CN106355290A - Electric vehicle charge load prediction method and system based on Markov chain - Google Patents

Abstract

The invention discloses an electric vehicle charge load prediction method and system based on Markov chain; the method comprises the steps of setting original data information and a number of simulations; determining decision-making behaviors; accumulating charge load curves to obtain total charge load; judging whether the number of currently performed simulations is greater than a set number of simulations or not; judging whether the total charge load acting as simulation results is converged or not; if yes, outputting the simulation results; if not, returning to reset the number of simulations; based on Markov theory, stage of charge value of a power battery of an electric vehicle is considered as a stage variable in Markov chain, the method provides a visual indication for changes in the state of charge value of the power battery during one-day outgoing of a user so as to determine real-time charging need in this process, charge load is calculated through Monte Carlo simulation on such basis, and therefore load prediction results obtained by using the method are more practical.

Description

One kind is based on markovian charging electric vehicle load forecasting method and system

Technical field

The present invention relates to charging electric vehicle load prediction field is and in particular to a kind of be based on markovian electronic vapour Car charging load forecasting method.

Background technology

Increasingly serious with energy crisis and environmental problem, electric automobile has become as the future automobile energy and environmental protection Development trend." the accelerating the instruction of new-energy automobile popularization and application " put into effect according to State Council, to the year two thousand twenty, the new energy of China Source car production and marketing is up to 5,000,000, and charging pile quantity is up to 4,500,000 it is seen that electric automobile quantity will have in the coming years There is suitable scale.Due to the aggregation properties of charging electric vehicle, and the uncertainty of automobile user demand and behavior With mutual difference, large-scale electric automobile accesses, and will bring huge challenge to the safe operation of electrical network and Optimized Operation, right The operation of the planning of power system, operation and electricity market produces profound influence, therefore, sets up effective charging electric vehicle Load forecasting model, the impact bringing for analysis charging electric vehicle load and electric automobile extensively access the regulating strategy of electrical network Theoretical foundation is provided.

Do numerous studies for charging electric vehicle load modeling both at home and abroad, be broadly divided into three classes: based on electric automobile Using the Monte Carlo Analogue Method of trip requirements, analyze the general of the charging electric vehicle power arriving at charging station using queuing theory Rate analytic process and physical analysis method.Wherein, the statistics modeling method application based on Monte Carlo simulation is more universal, generally examines Consider charging electric vehicle workload demand be subject to charging electric vehicle pattern, scale, moving law, battery behavior and electricity price regulation etc. because The impact of element, sets up the load model based on probability statistics.Research great majority are in given starting impetus battery charge state Set up simple model under conditions of (state of charge, soc), initiation of charge time and daily travel, ignore use The motility of family charging behavior and randomness, load prediction subjective, not from the feature of electric automobile itself, Real time charging demand during reflection user's trip, predicts the outcome and can not fully reflect the workload demand of reality.

Content of the invention

The technical problem to be solved is to provide one kind pre- based on markovian charging electric vehicle load Survey method, the method is based on Markov theory, and electric automobile power battery SOC is considered as in Markov Chain State variable, calculates charging load, therefore, the load prediction that obtains using the method by Monte Carlo simulation on this basis Result more conforms to practical situation.

The technical scheme is that

One kind is based on markovian charging electric vehicle load forecasting method it is characterised in that including step:

(1) setting primary data information (pdi) and number realization；

(2) battery charge according to primary data information (pdi) extract unique user one day travel time and trip moment first State value；

(3) extract single distance travelled, and calculate the battery lotus at the end of this section of stroke end time and this section of stroke Electricity condition value；

(4) judge whether the battery charge state value at the end of described stroke is less than the threshold value setting, if less than setting Threshold value, then execution step (5), otherwise, then execution step (7),

(5) randomly draw user's single charge duration, at the end of judging charging modes that user may take and charging Battery charge state value and end time；Record the battery lotus at the end of this charging duration, charging modes, and charging simultaneously Electricity condition value and end time；

(6) according to the battery charge state value at the end of this decision behavior and end time, judge that next step may be taken Decision behavior, if next decision behavior be stop, execution step (7)；If next decision behavior is to travel, execute Step (9)；

(7) randomly draw user's single parking duration, record battery charge state value and the knot at the end of this parking behavior The bundle time；

(8) described end time and the trip end time on the same day setting are compared, if the end time is more than or equal to setting The trip end time, then execution step (11), otherwise execution step (9)；

(9) extract single distance travelled, record the battery charge state value at the end of this traveling behavior and end time；

(10) above-mentioned end time and the trip end time on the same day setting are compared, if the end time is more than or equal to setting The trip end time, then execution step (11), otherwise execution step (4)；

(11) respectively add up fill soon, the charging load curve of trickle charge, obtain total bulk charging load；

(12) whether the number of times judging at present simulation is more than the number realization setting in step (1)；If so, then enter step Suddenly (13)；If it is not, then return to step (2) continues executing with；

(13) judge whether restrain as total bulk charging load of analog result；If convergence, export analog result；If no Convergence, then return to step (1) resets number realization.

The invention has the beneficial effects as follows this method is based on Markov theory, by electric automobile power battery SOC It is considered as the state variable in Markov Chain, automobile user trip process power battery charged shape under different decision behaviors The change procedure of state value is described by markovian state conversion process, and the method visually reflects user one day The state change of power battery charged state value during trip, and then determine the real time charging demand of this process, here basis Above charging load is calculated by Monte Carlo simulation, therefore, more conform to reality using the load prediction results that the method obtains Situation.

On the basis of technique scheme, the present invention can also do following improvement.

Further, the primary data information (pdi) in described step (1) includes:

Fill soon, the charge power of trickle charge, the battery capacity of electric automobile, the average overall travel speed of vehicle, every 100 kms consumption Electricity；

Initial time and initial SOC that electric automobile is gone on a journey in one day first；Travel every time and terminate rear battery lotus The threshold value of electricity condition；

The probability density function of user's single operating range；

The probability density function of the single charge duration of electric automobile；

The probability density function of single parking duration；

One day trip end time of user.

Further, the extraction single distance travelled in described step (3), and calculate this section of stroke end time and this section Battery charge state value at the end of stroke, the calculating of described battery charge state value is drawn using equation below:

${\mathrm{soc}}_j={\mathrm{soc}}_i-\frac{d_n{w}_{100}}{q_k}$

Wherein, d_nFor user's single operating range, w₁₀₀For the every 100 km power consumption of electric automobile, q_kFor the electronic vapour of kth The battery capacity of car, soc_i- for dnw1 electricity 00 electrical automobile any time battery charge state value, soc_j: for electric automobile at certain Plant the battery charge state value of the subsequent time under decision behavior.

Further, judge that the charging modes detailed process that user takes is in described step (5),

Charging duration scope is soon:

$\frac{0.2q_k}{p_k}\&le;t_i<\frac{q_k}{p_k}$

Trickle charge charging duration scope is:

$\frac{0.2q_k}{p_m}\&le;t_i<\frac{q_k}{p_m}$

It is assumed that p_k=5p_m, then

$\frac{q_k}{p_k}=\frac{0.2q_k}{p_m}=t_c$

It is a length of when user chargesThen take fast charge mode,

It is a length of when user chargesThen take trickle charge mode,

Wherein, p_k、p_mBe respectively fill soon, the charge power of trickle charge, t_iFor charging duration, q_kElectricity for kth electric automobile Tankage.

Further, the battery charge state value at the end of the charging in described step (5) draws according to formula below:

${\mathrm{soc}}_j=\left\{\begin{array}{cc}{\mathrm{soc}}_i+\frac{p_k{t}_i}{q_k}& a_i=1++\\ {\mathrm{soc}}_i+\frac{p_m{t}_i}{q_k}& a_i=1+\end{array}\right.$

Wherein, a_i=1++ represents and fills soon, a_i=1+ represents trickle charge, p_k、p_mBe respectively fill soon, the charge power of trickle charge, t_iFor Charging duration, q_kFor the battery capacity of kth electric automobile, soc_iFor the battery charge state value of electric automobile any time, soc_j: for the battery charge state value of electric automobile subsequent time at any time.

Further, in described step (11) respectively add up fill soon, the charging load curve of trickle charge, obtain total bulk charging bear Lotus, concrete calculating process is as follows:

A length of during charging soon:

$t_{i,1}=\frac{{\mathrm{soc}}_{i,1}-{\mathrm{soc}}_{i,0}}{p_k}{q}_k$

Trickle charge is a length of when charging:

$t_{i,2}=\frac{{\mathrm{soc}}_{i,1}-{\mathrm{soc}}_{i,0}}{p_m}{q}_k$

Then the charge power that fills soon of arbitrarily t is:

The trickle charge charge power of arbitrarily t is:

Wherein, t_i,1、t_i,2Represent respectively fill soon, trickle charge charging duration, soc_i,0Proceed by the lotus of charging for electric automobile Electricity condition, soc_i,1For the state-of-charge at the end of charging, t₀For starting to charge up the moment,

Charging electric vehicle power is added up, can predict fill soon, trickle charge load and total load, computing formula divides Not as follows:

$p_{t,1}=\underset{k=1}{\overset{n_1}{\&sigma;}}{p}_k$

$p_{t,2}=\underset{k=1}{\overset{n_2}{\&sigma;}}{p}_m$

$p_t=p_{t,1}+p_{t,2}=\underset{k=1}{\overset{n_1}{\&sigma;}}{p}_k+\underset{k=1}{\overset{n_2}{\&sigma;}}{p}_m$

Wherein, p_t,1、p_t,2Respectively represent t fill soon, trickle charge load, p_tFor the total charging load of t electric automobile, N1, n2 are natural number, represent the vehicle number filling soon with trickle charge respectively.

According to another aspect of the present invention, one kind is also provided to be based on markovian charging electric vehicle load prediction System is it is characterised in that include information setup module: for inputting primary data information (pdi), and sets number realization；

First abstraction module: for extracting unique user travel time and trip first in a day from primary data information (pdi) The battery charge state value in moment；

Second abstraction module: for extracting single distance travelled, and calculate this section of stroke end time and this section of stroke At the end of battery charge state value；

First comparison module: for the battery charge state value at the end of relatively described stroke whether less than the threshold setting Value；

3rd abstraction module: be used for randomly drawing user's single charge duration, and judge the charging side that user may take Battery charge state value at the end of formula and charging and end time；Record this charging duration, charging modes, Yi Jichong simultaneously Battery charge state value at the end of electricity and end time；

Decision-making module: for according to the battery charge state value at the end of this decision behavior and end time, judging next Walking the decision behavior taken is to stop or travel；

4th abstraction module: be used for randomly drawing user's single parking duration, record the battery at the end of this parking behavior SOC and time；

Second comparison module: for the trip end time on the same day of relatively described parking behavior end time and setting；

4th abstraction module: for extracting single distance travelled, record the battery charge state at the end of this traveling behavior Value and end time；

3rd comparison module: for the trip end time on the same day of relatively described traveling behavior end time and setting；

Accumulator module: for adding up respectively to fill soon, the charging load curve of trickle charge, obtain total bulk charging load；

4th comparison module: whether the number of times for relatively simulation at present is more than the number realization setting；

Judge module: for judging whether restrain as total bulk charging load of analog result.

Brief description

Fig. 1 is electric automobile state-of-charge transfer process schematic diagram of the present invention；

Fig. 2 is based on markovian load prediction flow chart for the present invention.

Specific embodiment

Below in conjunction with accompanying drawing, the principle of the present invention and feature are described, example is served only for explaining the present invention, and Non- for limiting the scope of the present invention.

It is illustrated in figure 1 electric automobile state-of-charge transfer process schematic diagram of the present invention, dynamic using Markov Chain description The change procedure of power battery charge state soc:

Markov, with the naming of Russia mathematician markov, markoff process be have markov property with Machine process, in the case of given current knowledge or information, the past (i.e. current former historic state) is for prediction in the future (i.e. Current later to-be) it is unrelated.Can be applicable to the research neck such as the hydrology, meteorology, earthquake, economic forecasting, management decision-making Domain.

Markov Chain (markov chain) is the discrete time stochastic process in mathematics with Markov property.It The model of one class stochastic systems is described, it is random for referring to system in each time state in which, from current time to The state of one time presses certain probability transfer, and to-be is only relevant with status praesenss and its transition probability, and with former State is unrelated, i.e. markov property.

According to Markov theory, if the soc value of electric automobile any time is considered as state s_i, under electric automobile for the moment The state-of-charge soc value carved is considered as state s_jIt is known that s_jState s only with current time_iAnd corresponding transition probability p_ijIt is relevant, So the state-of-charge change during electric automobile trip in a day can be described by Markov Chain.Due to various types of In type electric automobile, the proportion shared by private car is larger, and the charging behavior of private car user has very big motility and not true Qualitative, during trip, the change procedure of electrokinetic cell soc is complex.Therefore, this patent is mainly used using Markov Chain To analyze in private car one day soc value changes situation thus predicting charging load.

In figure a_iDifferent values represent different decision behaviors, wherein, a_i=-1 expression electric automobile during traveling；a_i=1+ +、a_i=1+ represents that electric automobile is filled and trickle charge soon respectively；a_i=0 expression is not charged and is not also travelled.Solid arrows represent State transfer case under certain decision behavior, heavy solid line arrows represent under current state, the possible decision behavior of next step, empty If line arrow represents is currently not one day end-state s_nWhen, the possible decision behavior of next step.p_ijSoc value for electric automobile Switch to the state transition probability of next state from preceding state, it has different representations as the case may be.

(1) decision behavior that user may take.State-of-charge transition probability p due to electric automobile_ijWith current time State-of-charge and current time decision behavior a that electric automobile is taken within the subsequent time time period_iRelevant.Therefore, root During going on a journey according to electric automobile, different travel behaviours (charging, not charging also does not travel and travel three kinds) are to electrokinetic cell soc The impact that change produces, the decision behavior that may take during user is gone on a journey is classified as follows.

(2) state transition probability.User's charging behavior is to meet demand, if charged terminate rear electrokinetic cell soc Value there is no change, then this charges just nonsensical.After a period of time it is therefore desirable to user charges, its soc value should There is certain change, in this patent, suppose that the excursion of its soc value is 0.2≤soc_k＜ 1, by filling soon, trickle charge power different, Can release its correspondingly fill soon, the distribution of trickle charge charging duration, derivation is as follows:

Charging duration scope soon:

$0.2\&le;\frac{p_k{t}_i}{q_k}<1$

$\frac{0.2q_k}{p_k}\&le;t_i<\frac{q_k}{p_k}$

Trickle charge charging duration scope:

$0.2\&le;\frac{p_m{t}_i}{q_k}<1$

$\frac{0.2q_k}{p_m}\&le;t_i<\frac{q_k}{p_m}$

In formula: p_k、p_mIt is respectively the fast, charge power of trickle charge, t_iFor charging duration, q_kBattery for kth electric automobile Capacity.It is assumed that the expectation of quick charge power is about 5 times of trickle charge, it is computed obtainingTherefore, basis in turn The current chargeable duration of user may determine that the charging modes that user may take, that is, when the chargeable duration of userWhen it is believed that private car takes fast charge mode, charging durationWhen, take is trickle charge mode.

In sum, according to taking different decision behaviors it can be deduced that the transition probability of corresponding decision, remember p_ijIt is from shape State s_iSwitch to state s_jState transition probability, its concrete representation is as follows.

$p_{ij}=\left\{\begin{array}{cc}\&integral;p(\frac{0.2q_k}{p_k}\&le;t_i<t_c)t\left(t_i\right){\mathrm{dt}}_i& a_i=1++\\ \&integral;p(t_c\&le;t_i<\frac{q_k}{p_m})t\left(t_i\right){\mathrm{dt}}_i& a_i=1+\\ 1& a_i=0\\ f\left(d_n\right)& a_i=-1\end{array}\right.$

In formula:Be respectively user take fill soon, the probability of trickle charge, t (t_i) it is single Secondary chargeable duration t_iProbability density function, f (d_n) for user's single operating range probability-distribution function.

(3) soc of subsequent time electric automobile.The soc of known current time and from current time start a period of time in The decision behavior taken, then draw the soc value of subsequent time, and calculating process is as follows.

Work as a_i=1+ or a_iDuring=1++, electric automobile is charged, thenWherein charge power p Value is:

$p=\left\{\begin{array}{cc}{p}_k& \frac{0.2q_k}{p_k}\&le;t_i<t_c\\ p_m& t_c\&le;t_i<\frac{q_k}{p_m}\end{array}\right.$

Work as a_iWhen=0, electric automobile is not charged also not travelling, then its state-of-charge does not change, i.e. soc_j= soc_i

Work as a_iWhen=- 1, electric automobile during traveling, then

${\mathrm{soc}}_j={\mathrm{soc}}_i-\frac{d_n{w}_{100}}{q_k}$

In formula: p_k、p_mIt is respectively the fast, charge power of trickle charge, d_nFor user's single operating range, w₁₀₀Every for electric automobile 100km power consumption, q_kBattery capacity for kth electric automobile.

To sum up can obtain the soc of subsequent time electric automobile:

${\mathrm{soc}}_j=\left\{\begin{array}{cc}{\mathrm{soc}}_i+\frac{p_k{t}_i}{q_k}& a_i=1++\\ {\mathrm{soc}}_i+\frac{p_m{t}_i}{q_k}& a_i=1+\\ {\mathrm{soc}}_i& a_i=0\\ {\mathrm{soc}}_i-\frac{d_n{w}_{100}}{q_k}& a_i=-1\end{array}\right.$

It is illustrated in figure 2 the present invention and is based on markovian load prediction flow chart, the present invention provides according to existing document Material and DOT issue American family trip survey data (national household travel survey, Nhts) give following condition:

(1) fill soon, the charge power of trickle charge, the battery capacity of electric automobile, the average overall travel speed of vehicle, every kilometer consumption Electricity；(2) electric automobile is gone on a journey in one day first initial time and initial state-of-charge soc value；(3) after traveling terminates every time When soc drops to certain threshold values soc, < when 0.2, electric automobile can select to be charged in charging station；(4) electric automobile list known to The probability density function of secondary distance travelled；(5) probability-distribution function of the single charge duration of electric automobile；(6) single stops The probability density function of persistent period；(7) one day trip finish time t of user_end.

The present invention describes the situation of change of electrokinetic cell soc during user's trip using Markov Chain, by electronic vapour Car electrokinetic cell soc, as state variable, goes on a journey according to automobile user during being accustomed to determining unique user trip in a day The decision behavior of day part, simulation produces user's trip sequence of a day, so that it is determined that all users charge period of a day, filling Electrically, adopt monte carlo method that the trip sequence of user is simulated based on conditions above, process is as follows:

(1) setting primary data information (pdi) and number realization；Wherein primary data information (pdi) includes:

Fill soon, the charge power of trickle charge, the battery capacity of electric automobile, the average overall travel speed of vehicle, every 100 kms consumption Electricity；

Initial time and initial SOC that electric automobile is gone on a journey in one day first；Travel every time and terminate rear battery lotus The threshold value of electricity condition；

The probability density function of user's single operating range；

The probability density function of the single charge duration of electric automobile；

The probability density function of single parking duration；

One day trip end time of user.

(2) battery charge according to primary data information (pdi) extract unique user one day travel time and trip moment first State value；

(3) extract single distance travelled, and calculate the battery lotus at the end of this section of stroke end time and this section of stroke Electricity condition value, the calculating of wherein battery charge state value is drawn using equation below:

${\mathrm{soc}}_j={\mathrm{soc}}_i-\frac{d_n{w}_{100}}{q_k}$

Wherein, d_nFor user's single operating range, w₁₀₀For the every 100 km power consumption of electric automobile, q_kFor the electronic vapour of kth The battery capacity of car, soc_i- for electric automobile any time battery charge state value, soc_j: for electric automobile in certain decision-making The battery charge state value of the subsequent time under behavior.

(4) judge whether the battery charge state value at the end of described stroke is less than the threshold value setting, if less than setting Threshold value, then execution step (5), otherwise, then execution step (7),

(5) randomly draw user's single charge duration, at the end of judging charging modes that user may take and charging Battery charge state value and end time；Record the battery charge shape at the end of this charging duration, charging modes, and charging State value and end time；

Wherein, judge that the charging modes that user may take are determined by formula below:

Charging duration scope is soon:

$\frac{0.2q_k}{p_k}\&le;t_i<\frac{q_k}{p_k}$

Trickle charge charging duration scope is:

$\frac{0.2q_k}{p_m}\&le;t_i<\frac{q_k}{p_m}$

It is assumed that p_k=5p_m, then

$\frac{q_k}{p_k}=\frac{0.2q_k}{p_m}=t_c$

It is a length of when user chargesThen take fast charge mode,

It is a length of when user chargesThen take trickle charge mode,

Wherein, p_k、p_mBe respectively fill soon, the charge power of trickle charge, t_iFor charging duration, q_kElectricity for kth electric automobile Tankage；

Battery charge state value at the end of charging draws according to formula below:

${\mathrm{soc}}_j=\left\{\begin{array}{cc}{\mathrm{soc}}_i+\frac{p_k{t}_i}{q_k}& a_i=1++\\ {\mathrm{soc}}_i+\frac{p_m{t}_i}{q_k}& a_i=1+\end{array}\right.$

Wherein, a_i=1++ represents and fills soon, a_i=1+ represents trickle charge, p_k、p_mBe respectively fill soon, the charge power of trickle charge, t_iFor Charging duration, q_kFor the battery capacity of kth electric automobile, soc_i- for electric automobile any time battery charge state value, soc_j: for the battery charge state value of subsequent time under certain decision behavior for the electric automobile.

(6) according to the battery charge state value at the end of this decision behavior and end time, judge that next step may be taken Decision behavior, if next decision behavior be stop, execution step (7)；If next decision behavior is to travel, execute Step (9)；

(7) randomly draw user's single parking duration, record battery charge state value and the knot at the end of this parking behavior The bundle time；

(8) described end time and the trip end time on the same day setting are compared, if the end time is more than or equal to setting The trip end time, then execution step (11), otherwise execution step (9)；

(9) extract single distance travelled, record the battery charge state value at the end of this traveling behavior and end time；

(10) above-mentioned end time and the trip end time on the same day setting are compared, if the end time is more than or equal to setting The trip end time, then execution step (11), otherwise execution step (4)；

(11) respectively add up fill soon, the charging load curve of trickle charge, obtain total bulk charging load；Concrete calculating process is such as Under:

A length of during charging soon:

$t_{i,1}=\frac{{\mathrm{soc}}_{i,1}-{\mathrm{soc}}_{i,0}}{p_k}{q}_k$

Trickle charge is a length of when charging:

$t_{i,2}=\frac{{\mathrm{soc}}_{i,1}-{\mathrm{soc}}_{i,0}}{p_m}{q}_k$

Then the charge power that fills soon of arbitrarily t is:

The trickle charge charge power of arbitrarily t is:

Wherein, t_i,1、t_i,2Represent respectively fill soon, trickle charge charging duration, soc_i,0Proceed by the lotus of charging for electric automobile Electricity condition, soc_i,1For the state-of-charge at the end of charging, t₀For starting to charge up the moment,

Charging electric vehicle power is added up, can predict fill soon, trickle charge load and total load, computing formula divides Not as follows:

$p_{t,1}=\underset{k=1}{\overset{n_1}{\&sigma;}}{p}_k$

$p_{t,2}=\underset{k=1}{\overset{n_2}{\&sigma;}}{p}_m$

$p_t=p_{t,1}+p_{t,2}=\underset{k=1}{\overset{n_1}{\&sigma;}}{p}_k+\underset{k=1}{\overset{n_2}{\&sigma;}}{p}_m$

Wherein, p_t,1、p_t,2Represent the filling soon of t, trickle charge load, p respectively_tBear for the total charging of t electric automobile Lotus, n1, n2 are natural number, represent the vehicle number filling soon with trickle charge respectively.

(12) whether the number of times judging at present simulation is more than the number realization setting in step (1)；If so, then enter step Suddenly (13)；If it is not, then return to step (2) continues executing with；

(13) judge whether restrain as total bulk charging load of analog result；If convergence, export analog result；If no Convergence, then return to step (1) resets number realization.

The present invention is based on Markov theory, by automobile user trip process electrokinetic cell under different decision behaviors The change procedure of soc is described by Markov Chain, obtains the situation of change of electrokinetic cell soc during this, and simulation is used The one day trip sequence in family, determines charging electric vehicle period, charging modes, calculates charging load by Monte Carlo simulation, should Method has vividly described the real time charging demand of user's trip process, is more accorded with based on the load prediction results that the method obtains Close practical situation.

The foregoing is only presently preferred embodiments of the present invention, not in order to limit the present invention, all spirit in the present invention and Within principle, any modification, equivalent substitution and improvement made etc., should be included within the scope of the present invention.

Claims (7)

1. one kind is based on markovian charging electric vehicle load forecasting method it is characterised in that including step:

(1) setting primary data information (pdi) and number realization；

(2) battery charge state in extract unique user one day from primary data information (pdi) travel time and trip moment first Value；

(3) extract single distance travelled, and calculate the battery charge shape at the end of this section of stroke end time and this section of stroke State value；

(4) whether the battery charge state value relatively at the end of described stroke is less than the threshold value setting, if less than the threshold setting Value, then execution step (5), otherwise, then execution step (7)；

(5) randomly draw user's single charge duration, the battery charge at the end of judging charging modes that user takes and charging State value and end time；Record the battery charge state value at the end of this charging duration, charging modes, and charging simultaneously And the end time；

(6) according to the battery charge state value at the end of this decision behavior and end time, judge the decision-making row that next step is taken For, if next decision behavior is to stop, execution step (7)；If next decision behavior is to travel, execution step (9)；

(7) randomly draw user's single parking duration, record battery charge state value and the time at the end of this parking behavior；

(8) relatively described parking behavior end time and the same day setting go on a journey the end time, if the end time is more than or equal to setting The fixed trip end time, then execution step (11), otherwise execution step (9)；

(9) extract single distance travelled, record the battery charge state value at the end of this traveling behavior and end time；

(10) relatively described traveling behavior end time and the same day setting go on a journey the end time, if the end time is more than or equal to setting The fixed trip end time, then execution step (11), otherwise execution step (4)；

(11) respectively add up fill soon, the charging load curve of trickle charge, obtain total bulk charging load；

(12) whether the number of times of relatively simulation at present is more than the number realization setting in step (1)；If so, then enter step (13)；If it is not, then return to step (2) continues executing with；

(13) judge whether restrain as total bulk charging load of analog result；If convergence, export analog result；If not receiving Hold back, then return to step (1) resets number realization.

2. according to claim 1 based on markovian charging electric vehicle load forecasting method it is characterised in that Primary data information (pdi) in described step (1) includes:

Fill soon, the charge power of trickle charge, the battery capacity of electric automobile, the average overall travel speed of vehicle, every 100 km power consumptions Amount；

Initial time and initial SOC that electric automobile is gone on a journey in one day first；Travel every time and terminate rear battery charge shape The threshold value of state；

The probability density function of user's single operating range；

The probability density function of the single charge duration of electric automobile；

The probability density function of single parking duration；

One day trip end time of user.

3. according to claim 1 based on markovian charging electric vehicle load forecasting method it is characterised in that Extraction single distance travelled in described step (3), and calculate the electricity at the end of this section of stroke end time and this section of stroke Pond SOC, the calculating of described battery charge state value is drawn using equation below:

${\mathrm{soc}}_j={\mathrm{soc}}_i-\frac{d_n{w}_{00}}{q_k}$

Wherein, d_nFor user's single operating range, w₁₀₀For the every 100 km power consumption of electric automobile, q_kFor kth electric automobile Battery capacity, soc_iFor the battery charge state value of any current time of electric automobile, soc_jFor electric automobile in certain decision-making row For under subsequent time battery charge state value.

4. according to claim 1 based on markovian charging electric vehicle load forecasting method it is characterised in that Judge that the charging modes detailed process that user takes is in described step (5): charging duration scope is soon:

$\frac{0.2q_k}{p_k}\&le;k_i<\frac{q_k}{p_k}$

Trickle charge charging duration scope is:

$\frac{0.2q_k}{p_m}\&le;t_i<\frac{q_k}{p_m}$

It is assumed that p_k=5p_m, then

$\frac{q_k}{p_k}=\frac{0.2q_k}{p_m}=t_c$

It is a length of when user chargesThen take fast charge mode,

It is a length of when user chargesThen take trickle charge mode,

Wherein, p_k、p_mBe respectively fill soon, the charge power of trickle charge, t_iFor charging duration, q_kBattery for kth electric automobile holds Amount.

5. according to claim 1 based on markovian charging electric vehicle load forecasting method it is characterised in that The battery charge state value at the end of charging in described step (5) draws according to formula below:

${\mathrm{soc}}_j=\left\{\begin{array}{cc}{\mathrm{soc}}_i+\frac{p_k{t}_i}{q_k}& a_i=1++\\ {\mathrm{soc}}_i+\frac{p_m{t}_i}{q_k}& a_i=1+\end{array}\right.$

Wherein, a_i=1++ represents and fills soon, a_i=1+ represents trickle charge, p_k、p_mBe respectively fill soon, the charge power of trickle charge, t_iFor charging Duration, q_kFor the battery capacity of kth electric automobile, d_nFor user's single operating range, w₁₀₀For the every 100 km consumptions of electric automobile Electricity, soc_iFor the battery charge state value of electric automobile any time, soc_jFor electric automobile under certain decision behavior under The battery charge state value in one moment.

6. according to claim 1 based on markovian charging electric vehicle load forecasting method it is characterised in that In described step (11) respectively add up fill soon, the charging load curve of trickle charge, obtain total bulk charging load, concrete calculating process is such as Under:

A length of during charging soon:

$t_{i,1}=\frac{{\mathrm{soc}}_{i,1}-{\mathrm{soc}}_{i,0}}{p_k}{q}_k$

Trickle charge is a length of when charging:

$t_{i,2}=\frac{{\mathrm{soc}}_{i,1}-{\mathrm{soc}}_{i,0}}{p_m}{q}_k$

Then the charge power that fills soon of arbitrarily t is:

The trickle charge charge power of arbitrarily t is:

Wherein, t_i,1、t_i,2Represent respectively fill soon, trickle charge charging duration, soc_i,0Proceed by the charged shape of charging for electric automobile State, soc_i,1For the state-of-charge at the end of charging, t₀For starting to charge up the moment,

Charging electric vehicle power is added up, predict fill soon, trickle charge load and total load, computing formula is as follows respectively:

$p_{t,1}=\underset{k=1}{\overset{n_1}{\&sigma;}}{p}_k$

$p_{t,2}=\underset{k=1}{\overset{n_2}{\&sigma;}}{p}_m$

$p_t=p_{t,1}+p_{t,2}=\underset{k=1}{\overset{n_1}{\&sigma;}}{p}_k+\underset{k=1}{\overset{n_2}{\&sigma;}}{p}_m$

Wherein, p_t,1、p_t,2Represent respectively fill soon, trickle charge load, p_tFor the total charging load of electric automobile, n1, n2 are natural number, Represent the vehicle number filling soon with trickle charge respectively.

7. one kind is based on markovian charging electric vehicle load prediction system it is characterised in that including information setting mould Block: for inputting primary data information (pdi), and set number realization；

First abstraction module: for extracting unique user one day travel time and the trip moment first from primary data information (pdi) Battery charge state value；

Second abstraction module: for extracting single distance travelled, and calculate this section of stroke end time and this section of stroke terminates When battery charge state value；

First comparison module: for the battery charge state value at the end of relatively described stroke whether less than the threshold value setting；

3rd abstraction module: be used for randomly drawing user's single charge duration, and judge charging modes that user may take and Battery charge state value at the end of charging and end time；Record this charging duration, charging modes simultaneously, and the knot that charges Battery charge state value during bundle and end time；

Decision-making module: for according to the battery charge state value at the end of this decision behavior and end time, judging that next step is adopted The decision behavior taking is to stop or travel；

4th abstraction module: be used for randomly drawing user's single parking duration, record the battery charge at the end of this parking behavior State value and time；

Second comparison module: for the trip end time on the same day of relatively described parking behavior end time and setting；

4th abstraction module: for extracting single distance travelled, record battery charge state value at the end of this traveling behavior and End time；

3rd comparison module: for the trip end time on the same day of relatively described traveling behavior end time and setting；

Accumulator module: for adding up respectively to fill soon, the charging load curve of trickle charge, obtain total bulk charging load；

4th comparison module: whether the number of times for relatively simulation at present is more than the number realization setting；

Judge module: for judging whether restrain as total bulk charging load of analog result.