, in 3.5, the particle position is updated according to the following equation:
in the formula (I), the compound is shown in the specification,
corresponding to the position of the h dimension of the (k + 1) th generation e particle in the EV charging space scheduling model;
representing attractors of h dimension of kth generation e particle in EV charging space scheduling model, wherein each particle in population iteration converges to each attractor
A random number represented on (0, 1);
representing the characteristics of h-dimension potential well of kth generation e-particle in EV charging space scheduling modelA length;
representing the position of an individual extreme value of the e-th particle in the h-th vector of the k generation in the EV charging space scheduling model,
the position of the group extreme value of the h-dimension vector of the kth generation is shown,
in step 3.5, the method specifically comprises the following steps:
calculating the average optimal position of the particles;
in the formula (I), the compound is shown in the specification,
representing the coordinates of the central point of the optimal position of all the particle individuals, and H represents the dimension of the particle; e represents the population size of the particles;
calculating a particle distance variation factor;
in the formula (I), the compound is shown in the specification,
represents the position of the e-th particle in the k +1 th generation, and the value range of mpu is (0, 0.01)]。
Adjusting compression-expansion factor
In the formula (I), the compound is shown in the specification,
τ
initialrepresent
The value of the first generation of the coefficients,
can be used as dynamic feedback information during particle iteration to realize
Self-adaptive adjustment of (2);
all particle positions are updated according to the compression-expansion factor:
the application also discloses a power distribution network double-layer scheduling system based on dynamic matching of the electric vehicle and the charging facilities by utilizing the power distribution network double-layer scheduling method, and the power distribution network double-layer scheduling system comprises an EV charging time scheduling model module, a private charging pile sharing model module and an EV charging space scheduling model module.
The EV charging time scheduling model module establishes an EV charging time scheduling model in a power distribution network based on travel data such as the charge state of an EV battery, the charging demand and the like, and aims at minimizing the load variance of the power distribution network and the economic cost, and transmits the EV load time distribution to a private charging pile sharing model module;
the private charging pile sharing model module determines the capacity range of the shared charging pile based on EV load time distribution and establishes a private charging pile sharing model based on a non-cooperative game;
the EV charging space scheduling model module is used for coordinating concentrated charging stations in various areas and shared private charging piles to charge the EV, and establishing an EV charging space scheduling model and determining the specific charging position of each EV according to the EV charging time scheduling model by taking the optimized distribution network voltage deviation extreme value as a target.
Further preferably, the power distribution network double-layer dispatching system further comprises an EV travel data acquisition and prediction module, wherein the EV travel data acquisition and prediction module is used for acquiring individual characteristic data and historical travel data of an electric vehicle in the power distribution network, extracting and analyzing EV user characteristics by using a deep convolutional neural network, mapping and outputting corresponding EV travel data, determining single EV load distribution according to the EV travel data, and accumulating to obtain total EV load distribution in the whole power distribution network; and transmitting the EV travel data to an EV charging time scheduling model module.
The power distribution network double-layer scheduling system further comprises a dynamic self-adaptive feedback quantum particle swarm algorithm module, and the EV charging time scheduling model and the EV charging space scheduling model in the power distribution network are optimized and solved.
The application also protects a terminal, which comprises a processor and a storage medium;
the storage medium is used for storing instructions; the processor is used for operating according to the instruction to execute the steps of the double-layer scheduling method for the power distribution network.
And protecting a computer readable storage medium, wherein a program, power distribution network basic data and a power distribution network domestic vehicle trip investigation data set which run on a terminal are stored, and the program is executed by a processor to realize the steps of the power distribution network double-layer scheduling method.
Compared with the prior art, the invention has the beneficial effects that:
1. an EV double-layer space-time scheduling algorithm cooperating with an original charging station and a shared charging pile is constructed from a power grid side. The algorithm optimizes the charging time and position of each EV in a layered partition manner, completes the coordination between the upper and lower layers, realizes the optimal matching of charging facilities and the EV on the premise of not needing to frequently expand the charging station, solves the problems of voltage out-of-limit and the like caused by concentrated charging of EV load, and improves the electric energy quality and the economy of a power distribution network. The impact of accessing a large number of EVs into the power grid can be avoided to the maximum extent by the power grid.
2. A charging pile sharing model based on a complete information static game is established, and optimal configuration of sharing capacity is achieved. The model constructs the optimization problem of the shared capacity as a complete information static game balance problem considering capacity coupling constraint, and further determines the optimal sharing scheme of the charging pile when generalized Nash balance is achieved.
3. On the basis of the global convergence of the quantum particle swarm algorithm, an improved algorithm is further provided, the compression-expansion factor coefficient is adaptively fed back and adjusted by the algorithm, the optimization of an EV scheduling scheme is realized, the solution precision of an EV scheduling model is improved, and the algorithm is suitable for a scene of charging a large-scale EV access power distribution network.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. The embodiments described herein are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by a person skilled in the art without any inventive step based on the spirit of the present invention are within the scope of the present invention.
As shown in the attached figure 1, the invention discloses a power distribution network double-layer scheduling method based on dynamic matching of an electric automobile and a charging facility, which comprises the following steps:
step 1, establishing an EV charging time scheduling model in a power distribution network based on EV battery charge state and charging demand trip data and with the goals of minimum load variance and minimum economic cost of the power distribution network as targets;
the intra-power-distribution-network EV charging time scheduling model is also referred to as an upper-layer EV charging time scheduling model in the embodiment of the present invention.
EV travel data are predicted based on a user characteristic data driven electric vehicle load calculation method. And the upper-layer power distribution network dispatching center dispatches the EV charging time based on the travel data such as the charge state of the EV battery, the EV charging demand and the like and the basic load data of the power distribution network. Due to the fact that large-scale EV disordered charging behaviors can cause the phenomenon of 'peak-to-peak' of load of a power distribution network, an upper-layer EV charging time scheduling model selects load variance and economic cost as scheduling targets, EV load in each time period of the power distribution network is transferred, variance is reduced, network loss and load peak-to-valley difference are reduced, cost is reduced, and economy and electric energy quality of operation of the power distribution network are improved. And the upper-layer scheduling model does not consider whether the EV load is matched with the charging facility or not, and only transfers the EV load to enable f to reach the minimum value, so that an EV charging time scheduling scheme is provided for the lower-layer scheduling model. According to the scheduling model of the EV charging time in the power distribution network, the power distribution network load variance is minimum, the economic cost is minimum, the scheduling target is selected, the EV load in each time period of the power distribution network is transferred, the power distribution network load variance is reduced, and the network loss and the load peak-valley difference are reduced, so that the initial charging time of each EV when the power distribution network load variance is minimum is determined.
According to the scheduling target of the EV charging time scheduling model, the skilled person can obtain the following scheduling model:
in the formula (f)lv、fcostAnd f are the load variance of the power distribution network, the economic cost and the comprehensive objective function respectively. T denotes a scheduling period, T sum1,2, … T represents that the scheduling cycle is divided into T periods; f. oflvRepresenting the load variance, L, of the EV load after it has been switched into the distribution networksum(t) represents the total load after the EV is accessed to the power distribution network in the t-th period; l is a radical of an alcoholb(t) represents a base load of a t-th period; l isc(t) a total EV load corresponding to a t-th period; l isavRepresenting the average load in the whole scheduling period; peg(t) represents the exchange power of the distribution network and the external network, peg(t) represents an external power grid purchase price; p is a radical ofbFor the price of energy-storing batteries for distribution networks, Delta Ek(t) is the battery charge-discharge power, NcIs the nominal cycle number of the battery, Ek maxIs the energy storage battery capacity; pdeg(t) is the power generated by the diesel engine, PdegNRated power for the firewood machine, a1、a2Respectively has a first coefficient and a second coefficient of the polynomial of more than or equal to a and 0.15L/(kW.h)1≤0.21L/(kW.h),0.22 L/(kW.h)≤a2Less than or equal to 0.28L/(kW.h). Preferred embodiments in the present applicationIn the examples, a1、a2Respectively set as 0.18L/(kW.h), 0.025L/(kW.h), pgIs monovalent for diesel oil, ceGenerating environmental cost coefficient for the diesel engine; f. oflv,min、 flv,maxRepresenting the minimum value and the maximum value of the load variance of the power distribution network; f. ofcost,min、fcost,maxRepresenting the minimum value and the maximum value of the economic cost of the power distribution network; omega1、ω2The weight coefficients of the first target and the second target are respectively. Wherein the energy storage battery does not comprise an EV battery. First target weight coefficient omega1The weight coefficient omega of the second target2Should be a number between (0,1), and ω1>ω2. In a preferred embodiment of the present application, ω1、 ω2The values are 0.55 and 0.45 respectively.
The model comprises EV charging time constraint, EV charging behavior continuity constraint, driving mileage constraint, EV state-of-charge constraint and total power constraint of a power distribution network allowing EV load access, and the EV charging behavior continuity constraint, the EV charging behavior continuity constraint and the total power constraint are respectively expressed as follows:
1) EV charge time constraints:
in the formula, NUsumRepresenting a set of EVs to be charged; t is tduration(nu) represents a charging duration of the nu-th vehicle EV; t isdwell(nu) represents the stay duration of the nu-th vehicle EV; t is tcharing(nu) represents a charging time of the nu-th vehicle EV; t is tec(nu) and tlc(nu) correspond to the stop start and end times of the nu-th EV, respectively.
2) EV charging behavior continuity constraint:
in the formula, T
start(nu) represents a charge start time of the nu-th vehicle EV;
and a state-of-charge flag indicating that the nu-th EV is in the t-th period, where the state-of-charge of the EV is 1, and otherwise, the state-of-charge of the EV is 0.
3) And (3) restriction of the driving mileage:
in the formula (d)s(nu) represents the mileage of the nu-th EV; dsmaxAnd represents the upper limit of the single driving mileage of the EV.
4) EV state of charge constraint:
in the formula, SOC (nu, t) represents the state of charge of the nu-th vehicle EV during the t-th period; SOCmaxAnd SOCminRepresenting upper and lower limits respectively corresponding to EV state of charge; pcp(nu) represents the charging power of the nu-th vehicle EV; cbatteryIndicating the EV battery capacity.
5) Total power constraint of the distribution grid to allow EV load access:
in the formula, Lpmax(t) represents the maximum charging power the distribution grid allows for EV access during the t-th period. According to the technical scheme, the EV travel data are actively uploaded by EV owners or are obtained through prediction based on user characteristic data. The embodiment of the invention discloses a data driver based on user characteristicsAnd (4) a preferable scheme for predicting the load of the electric automobile is implemented. The logic structure is shown in fig. 2:
1) and carrying out data preprocessing operation on the vehicle travel survey data set. The user data collected by the data set has typical big data characteristics: the source is various and heterogeneous, the data is numerous and complex, the calculation processing scale is large, and the value realization of the data depends on the high integration of the data. Therefore, before the data feature extraction and analysis, a data preprocessing operation is firstly performed on the data set.
In a preferred embodiment of the application, the user data comprises individual characteristic data and historical exit data of the electric vehicle, wherein the individual characteristic data of the user comprises the age, the sex, the position of a vehicle home area, the number of family members and the like of the EV user; the travel data includes a stop start time, a travel distance, a stop duration, and the like after the EV reaches the destination.
In order to make each data have comparability, the method for processing abnormal data such as defective data, redundant data and the like is a direct elimination method, and an extreme difference standardization method is adopted to convert individual characteristic data and each class of characteristic data in a historical output data set of an electric automobile in a power distribution network into data without dimensional units:
in the formula, data (nu, D) represents the normalized nu sample D-th class feature data, and D belongs to {1, 2.. D }, namely the common D class feature data; data*(nu, d) represents the nu sample class d feature data before normalization.
After data standardization processing, the numerical value of each EV user characteristic data is between [0 and 1], but the EV user characteristic data are still numerous in variety, a data class with extremely small correlation with EV travel data is eliminated through correlation analysis, and the definition of a correlation coefficient and intermediate variables are as follows:
wherein Corrc (d)1,d2) Representing a class d of data1And class d of data2The correlation coefficient of (a); cov (d)1,d2) Representative data class d1And class d of data2The covariance of (a); var (d) is the standard deviation of class d data; average (d) is the average value of d types of data; corrc (d)1,d2) Is at [ -1,1 [)]When the absolute value is 1, the correlation is positive, when the absolute value is greater than 0, the correlation is negative, when the absolute value is less than 0, the correlation is completely irrelevant, and when the absolute value is 1, the correlation is completely relevant.
2) A Deep Convolutional Neural Network (DCNN) is constructed to extract and analyze EV user features. The structure of a DCNN input layer, a convolution layer, a pooling layer, a full-connection layer and an output layer is respectively built, and a proper activation function and a proper loss function are selected. And setting super parameters such as DCNN sampling intervals, training iteration times, learning rate and the like, and initializing the weight and threshold of the DCNN.
3) Training the deep neural network by an error back propagation method, continuously comparing loss values between the predicted output value and the expected output value, and realizing adjustment and update of parameters such as DCNN weight values, threshold values and the like until the loss function meets the precision requirement.
4) Importing EV user characteristic data into a deep convolutional neural network which is trained, and mapping and outputting corresponding EV travel data; thereby accurately acquiring travel data of each EV.
As can be seen from the above description, in step 1 of the present application, establishing an EV charging time scheduling model in a power distribution network includes: the method comprises the steps of scheduling EV charging time based on the charge state of an EV battery, charging requirements and basic load data of a power distribution network, selecting the power distribution network with the minimum load variance and the minimum economic cost as scheduling targets, and determining the initial charging time of each EV when the load variance of the power distribution network is minimum by transferring the EV load in each period of the power distribution network, reducing the load variance of the power distribution network, and reducing the network loss and the load peak-valley difference.
Based on the above, it is clear to one of ordinary skill in the art that the aforementioned EV charge time scheduling model can be solved under constraint conditions expressed as a mixed integer nonlinear programming model. However, as the number of electric vehicles increases, the solution space of the scheduling model tends to be infinite, and in order to increase the calculation speed, the embodiment of the invention provides an optimization algorithm with higher convergence accuracy for solution. However, it should be noted that the solution of the EV charging time scheduling model by using the algorithm is only a preferred technical solution in the implementation process of the present invention, and is not a necessary limitation for implementing the power distribution network double-layer scheduling method based on the dynamic matching of the electric vehicle and the charging facility. The preferred scheme for solving the EV charging time scheduling model by adopting the optimization algorithm with higher convergence accuracy is as follows (see the attached figure 3):
in step 1, establishing an EV charging time scheduling model in a power distribution network, specifically including:
1.1 initializing each EV initial charging time as a population particle position;
1.2 calculating the load variance of the power distribution network according to the initialized EV initial charging time;
1.3, judging whether the EV initial charging time, the end charging time, the driving mileage, the charge state and the total EV load access power meet constraint conditions (6-12), if not, returning to 1.1 to reinitialize each EV initial charging time, and if so, entering 1.4;
1.4 calculating the initial charging time of each EV when the load variance of the power distribution network is minimum, acquiring the individual optimal position and the group optimal position in the particle swarm, entering 1.5 when the lower convergence condition is not met, or entering 1.6:
position of kth generation population in the formula
Wherein
The position of the extreme value of the group of the kth-dimension vector is expressed, | | | calving
2Represents L2 norm, and the convergence error ε is 10
-5。
1.5, after updating all particle positions and the initial charging time of each EV, continuing to execute 1.2-1.4;
the particle position is updated according to:
in the formula (I), the compound is shown in the specification,
respectively corresponding to the h-dimension position of the (k + 1) -th generation e-th particle;
representing attractors of the h dimension of the kth generation of the e particle, each particle in the population iteration converges to the respective attractor
A random number represented on (0, 1);
representing the characteristic length of the h-dimension potential well of the kth generation e-particle;
representing the position of the individual extreme value of the h-dimension vector of the e-th particle in the k generation,
in the formula for updating the position of the particle, when
When the positive polarity is larger than 0.5, the positive polarity is taken out, otherwise, the negative polarity is taken out.
In the present application, the particle position is updated according to the following equation:
in the formula (I), the compound is shown in the specification,
respectively corresponding to the h-dimension position of the (k + 1) -th generation e-th particle;
representing attractors of the h dimension of the kth generation of the e particle, each particle in the population iteration converges to the respective attractor
Representing adaptive feedback coefficients
The initial generation value of (a) is,
to describe the indirect variable of the distance variation factor of the particle,
the coordinates of the central point of the optimal position of all the particle individuals are obtained,
represents a random number on (0, 1);
representing the position of the individual extreme value of the h-dimension vector of the e-th particle in the k generation,
the position of the group extreme value of the h-dimension vector of the kth generation is shown,
wherein the indirect variables are
Distance change factor
Respectively calculated according to the following formula:
wherein the content of the first and second substances,
represents the k +1 thThe value of mpu is (0, 0.01) instead of the position of the e particle]. When the particle is
To
Distance ratio of
To
When the distance is large, the description will be given
And
the relative distance between them is increasing and the current position of the particle is
Possibly for a worse solution position, falling into a local optimum,
should be increased gracefully to enhance the algorithm's global search capability. When the particle is
To
Distance ratio of
To
Distance of (2) is small, show
And with
The relative distance between the particles is becoming smaller, and there may be a better solution in the solution space near the current particle, at which time the algorithm global search capability should be slowed down,
should be reduced moderately to enhance the local search capability of the algorithm.
Can be used as dynamic feedback information during particle iteration
Adaptive adjustment of (3).
And 1.6, finishing iteration, and obtaining EV load time distribution according to the initial charging time of each EV when the load variance of the power distribution network is minimum.
Step 2, determining the private charging pile capacity needing to be shared in the power distribution network according to the EV charging time scheduling model, and establishing a private charging pile sharing model;
fill the electric pile shared aim at and fill the breach of charging that the charging station produced when facing EV load peak period, improve the utilization ratio of idle electric pile of filling, increase private electric pile's sharing income, therefore the electric wire netting provides a nimble feasible sharing scheme and fills the prerequisite that electric pile owner participated in the shared service. Actually, the EV load and the charging station capacity do not reach a completely matched state, and the demands of the power distribution network for sharing the charging piles at different positions are different, so that the sharing quotation provided by the power distribution network to the charging pile owner is different, and a competitive relationship exists among private charging pile aggregators belonging to different positions.
The game model is generally composed of three elements: participants, policies of the participants, and revenue functions. The participators of the game model shared by the private charging piles are private charging pile aggregators at each position, J belongs to JsumJ represents that a private charging pile aggregator with J distribution network node locations in total participates in the game. Private charging pileThe strategy of the shared game model is that the shared capacity provided by the private charging pile aggregators at each position is selected, and the theoretically sharable charging pile capacity range is represented as follows:
in the formula (I), the compound is shown in the specification,
indicating the shared capacity selected by the charging pile aggregator at node j;
representing the available charging pile capacity of the charging pile aggregator for node j.
The process of the game of each private charging pile aggregator in the invention is a complete information static game process, and the private charging pile aggregator of the node j refers to the compensation price
Compensating price depreciation coefficients
And sharing compensation quotes
Etc. are fully disclosed. According to the market economic supply theory, along with the increase of the number of the charging piles participating in the sharing service, the demand of the power distribution network on the charging piles is correspondingly reduced, and each position is provided with
Will decrease with increasing shared capacity, as shown below:
the income objective function of each node aggregator in the private charging pile sharing game model is expressed as follows:
in the formula (I), the compound is shown in the specification,
representing the sharing income of the node j charging pile aggregator; Δ t represents a unit time length;
and representing the charging pile sharing cost of the node j.
The present invention labels all participants except participant j (the aggregator for node j) as the opponents of participant j, as "-j". Whatever strategy the other participants-j make
Participant j will choose a strategy that maximizes his own profit. Since private charging pile aggregators do not harm their own interests for charging pile aggregators in other locations, the essence of sharing the gaming model is a complete information static gaming problem.
If the policy made by participant j
Enable an objective function
The policy is the optimal policy of the participant j:
in the formula (I), the compound is shown in the specification,
the optimal strategy for representing participant j, if for any participantBoth satisfy the following formula:
i.e. any policy chosen by the participant j
Are all no better than
Under the condition that any one participant in the game determines the strategy by other participants, the selected optimal strategy is the optimal strategy
Is defined as a game equilibrium solution. Each participant in the game equilibrium solution interacts with other participants based on the objective function, but the participants are mutually independent in the strategy space. Namely, the sharing strategy selected by each aggregator is the most beneficial strategy for the aggregator, and the strategy is used as a charging pile sharing scheme.
The charging demand of the EV is limited, so the capacity demand of the power distribution network on the shared charging pile has a certain range, as shown in formula (18), the power distribution network needs to ensure that the charging piles participating in the shared service cannot be excessively redundant, and the charging demand of the EV cannot be met even after the power distribution network supplements the shared charging pile.
In the formula, PdcmaxAnd PdcminRepresenting the maximum and minimum capacity requirements of the distribution grid for the shared charging poles.
And 3, cooperating the centralized charging stations in each area and the shared private charging piles to charge the EVs, establishing an EV charging space scheduling model by taking the optimized distribution network voltage deviation extreme value as a target according to the EV charging time scheduling model, and determining the specific charging position and charging time of each EV.
In the present invention, the EV charging time scheduling model is also referred to as a lower layer EV charging space scheduling model. After the EV charging time scheduling model at the upper layer determines the EV charging time scheduling scheme, issuing the P in the private charging pile sharing playing modeldcmin(t) and PdcmaxAnd (t), determining the capacity requirement of the shared charging pile at each time interval of the power distribution network, and determining a charging pile sharing scheme achieving game balance by charging pile aggregators at each position. And the centralized charging stations in the lower layer and the regions cooperate with the original charging stations and the private shared charging piles in the regions to provide charging service for the EV according to the EV charging time scheduling scheme and the private charging pile sharing scheme, so that the EV charging position scheduling is completed.
The lower-layer scheduling model takes the minimum voltage deviation extreme value of each time interval after the EV load is connected into the power distribution network as a scheduling target, and an expression f of the lower-layer scheduling modelvdIs represented as follows:
in the formula, Vdmax(t) represents the maximum voltage offset for the t-th period after EV access; vdAnd (j, t) represents the voltage offset of the distribution network node j in the t-th period after the EV is accessed.
The model comprises maximum charging power constraint allowing EV load access in each time period of each region, capacity constraint of the charging stations and shared private charging piles, utilization constraint of the shared charging piles, equation constraint of EV charging loads of each region and scheduling range constraint of a region centralized charging station, namely the region centralized charging station can only use charging facilities in the region to charge the EV.
1) And (3) the maximum charging power constraint of EV load access allowed in each period of each region:
in the formula, Pc(r, t) represents the r-th region EV load during the t-th period; p iscmax(r, t) represents the maximum accessible EV load of the r-th region during the t-th period; pcsmax(r, t) represents the maximum available capacity of the mth zone charging station for the tth period; pdc(r, t) represents the shared charging pile capacity of the r-th area at the t-th time period; rsum{1,2, … R } represents a set containing R sub-regions.
2) Charging station, the capacity constraint of sharing private electric pile:
in the formula, Pcsload(r, t) a charging load corresponding to the mth zone charging station for the tth time period; pdcloadAnd (j, t) represents the charging load of the j node accessed to the shared charging pile in the t period.
3) The electric pile utilization ratio constraint is filled in the sharing:
4) the EV charging load equation constraint of each area and the dispatching range constraint of the area centralized charging station are as follows:
in the formula, Lc(t) represents the total EV load for the t-th period; ddc(j, r) represents the distance from the charging pile belonging to the node j to a centralized charging station in the region r; ddc(j, other) represents the distance from the charging pile belonging to node j to the other regional centralized charging station.
The EV charging space dispatching model function can be solved by using the existing intelligent optimization algorithms such as particle swarm search, genetic algorithm and the like, and the expected basic technical effect can be obtained, but in order to obtain a faster solving speed, referring to the attached figure 3, the EV charging space dispatching model function is solved by using the optimization algorithm with higher convergence precision, which is specifically as follows:
in step 3, solving the EV charging space scheduling model in the power distribution network specifically includes:
3.1 determining EV loads of all charging positions in the current time period, and taking the EV loads as initialized population particle positions in the t-th time period;
3.2 calculating voltage deviation and voltage deviation extreme values of all nodes according to the particle positions, namely EV loads of all charging positions in the current time period;
3.3 determination of Pc(r,t)、Pcsload(r,t)、Pdcload(j,t)、Ddc(j, r) whether constraint conditions (21-27) are met, if not, returning to 3.1 to reinitialize EV loads of all charging positions in the current time period, and if so, entering 3.4;
3.4 obtaining the individual optimal position and the group optimal position, comparing and recording the EV load of each charging position when the voltage deviation is minimum, and entering a step 3.5 when the following convergence condition is not met, or entering a step 3.6:
position of kth generation particle population in formula
Wherein
Representing the position of the extreme value of the population of the kth-dimensional vector, | | | | | | non-calculation
2Represents L2 norm, and the convergence error ε is 10
-5。
3.5 adding 1 to the iteration number k to regenerate the particle position, and repeating for 3.1-3.4;
the particle position is updated according to:
in the formula (I), the compound is shown in the specification,
respectively corresponding to the h-dimension position of the (k + 1) -th generation e-th particle;
representing attractors of the h dimension of the kth generation of the e particle, each particle in the population iteration converges to the respective attractor
A random number represented on (0, 1);
representing the characteristic length of the h-dimension potential well of the kth generation e-particle;
representing the position of the individual extreme value of the h-dimension vector of the e-th particle in the k generation,
in the formula for updating the position of the particle, when
When the positive polarity is larger than 0.5, the positive polarity is taken out, otherwise, the negative polarity is taken out.
In the present application, the particle position is updated according to the following steps:
calculating the average optimal position of the particles;
in the formula (I), the compound is shown in the specification,
representing the coordinates of the central point of the optimal position of all the particle individuals, and H represents the dimension of the particle; e represents the population size of the particles.
Calculating a particle distance variation factor;
in the formula (I), the compound is shown in the specification,
represents the position of the e-th particle in the k +1 th generation, and the value range of mpu is (0, 0.01)]. When the particle is
To
Distance ratio of
To
When the distance is large, the description will be given
And
the relative distance between them is increasing and the current position of the particle is
Possibly for a worse solution position, falling into a local optimum,
should be increased gracefully to enhance the algorithm's global search capability. When the particle is
To
Distance ratio of
To
Distance of (2) is small, show
And
the relative distance between the particles is becoming smaller, and there may be a better solution in the solution space near the current particle, at which time the algorithm global search capability should be slowed down,
should be reduced moderately to enhance the local search capability of the algorithm.
Adjusting compression-expansion factor
In the formula (I), the compound is shown in the specification,
τ
initialto represent
The initial value of the coefficient.
Can be used as dynamic feedback information during particle iteration to realize
Adaptive adjustment of (3).
All particle positions are updated according to the compression-expansion factor.
3.6, determining the EV load at each position when the voltage deviation extreme value in the T-th period is minimum, obtaining the spatial distribution of the EV load in the T-th period, and if the period T is less than the total period number T in the set scheduling time, returning T to be T +1 by 3.1;
3.7 determining the EV load spatial distribution in each time period according to the average voltage deviation extreme value in all the time periods.
The application also discloses a power distribution network double-layer scheduling system based on dynamic matching of the electric automobile and the charging facilities by utilizing the power distribution network double-layer scheduling method, and the power distribution network double-layer scheduling system comprises an EV charging time scheduling model module, a private charging pile sharing model module and an EV charging space scheduling model module.
The EV charging time scheduling model module establishes an EV charging time scheduling model in a power distribution network based on travel data such as the charge state of an EV battery, the charging demand and the like, and aims at minimizing the load variance of the power distribution network and the economic cost, and transmits the EV load time distribution to a private charging pile sharing model module;
the private charging pile sharing model module determines the capacity range of the shared charging pile based on EV load time distribution and establishes a private charging pile sharing model based on a non-cooperative game;
the EV charging space scheduling model module is used for coordinating concentrated charging stations in various areas and shared private charging piles to charge the EV, and establishing an EV charging space scheduling model and determining the specific charging position of each EV according to the EV charging time scheduling model by taking the optimized distribution network voltage deviation extreme value as a target.
Furthermore, the power distribution network double-layer dispatching system also comprises an EV travel data acquisition and prediction module, wherein the EV travel data acquisition and prediction module is used for acquiring individual characteristic data and historical travel data of an electric vehicle in the power distribution network, extracting and analyzing EV user characteristics by using a deep convolutional neural network, mapping and outputting corresponding EV travel data, determining single EV load distribution according to the EV travel data, and accumulating to obtain total EV load distribution in the whole power distribution network; and transmitting the EV travel data to an EV charging time scheduling model module.
In order to obtain a faster calculation effect, the power distribution network double-layer scheduling system further comprises a dynamic self-adaptive feedback quantum particle swarm algorithm module for performing optimization solution on an EV charging time scheduling model and an EV charging space scheduling model in the power distribution network.
The present disclosure may be systems, methods, and/or computer program products. The computer program product may include a computer-readable storage medium having computer-readable program instructions embodied thereon for causing a processor to implement various aspects of the present disclosure.
The computer readable storage medium may be a tangible device that can hold and store the instructions for use by the instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, semiconductor memory device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a Static Random Access Memory (SRAM), a portable compact disc read-only memory (CD-ROM), a Digital Versatile Disc (DVD), a memory stick, a floppy disk, a mechanical coding device, such as punch cards or in-groove projection structures having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media as used herein is not to be construed as transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission medium (e.g., optical pulses through a fiber optic cable), or electrical signals transmitted through electrical wires.
The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to a respective computing/processing device, or to an external computer or external storage device via a network, such as the internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. The network adapter card or network interface in each computing/processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in the respective computing/processing device.
The computer program instructions for carrying out operations of the present disclosure may be assembler instructions, Instruction Set Architecture (ISA) instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider). In some embodiments, the electronic circuitry can execute computer-readable program instructions to implement aspects of the present disclosure by utilizing state information of the computer-readable program instructions to personalize the electronic circuitry, such as a programmable logic circuit, a Field Programmable Gate Array (FPGA), or a Programmable Logic Array (PLA).
Various aspects of the present disclosure are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the disclosure. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer-readable program instructions.
These computer-readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer-readable program instructions may also be stored in a computer-readable storage medium that can direct a computer, programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer-readable medium storing the instructions comprises an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.
The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer, other programmable apparatus or other devices implement the functions/acts specified in the flowchart and/or block diagram block or blocks.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
Experimental examples and analysis: the invention is based on MATLAB R2016b programming, the charging pile sharing scheme is solved by a smooth Newton method, and the EV hierarchical scheduling model is solved by the particle swarm optimization algorithm. The convergence accuracy of the smooth Newton method is set to 10-5. The maximum iteration of the particle swarm optimization algorithm is 200 generations, the swarm comprises 20 particles, and the initial value of the adaptive feedback coefficient is set to be 0.5. By adopting IEEE33 node system simulation, the load of the power distribution network is obtained by multiplying the per unit value of load data of a certain day in 2018 of a power company in Shaoyang city of Hunan province by the active load of each node of IEEE33, and the change trend of the daily load of the power distribution network can be reflected. The total basic load of each time interval of the power distribution network is shown in table 1, and the basic load distribution of each node of each time interval is shown in fig. 5. The charging station position is shown in FIG. 6, after roundingSee table 2 for charging station capacity and service range data. The number of the private charging piles available in each position is shown in a table 3, the private charging piles adopt a slow charging mode to charge for the EV, and the charging power is 3 kW.
TABLE 1 Total base load of distribution network at each time interval
TABLE 2 charging station building Capacity and service Range
Table 3 number of available private charging piles at each location
The invention sets three groups of strategies for comparison experiment:
strategy 1: EV disordered charging; the EV disordered charging means that the power grid dispatching is not accepted when the charging requirement exists, and the charging is directly carried out at a charging station;
strategy 2: EV traditional scheduling shared by charging piles is not considered; different from disordered charging, the conventional EV scheduling without considering shared charging means that an EV receives power grid scheduling and is charged only in a charging station;
strategy 3: EV double-layer charging space-time scheduling considering a charging pile sharing game is adopted.
The load curve after the disordered charging and the upper-layer time scheduling is shown in fig. 7, and the voltage deviation extreme values of each time period of the distribution network under the three strategies are shown in tables 4, 5 and 6.
As shown in fig. 7, in the case of chaotic charging, most EVs are charged in the 14 th to 24 th periods, the load after the EV is accessed is the largest in the 18 th period, the smallest in the 5 th period, the load peak-to-valley difference is 2347kW, and the load variance is 658321. In the disordered charging state, the EV is not charged in the load valley period, but a new load peak is formed on the basis of the original load peak period, the load peak value of the power distribution network in the disordered charging of the EV reaches 4640kW, and the load peak value is increased by 24.80 percent compared with the peak value when the EV is not connected for charging, so that great power supply pressure is brought to the power distribution network.
According to the difference strategy 1, strategy 2 and strategy 3, through upper-layer time scheduling, most EV loads are transferred to a valley period, loads after access of the EV are the largest in the 18 th period, the loads are the smallest in the 2 nd period, the maximum load is 3897kW, and the minimum load is 3037 kW. The load peak-valley difference is only 860kW, the load peak value is reduced by 17.02% compared with the disordered charging, the total load variance is 97017, the load variance is reduced by 561304 compared with the disordered charging, the blank of the load valley period is filled after the EV load is scheduled by the upper layer, and the adverse effect of 'peak-up and peak-up' of a power distribution network caused by the disordered charging of the EV is also effectively avoided.
Based on table 4, it can be seen that, in the disordered charging state, even though the charging station can accept a large number of EVs for centralized charging, the transformation of the distribution network line and the capacity expansion scheme of the transformer are difficult to follow in a short time, so that the voltages of all nodes of the distribution network charged by the EVs accessed to the charging station in a centralized manner generally decrease, the voltage deviation extreme values in the periods from 13 th to 24 th exceed-7%, the average voltage deviation extreme value in the period from 24 th to 24 th reaches-7.51%, the maximum voltage deviation extreme value reaches-11.13%, and the EV centralized charging will cause great negative effects on the operation of the distribution network.
As can be seen from table 5, although the upper-layer time scheduling transfers the charging time of most of the EV loads and reduces the load variance, the total charging demand of the EV does not change, which still does not solve the problem of shortage of charging facilities in some regions due to the uneven spatial distribution of the EVs, and the capacity of each charging station in each region cannot fully satisfy the charging demand of the EV loads. If the lower-layer scheduling model does not supplement the shared charging pile to charge the EV, the spatial position of the EV load accessed to the power distribution network is still not changed, and the problem of voltage out-of-limit caused by concentrated charging of the EV in the load peak period is still not solved.
Comparing table 4, table 5 analysis, regional concentrated type charging station of lower floor has sufficient facility of charging to the EV charges, has filled the charging gap of EV load peak period, and the EV load no longer concentrates on 3 charging stations to charge, and then the voltage offset of each node is less relatively in each period. The average voltage deviation extreme value in the 24 periods is-5%, the average voltage deviation extreme value is reduced by 2.51% compared with the average voltage deviation extreme value in the disordered charging process, and the optimization effect of the lower-layer scheduling model after the shared charging piles are supplemented based on the sharing scheme is extremely obvious.
The double-layer scheduling method for the power distribution network based on dynamic matching of the electric automobile and the charging facilities, provided by the invention, realizes the cooperation of a centralized charging station and a distributed idle charging pile, determines the charging position and the charging time of the EV by using a double-layer scheduling algorithm on the premise of not frequently expanding the capacity of the charging station and supporting facilities thereof, and solves the problems of charging gaps existing in the peak period of the EV load of the original charging station and voltage overrun caused by centralized access of the EV load.
Table 4 voltage deviation extreme value of distribution network in each period under strategy 1
Table 5 voltage deviation extreme value of distribution network in each period under strategy 2
Table 6 strategy 3 lower distribution network voltage deviation extreme value in each period
Finally, it should be noted that: although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.