CN104899667A - Electric car charging-discharging behavior prediction method - Google Patents

Abstract

The invention provides an electric car charging-discharging behavior prediction method. The method includes the steps of a, setting electric car charging-discharging basic constraint; b, setting a target function; c, calculating MSN influence omega 2; d, using omega 2 to correct cross inheritance particle swarm algorithm parameters; e, using the electric car charging-discharging basic constraint as premise, and using the corrected cross inheritance particle swarm algorithm so solve the target function so as to obtain an electric car charging-discharging plan and power distribution network load. The method has the advantages that the cross inheritance particle swarm algorithm is used to predict the charging-discharging plan of an electric car user, influence of a mobile social network on the charging-discharging plan is fully considered, prediction result accuracy is increased greatly, and reliable reference data can be provided to the power supply department during power grid load adjusting.

Description

A kind of Forecasting Methodology of electric automobile discharge and recharge behavior

Technical field

The present invention relates to a kind of Forecasting Methodology considering the electric automobile discharge and recharge behavior of mobile community network impact, belong to power transmission and distribution technical field.

Background technology

The extensive use of electric automobile defines huge charge requirement, also brings huge challenge to the planning of electrical network, operation, therefore predicts the discharge and recharge behavior of electric automobile and study just to seem particularly important.

Along with improving and the V2G (abbreviation of Vehicle-to-grid of tou power price strategy, it is described that such a system: when hybrid electric vehicle or pure electric vehicle be not when running, by the electro-motor being connected to electrical network, energy is defeated by electrical network, conversely, when the battery of electric motor car needs to be full of, electric current can extract and be given to battery from electrical network) development of technology, electric automobile while as the vehicles, progressively starts the role playing the part of network load attemperator.Meanwhile, mobile community network (Mobile Social Network, MSN) expanding day, becomes a part indispensable in people's life.The influence power of MSN can affect the discharge and recharge plan even changing user to a certain extent, and then optimizes " peak load shifting ", even for electric automobile user creates larger income.

The existing electric automobile discharge and recharge Multiobjective Optimal Operation scheme based on tou power price utilizes crisscross inheritance particle cluster algorithm to predict the discharge and recharge plan of electric automobile user, and idiographic flow as shown in Figure 1.The program more adequately can predict the discharge and recharge plan of electric automobile, and network load regulates relatively obvious, and user's cost is less.But this scheme does not consider that MSN is on the impact of user's discharge and recharge plan, makes precision of prediction be subject to certain impact, is therefore necessary to be improved.

Summary of the invention

The object of the invention is to the drawback for prior art, the Forecasting Methodology of a kind of Consideration comprehensive electric automobile discharge and recharge behavior is provided, for power supply department regulates network load to provide reliable reference data.

Problem of the present invention realizes with following technical proposals:

A Forecasting Methodology for electric automobile discharge and recharge behavior, said method comprising the steps of:

A. formulate electric automobile discharge and recharge substantially to retrain:

P _Lij＜P _ij＜P _Hij

-C/5＜I _ij＜C/3

SOC _ijmin＜SOC _ij＜SOC _ijmax，

Wherein, P _ijbe the charge-discharge electric power of i-th car at time period j, on the occasion of being electric discharge, negative value is charging; P _lijrepresenting electric automobile maximum charge power, is negative value; P _hijrepresent maximum discharge power, on the occasion of; I _ijfor electric automobile i is at the charging and discharging currents of moment j, C is that vehicle lithium battery 1h is full of required electric current; SOC _ijbe the quantity of electric charge of i-th electric automobile when time j, SOC _ijminand SOC _ijmaxrepresent lowest charge amount and maximum amount of charge respectively;

B. set objectives function:

1. network load mean square deviation computation model during structure electric automobile discharge and recharge:

${\mathrm{minp}}_1=\underset{j=1}{\overset{24}{\Σ}}{(p_{Tj}-\underset{i=1}{\overset{n}{\Σ}}{p}_{ij}-p_{ave})}^2$

$p_{ave}=\underset{j=1}{\overset{24}{\Σ}}\frac{p_{Tj}-\underset{i=1}{\overset{24}{\Σ}}{p}_{ij}}{24},$

In formula, P ₁for network load mean square deviation; p _tjduring for networking without electric automobile, network is at the load of time period j; P _avefor the average load after electric automobile networking; N represents the quantity of electric automobile;

2. electric automobile user income calculation model is built:

${\mathrm{minT}}_1=\underset{j=1}{\overset{24}{\Σ}}\underset{i=1}{\overset{n}{\Σ}}\left(\right|P_{\mathrm{ij}}\left|M_j\right),$

In formula, T ₁for the income of electric automobile, negative value representative profit, on the occasion of representative loss; M _jfor electricity price, on the occasion of representative charging electricity price, negative value representative is to electrical network feed electricity price;

3. convert above-mentioned two models to single model by weighting process, obtain final objective function:

$\min T=\mathrm{\α}\left(\frac{T_1}{T_{\max }}\right)+\mathrm{\β}\left(\frac{p_1}{p_{\max }}\right),$

α+β＝1，

In formula, T is T ₁and P ₁final goal value after weighting merges, α, β are weight coefficient; P _maxfor former network load; T _maxfor the cost needed from electricity minimum to electricity maximal value when electric automobile accepts adjustment;

C. MSN influence power ω is calculated ₂:

ω ₂＝p+q，

In formula, p is external action, and q is internal influence, with reference to the impact of different p and q values for network, and initialization p=0.005, q=0.7 respectively;

D. ω is utilized ₂crisscross inheritance particle cluster algorithm parameter is revised:

Basic particle group algorithm module in crisscross inheritance particle swarm algorithm model is as follows:

$v_{ld}^{k+1}={\mathrm{\ωv}}_{ld}^k+C_1{r}_{and1}^k(p_{ld}^k-x_{ld}^k)+C_2{r}_{and2}^k(g_{ld}^k-x_{ld}^k),$

$x_{ld}^{k+1}=x_{ld}^k+v_{ld}^k,$

In formula, represent the position of particle l d dimension in kth time iteration; for representing the speed of particle l d dimension in kth time iteration; for particle l optimal location in d dimension in k iteration; for all particles optimal location in d dimension in k iteration; ω is inertial factor; C ₁, C ₂for Studying factors;

Utilize ω ₂to particle cluster algorithm inertia weight and Studying factors correction:

ω＝aω ₁+b(1－ω ₂) _，

C ₁＝C _1s+sinω，

C ₂＝C _2s+cosω，

Wherein:

${\mathrm{\ω}}_1={\mathrm{\ω}}_{max}-({\mathrm{\ω}}_{max}-{\mathrm{\ω}}_{min})\sqrt{\frac{n}{N}},$

In formula, a, b are the weight coefficient of MSN influence power and inertial factor weighting, meet a+b=1, C _1sand C _2sbe respectively Studying factors C ₁and C ₂adjusted value, ω ₁for primary particle group algorithm dynamic change inertial factor, ω _maxfor ω ₁theoretical maximum 0.9, ω _minfor ω ₁theoretical minimum value 0.4, n be current particle iterations in particle cluster algorithm, N is particle cluster algorithm particle iterations altogether;

E. be substantially constrained to prerequisite with electric automobile discharge and recharge, utilize the crisscross inheritance particle cluster algorithm revised to solve objective function, obtain discharge and recharge plan and the distribution network load of electric automobile.

The present invention predicts the discharge and recharge plan of electric automobile user and has taken into full account the impact of mobile community network on discharge and recharge plan, substantially increases the accuracy predicted the outcome, and network load can be regulated to provide reliable reference data for power supply department.

Accompanying drawing explanation

Below in conjunction with accompanying drawing, the invention will be further described.

Fig. 1 is the transfer automobile discharge and recharge Multiobjective Optimal Operation protocol procedures figure based on tou power price;

Fig. 2 is the lower electric automobile discharge and recharge concept map of MSN impact;

Fig. 3 is the electric automobile discharge and recharge prediction process flow diagram based on MSN impact;

Fig. 4 is the network load under 3 kinds of scheduling strategies.

In figure neutralization literary composition, each symbol inventory is: P _ijbe the charge-discharge electric power of i-th car at time period j; P _lijrepresent electric automobile maximum charge power; P _hijrepresent maximum discharge power; I _ijfor electric automobile i is at the charging and discharging currents of moment j; C is that vehicle lithium battery 1h is full of required electric current; SOC _ijbe the quantity of electric charge of i-th electric automobile when time j, SOC _ijminand SOC _ijmaxrepresent lowest charge amount and maximum amount of charge respectively; P ₁for network load mean square deviation; p _tjduring for networking without electric automobile, network is at the load of time period j; P _avefor the average load after electric automobile networking; N represents the quantity of electric automobile; T ₁for the income of electric automobile; M _jfor electricity price; T is T ₁and P ₁final goal value after weighting merges, α, β are weight coefficient; P _maxfor former network load; T _maxfor the cost needed from electricity minimum to electricity maximal value when electric automobile accepts adjustment; ω ₂for MSN influence power; P is external action; Q is internal influence; represent the position of particle l d dimension in kth time iteration; for representing the speed of particle l d dimension in kth time iteration; for particle l optimal location in d dimension in k iteration; for all particles optimal location in d dimension in k iteration; W is inertial factor; C ₁, C ₂for Studying factors; A, b are weight coefficient; C _1sand C _2sbe respectively Studying factors C ₁and C ₂adjusted value; ω ₁for primary particle group algorithm dynamic change inertial factor, ω _maxfor ω ₁theoretical maximum 0.9, ω _minfor ω ₁theoretical minimum value 0.4, n be current particle iterations in particle cluster algorithm, N is particle cluster algorithm particle iterations altogether.

Embodiment

The present invention includes two parts content: MSN is to the influence power of electric automobile discharge and recharge behavior; The discharge and recharge plan of electric automobile under this influence power.Below in conjunction with Fig. 2, Fig. 3, the present invention is elaborated:

(1) Fig. 2 is the influence power concept map of MSN to electric automobile discharge and recharge behavior.As the constituent particle of in MSN, electric automobile is subject to the informational influence (being called external action p) that Information Communication node in network is propagated, and network internal is associated individuality (other electric automobiles) idea, behavioral implications (being called internal influence q).P and q constitutes the influence power of MSN to electric automobile discharge and recharge jointly, is formulated as:

ω ₂＝p+q (1)

In formula, ω ₂be the influence power of MSN to electric automobile discharge and recharge.

MSN influence power affects electric automobile discharge and recharge behavior, regulates the former network load of distribution region composition, reaches the object of " peak load shifting ".

(2) Fig. 3 is the electric automobile discharge and recharge prediction process flow diagram based on MSN impact, and concrete steps are described below:

Step S1: start.

Step S2: formulate electric automobile discharge and recharge and substantially retrain.Following constraint must be met: electric automobile charge-discharge electric power within the tolerance range of batteries of electric automobile, must protect battery to greatest extent during electric automobile discharge and recharge; Meanwhile, the quantity of electric charge of batteries of electric automobile also will meet user at any time and use needs, within quantity of electric charge bound scope.Restricted model is as follows:

P _Lij＜P _ij＜P _Hij(2)

－C/5＜I _ij＜C/3 (3)

SOC _ijmin＜SOC _ij＜SOC _ijmax(4)

P _ijbe the charge-discharge electric power of i-th car at time period j, on the occasion of being electric discharge, negative value is charging; P _lijrepresenting electric automobile maximum charge power, is negative value; P _hijrepresent maximum discharge power, on the occasion of; I _ijfor electric automobile i is at the charging and discharging currents of moment j, C is that vehicle lithium battery 1h is full of required electric current; SOC _ijbe the quantity of electric charge of i-th electric automobile when time j, SOC _ijminand SOC _ijmaxrepresent lowest charge amount and maximum amount of charge respectively.

Step S3: set objectives function.Model of the present invention is to regulate network load stable and ensure the electric automobile income objective function that has been target making, and objective function meets the constraint condition in step S2.

The stability of network load can represent by load mean square deviation, and mean square deviation less expression network load is more stable.Network load mean square deviation computation model during structure electric automobile discharge and recharge:

${\mathrm{minp}}_1=\underset{j=1}{\overset{24}{\Σ}}{(p_{Tj}-\underset{i=1}{\overset{n}{\Σ}}{p}_{ij}-p_{ave})}^2---\left(5\right)$

$p_{ave}=\underset{j=1}{\overset{24}{\Σ}}\frac{p_{Tj}-\underset{i=1}{\overset{24}{\Σ}}{p}_{ij}}{24}---\left(6\right)$

In formula, P ₁for network load mean square deviation; During for networking without electric automobile, network is at the load of time period j; P _ijbe the charge-discharge electric power of i-th car at time period j, on the occasion of being electric discharge, negative value is charging; P _avefor the average load after electric automobile networking; N represents the quantity of electric automobile.

By tou power price scheduling strategy and V2G technology, electric automobile user can obtain income by the Reasonable Regulation And Control batteries of electric automobile quantity of electric charge: during charging, this moment income is negative, and during electric discharge, this moment income is just.

Electric automobile user income calculation model:

${\mathrm{minT}}_1=\underset{j=1}{\overset{24}{\Σ}}\underset{i=1}{\overset{n}{\Σ}}\left(\right|P_{\mathrm{ij}}\left|M_j\right)---\left(7\right)$

In formula, T ₁for the income of electric automobile, for negative representative of consumer income is just, for positive representative of consumer is lost; P _ijbe the charge-discharge electric power of i-th car at time period j; M _jfor on the occasion of representing charging electric vehicle electricity price, for negative value represents electric automobile to electrical network feed electricity price.

In order to ensure that network load optimum ensures user's income simultaneously, above-mentioned two models convert single model to by weighting process:

$\min T=\mathrm{\α}\left(\frac{T_1}{T_{\max }}\right)+\mathrm{\β}\left(\frac{p_1}{p_{\max }}\right)---\left(8\right)$

Formula (8) meets constraint condition:

α+β＝1 (9)

In formula, T is the two weighting final goal value, and α, β are weight coefficient; P _maxfor former network load; T _maxfor the cost needed from electricity minimum to electricity maximal value when electric automobile accepts adjustment.

Formula (8) is final goal function, namely network load minimum while ensure electric automobile user income.Inequality (2), (3), (4) are the base region constraint of electric automobile discharge and recharge.

Step S4: calculate MSN influence power.MSN influence power computing method are formula (1), do not repeat at this.

Step S5: add MSN influence power, crisscross inheritance particle cluster algorithm parameter is revised.

Basic particle group algorithm module in crisscross inheritance particle swarm algorithm model is as follows:

$V_{ld}^{k+1}={\mathrm{\ωv}}_{ld}^k+C_1{r}_{and1}^k(p_{ld}^k-x_{ld}^k)+C_2{r}_{and2}^k(g_{ld}^k-x_{ld}^k)---\left(10\right)$

$x_{ld}^{k+1}=x_{ld}^k+v_{ld}^k---\left(11\right)$

In formula, represent the position of particle l d dimension in kth time iteration, i.e. the charge-discharge electric power in an electric automobile a certain moment; for representing the speed of particle l d dimension in kth time iteration; for particle l optimal location in d dimension in k iteration; for all particles optimal location in d dimension in k iteration; ω is inertial factor, represents the coefficient that electric automobile does not change discharge and recharge behavior; C ₁, C ₂for Studying factors, represent electric automobile carries out discharge and recharge Behavioral change coefficient according to own situation, extraneous circumstance respectively, all the other parameters are particle cluster algorithm basic parameter, and therefore not to repeat here.

MSN is on the impact of electric automobile discharge and recharge, and be embodied on particle cluster algorithm inertia weight and Studying factors correction, it is described below:

ω＝aω ₁+b(1－ω ₂) (12)

C ₁＝C _1s+sinω (13)

C ₂＝C _2s+cosω (14)

Wherein:

${\mathrm{\ω}}_1={\mathrm{\ω}}_{max}-({\mathrm{\ω}}_{max}-{\mathrm{\ω}}_{\min })\sqrt{\frac{n}{N}}---\left(15\right)$

In formula, a, b are the weight coefficient of MSN influence power and inertial factor weighting, meet a+b=1, C _1sand C _2sbe respectively Studying factors C ₁and C ₂adjusted value, ω ₁for primary particle group algorithm dynamic change inertial factor, ω _maxfor ω ₁theoretical maximum 0.9, ω _minfor ω ₁theoretical minimum value 0.4, n be current particle iterations in particle cluster algorithm, N is particle cluster algorithm particle iterations altogether;

Step S6: initialization master data.The basic parameter of model is set, prepares emulation.

Step S7: utilize crisscross inheritance PSO Algorithm, be wherein constrained to prerequisite with discharge and recharge, solve adaptive value by objective function.

Step S8: try to achieve optimum solution, is electric automobile discharge and recharge plan.

Step S9: terminate.

The present invention is closeness to life actual conditions more, can be the discharge and recharge behavior that power department predicts electric automobile more truly by MSN influence power and tou power price strategy, and then electric automobile discharge and recharge behavior is regulated and controled both controlling, thus reach the object regulating network load, be electric automobile user extra earning, excitation user participates in network load and regulates simultaneously.The plan of travel of indirect predictions electric automobile, for the prediction electric automobile trip of municipal traffic department provides technical support.

Table 1 is the simulation comparison result of the present invention and prior art, and compared with the existing technology, when adopting this method to predict, load mean square deviation reduces 1/4, and user's income increases substantially, and meets our simulated target.

Table 1 simulated grid load stability of the present invention (load mean square deviation) and user's income

Fig. 4 is the network load under 3 kinds of scheduling strategies, can be found by Fig. 4, same tou power price, and the network load under MSN impact (the inventive method) is more mild, " peak load shifting " better effects if; Meanwhile, the grid load curve without MSN impact is shifted 0.5h to the right relative to the curve under MSN impact, shows that the electric automobile discharge and recharge under MSN impact regulates more timely, efficient.

Claims (1)

1. a Forecasting Methodology for electric automobile discharge and recharge behavior, is characterized in that, said method comprising the steps of:

A. formulate electric automobile discharge and recharge substantially to retrain:

P _Lij＜P _ij＜P _Hij

-C/5＜I _ij＜C/3

SOC _ijmin＜SOC _ij＜SOC _ijmax；

Wherein, P _ijbe the charge-discharge electric power of i-th car at time period j, on the occasion of being electric discharge, negative value is charging; P _lijrepresenting electric automobile maximum charge power, is negative value; P _hijrepresent maximum discharge power, on the occasion of; I _ijfor electric automobile i is at the charging and discharging currents of moment j, C is that vehicle lithium battery 1h is full of required electric current; SOC _ijbe the quantity of electric charge of i-th electric automobile when time j, SOC _ijminand SOC _ijmaxrepresent lowest charge amount and maximum amount of charge respectively;

B. set objectives function:

1. network load mean square deviation computation model during structure electric automobile discharge and recharge:

${\mathrm{minp}}_1=\underset{j=1}{\overset{24}{\Σ}}{(p_{Tj}-\underset{i=1}{\overset{n}{\Σ}}{p}_{ij}-p_{ave})}^2$

$p_{ave}=\underset{j=1}{\overset{24}{\Σ}}\frac{p_{Tj}-\underset{i=1}{\overset{24}{\Σ}}{p}_{ij}}{24};$

In formula, P ₁for network load mean square deviation; p _tjduring for networking without electric automobile, network is at the load of time period j; P _avefor the average load after electric automobile networking; N represents the quantity of electric automobile;

2. electric automobile user income calculation model is built:

${\mathrm{minT}}_1=\underset{j=1}{\overset{24}{\Σ}}\underset{i=1}{\overset{n}{\Σ}}\left(\right|P_{ij}\left|M_j\right);$

In formula, T ₁for the income of electric automobile, negative value representative profit, on the occasion of representative loss; M _jfor electricity price, on the occasion of representative charging electricity price, negative value representative is to electrical network feed electricity price;

3. convert above-mentioned two models to single model by weighting process, obtain final objective function:

$\min T=\mathrm{\α}\left(\frac{T_1}{T_{\max }}\right)+\mathrm{\β}\left(\frac{p_1}{p_{\max }}\right)$

α+β＝1；

In formula, T is T ₁and P ₁final goal value after weighting merges, α, β are weight coefficient; P _maxfor former network load; T _maxfor the cost needed from electricity minimum to electricity maximal value when electric automobile accepts adjustment;

C. MSN influence power ω is calculated ₂:

ω ₂＝p+q；

In formula, p is external action, and q is internal influence, with reference to the impact of different p and q values for network, and initialization p=0.005, q=0.7 respectively;

D. ω is utilized ₂crisscross inheritance particle cluster algorithm parameter is revised:

Basic particle group algorithm module in crisscross inheritance particle swarm algorithm model is as follows:

$v_{ld}^{k+1}={\mathrm{\ωv}}_{ld}^k+C_1{r}_{and1}^k(p_{ld}^k-x_{ld}^k)+C_2{r}_{and2}^k(g_{ld}^k-x_{ld}^k);$

$x_{ld}^{k+1}=x_{ld}^k+v_{ld}^k;$

In formula, represent the position of particle l d dimension in kth time iteration; for representing the speed of particle l d dimension in kth time iteration; for particle l optimal location in d dimension in k iteration; for all particles optimal location in d dimension in k iteration; ω is inertial factor; C ₁, C ₂for Studying factors;

Utilize ω ₂to particle cluster algorithm inertia weight and Studying factors correction:

ω＝aω ₁+b(1-ω ₂)

C ₁＝C _1s+sin ω

C ₂＝C _2s+cos ω；

Wherein:

${\mathrm{\ω}}_1={\mathrm{\ω}}_{max}-({\mathrm{\ω}}_{max}-{\mathrm{\ω}}_{\min })\sqrt{\frac{n}{N}};$

In formula, a, b are the weight coefficient of MSN influence power and inertial factor weighting, meet a+b=1, C _1sand C _2sbe respectively Studying factors C ₁and C ₂adjusted value, ω ₁for primary particle group algorithm dynamic change inertial factor, ω _maxfor ω ₁theoretical maximum 0.9, ω _minfor ω ₁theoretical minimum value 0.4, n be current particle iterations in particle cluster algorithm, N is particle cluster algorithm particle iterations altogether;

E. be substantially constrained to prerequisite with electric automobile discharge and recharge, utilize the crisscross inheritance particle cluster algorithm revised to solve objective function, obtain discharge and recharge plan and the distribution network load of electric automobile.