CN203933038U - From the grid-connected mixing photovoltaic power generation control system of net - Google Patents

Abstract

The utility model relates to a kind of from the grid-connected mixing photovoltaic power generation control system of net.This control system is by photovoltaic array, and combining inverter, batteries, two-way inverter, load, wattmeter, public electric wire net, demand response control system and switches set form.Described photovoltaic array is by combining inverter incoming transport side, and batteries is by two-way inverter incoming transport side; Combining inverter is connected, is connected by switches set (S1-S4) between public electric wire net and load by, two-way inverter; Whole from netting grid-connected hybrid power system by demand response control system unified management control.By the switching combination of switches set, system is supported 8 kinds of different operational modes, to realize the optimization of economic benefit and the peak load shifting of electrical network, and extends its useful life by the restriction battery discharging degree of depth with the method that discharges and recharges power.

Description

From the grid-connected mixing photovoltaic power generation control system of net

Technical field

The utility model relates to solar energy generation of electricity by new energy and application, relates in particular to a kind of from the grid-connected mixing photovoltaic power generation control system of net.

Background technology

Along with the aggravation of energy crisis, in developing regenerative resource, how more reasonably to utilize the energy also to become gradually the problem of social concerns.The load increasing speed of electric energy is greater than the growth of electric weight in recent years, causes load rate of grid to decline, and peak-valley difference strengthens.For electricity consumption enterprise, this can make electric cost greatly improve, and be unfavorable for economic benefit, and for public electric wire net, this will affect the reliability and stability of operation of power networks again.Therefore, by photovoltaic generation and the comprehensive utilization of electrical network electric energy, and it is significant based on time-of-use tariffs, distributed power supply system to be optimized to scheduling.

And present most distributed power supply system, although can realize the grid-connected from network operation of system, and there is energy storage device to improve energy utilization rate, but still there is following deficiency:

1. existing system is not optimized the economical operation of system for time-of-use tariffs.Although peer machine can support time-of-use tariffs, but the scheduling scheme of these systems is comparatively simple, only carry out pattern switching according to several threshold values or condition, though can reduce to a certain extent like this operating cost, but be not optimized control from the angle of global optimum, therefore effect of optimization is limited.

2. existing system lacks the prediction to day part photovoltaic power generation quantity conventionally, though peer machine relates to energy output prediction, prediction algorithm is comparatively simple, and predicted value confidence level is lower.Lack the prediction of day part energy output, just cannot arrange this day scheduling scheme by energy output, be difficult to realize economic optimum.

3. existing system is not considered service lifetime of accumulator, in running, and the depth of discharge of uncontrollable storage battery, and deep discharge will reduce the useful life of storage battery greatly.

Utility model content

In order to solve the deficiency existing in prior art, it is a kind of according to tou power price that the utility model provides, the microgrid that photovoltaic power generation apparatus, batteries and public network are formed be optimized control from the grid-connected mixing photovoltaic power generation control system of net.

For achieving the above object, the utility model by the following technical solutions:

A kind of from the grid-connected mixing photovoltaic power generation control system of net, it is by photovoltaic array, combining inverter, batteries, two-way inverter, load, wattmeter, public electric wire net, demand response controller, supervisory computer and switches set form, and switches set comprises switch S 1-S4; Described photovoltaic array is by combining inverter incoming transport side, and batteries is by two-way inverter incoming transport side; Combining inverter is connected with electrical network by switch S 1, is connected with two-way inverter by switch S 3, is connected with load by switch S 3, S4; Two-way inverter is connected with load by switch S 4, is connected with electrical network by switch S 4, S2; Electrical network is connected with load by switch S 2; Whole from netting grid-connected hybrid power system by demand response controller and supervisory computer unified management control; Difference by switch S 1-S4 opens and closes combination, carry out system support shutdown mode, from net storage battery power supply pattern, from the switching of net photovoltaic-battery-operated pattern, the independent powering mode of electrical network, mains supply-charge mode, photovoltaic power supply-grid-connected-charge mode, mains supply-grid-connected-charge mode and mains supply-grid-connected-discharge mode.

Described supervisory computer is provided with historical information database, according to the historical data of photovoltaic power generation quantity and power load and local weather information, photovoltaic generation discharge curve and the power load curve on prediction same day, and will predict the outcome and be sent to demand response controller by Ethernet.

Described demand response controller is made up of microcontroller, ethernet interface and A.C. contactor control interface, receive the power prediction value from supervisory computer, be optimized on this basis scheduling, and final control switch group is by the switching of each switch, carry out the adjustment of mode of operation, realize scheduling scheme.

Described each operational mode:

1) shutdown mode, all switches are off-state;

2), from net storage battery power supply pattern, S1, S2, S3 disconnect, S4 closure, and load is powered separately by storage battery, and energy flows to load by storage battery;

3) from net photovoltaic-battery-operated pattern, this pattern is again because the charging and discharging state difference of storage battery is divided into two kinds of situations: when switch S 1, S2 disconnect, S3, S4 closure, when photovoltaic array underpower is when meeting loading demand, supplemented by storage battery, common is load supplying, and energy flows to load by photovoltaic array and storage battery;

Switch S 1, S2 disconnect, and S3, S4 closure, in the time that photovoltaic array power is enough to meet loading demand, charge to storage battery, and energy flows to storage battery and load by photovoltaic array;

4) the independent powering mode of electrical network, switch S 1, S3, S4 disconnect, S2 closure, public electric wire net is load supplying separately, energy flows to load by public electric wire net;

5) mains supply-charge mode, S1, S3 disconnect, S2, S4 closure, public electric wire net is to load supplying and to charge in batteries, and energy flows to storage battery and load by public electric wire net;

6) photovoltaic power supply-grid-connected-charge mode, switch S 2 disconnects, S1, S3, S4 closure, photovoltaic array power is enough to meet loading demand, and after charge in batteries demand, unnecessary power delivery is to public electric wire net, energy flows to load from photovoltaic array, storage battery and public electric wire net;

7) mains supply-grid-connected-charge mode, S3 disconnects, S1, S2, S4 closure, photovoltaic array power output is to public electric wire net on the one hand, and public electric wire net is to load supplying, to charge in batteries on the other hand;

8) mains supply-grid-connected-discharge mode, S4 disconnects, S1, S2, S3 closure, photovoltaic array and storage battery power output are to public electric wire net, and public electric wire net is to load supplying.

A kind of optimal control method from the grid-connected mixing photovoltaic power generation control system of net, according to the historical data of photovoltaic power generation quantity and power load and weather information, photovoltaic generation discharge curve and the power load curve of predicting the same day, according to photovoltaic power generation quantity and the power load of prediction, taking operating cost minimum as optimization aim, with the accumulator cell charging and discharging power in per hour, with the exchange power of electrical network and the operational mode of system be optimised variable, with electric energy balance condition, bound with electrical network exchange power, accumulator cell charging and discharging Power Limitation and state-of-charge are restricted to constraints, carry out scheduling decision by particle swarm optimization algorithm system running pattern is carried out to decision-making scheduling, thereby make system operation cost minimum, and extend its useful life by the restriction battery discharging degree of depth with the mode that discharges and recharges power.

Photovoltaic power output is predicted as: add up under various weather, in the research period, the photovoltaic power output of every day, is evenly divided into some intervals by power output from 0 to maximum, and the power in same interval is as a state; Be divided into multiple time periods by one day, at least 1 hour time period, in the statistical research period, the transfer number of each time period photovoltaic power output, obtains state-transition matrix corresponding to this time period; After system commencement of commercial operation, the information first providing according to RSMC, finds the statistics under corresponding weather, then by markovian method, daylong photovoltaic power output is predicted.

Electric loading is predicted as: add up in the research period, the power load of every day, is evenly divided into some intervals by power output from 0 to maximum, and the power in same interval is as a state; Be divided into multiple time periods by one day, at least 1 hour time period, in the statistical research period, the transfer number of each time period power load, obtains state-transition matrix corresponding to this time period; After system commencement of commercial operation, according to statistics, by markovian method, daylong power load is predicted.

While carrying out power prediction, first obtain the initial condition probability mass function of the first unit interval, i.e. initial distribution p ₁, making the state probability corresponding to measured power of initial time is 1, and all the other state probabilities are 0, then utilize state-transition matrix to calculate the distribution probability of a moment state, and formula is as follows:

p _m+1＝p _mP _m

Wherein p _mand p _m+1represent respectively the distribution probability row vector of m time period and m+1 time period, P _mbe m state-transition matrix corresponding to each time period, obtain the distribution probability p of m+1 time period _m+1after, the method for recycling mathematic expectaion obtains the predicted value in m+1 moment, and computing formula is as follows:

F _m+1＝p _m+1P _EXP

Wherein, P _eXPfor mathematic expectaion matrix, F _m+1be the power prediction value of m+1 time period, obtain F _m+1after, repeat said process, until obtain the power prediction value of all periods of whole day, then by Ethernet, power output is sent to demand response controller.

The handling process of decision making algorithm is:

(1) obtain the every operational factor of predefined system, cost parameters, electric price parameter and photovoltaic power and power load;

(2) obtain predefined particle cluster algorithm parameter, mainly comprise population scale, maximum iteration time, the study factor and inertia weight coefficient, and Offered target function and containing the confidence level of the constraints of stochastic variable;

Wherein target function is:

$\left\{\begin{array}{c}\mathrm{Min}\stackrel{\‾}{f}\\ s.t.\mathrm{Pr}\{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{C}_i\≤\stackrel{\‾}{f}\}\≥\mathrm{\β}\end{array}\right.$

it is target function

β is given confidence level

C _ithe operating cost of i period, wherein C _i=T[J _{buy, i}p _{buy, i}+ P _{pv, i}c _{pv_m}+ | P _{bt, i}| C _{bt_m}-J _{sel, i}p _{sel, i}]

T is the time interval of unit period

N is the period sum in dispatching cycle

P _{buy, i}it is the electrical power of buying from public electric wire net the i period

P _{sel, i}it is the electrical power that the i period exports public electric wire net to

P _{pv, i}it is the generated output of i period photovoltaic array

P _{bt, i}be the power that discharges and recharges of i period storage battery, discharge for just, be charged as negative

J _{buy, i}be the electricity price of i period from public electric wire net power purchase

J _{sel, i}be the i period to sell the electricity price to public electric wire net

C _{pv_m}for the unit operating cost of photovoltaic array

C _{bt_m}for the maintenance cost of storage battery;

(3) initialization population, generates discharging and recharging power and buying the power from public electric wire net of each scheduling slot storage battery at random, forms a particulate, and utilizes the feasibility of constraints inspection particulate, until all particulate initialization is complete; Meanwhile, generate at random the initial velocity of each particulate;

(4) calculate the fitness value of each particulate, and contrast fitness value and the individual extreme value of each particulate, if the former is more excellent, upgrade the individual extreme value of current particulate and personal best particle; Otherwise remain unchanged;

(5) contrast current whole individual extreme value and global extremum, get the superior and upgrade current global extremum and global optimum position thereof;

(6) upgrade speed and the position of each particulate, and check the feasibility of particulate by constraints, until all particulate is feasible;

(7) repeat (4)～(8), until meet end condition;

(8) output optimal solution, system is in the operational mode of each hour, accumulator cell charging and discharging power, and with the power of electrical network exchange.

In described step (3), wherein constraints comprises electrical power Constraints of Equilibrium, retrains with electrical network exchange power constraint and storage battery;

Electrical power Constraints of Equilibrium formula is:

$P_{\mathrm{buy},i}+P_{\mathrm{pv},i}{\mathrm{\η}}_{\mathrm{inv}}+\frac{P_{\mathrm{bt},i}}{{\mathrm{\η}}_{\mathrm{ch}}}-P_{\mathrm{sel},i}-P_{\mathrm{ld},i}=0$

P _buy,i+P _pv,iη _inv+P _bt,iη _dis-P _sel,i-P _ld,i＝0

In formula, P _{pv, i}be the photovoltaic generation power of i period, P _{buy, i}and P _{sel, i}be respectively the power that the i period buys in and sell from electrical network, P _{bt, i}be the power that discharges and recharges of i period storage battery, P _{ld, i}be the load of i period, η _invfor the efficiency of inverter, η _chfor the charge efficiency of storage battery, η _disfor the discharging efficiency of storage battery;

With the formula of electrical network exchange power be:

$0\≤P_{\mathrm{buy},i}\≤P_{\mathrm{buy}}^{\max }$

$0\≤P_{\mathrm{sel},i}\≤P_{\mathrm{sel}}^{\max }$

In formula, it is respectively the maximum power value of power purchase and sale of electricity;

Storage battery constraint formulations is:

$P_{\mathrm{cbt},i}^{\max }\≤P_{\mathrm{bt},i}\≤P_{\mathrm{abt},i}^{\max }$

SOC _min≤SOC _i≤SOC _max

$\underset{i=1}{\overset{n}{\text{\Σ}}}{P}_{\mathrm{bt},i}T=0$

Wherein, be respectively the maximum power of the charge and discharge of storage battery i period, SOC _ifor the storage battery state-of-charge of i period, SOC _min, SOC _maxbe respectively minimum and the peak of storage battery charge state, suppose that the capacity of storage battery is constant herein, represent that storage battery equates at the initial time of dispatching cycle and the energy storage capacity of the finish time.

The beneficial effects of the utility model

1. power prediction accurately: existing system lacks the prediction of photovoltaic generation power conventionally, although peer machine relates to power prediction, but algorithm is simpler, as by the same day weather with the weather comparison of certain day in history, result coupling, just using certain day day part energy output in history as the predicted value of day part energy output on the same day.And the utility model adopts markovian Forecasting Methodology, the method is the transition probability between initial probability and the each state in different conditions by things, the general morphologictrend of judgement state, to realize the prediction to to-be, therefore the confidence level predicting the outcome is higher.

2. economical operation optimization: the economical operation prioritization scheme of existing system is comparatively simple, generally only carry out pattern switching according to several threshold values or condition, and reality is ever-changing, in system, contain again this stochastic variable of photovoltaic power, therefore this method cannot realize global optimum, cannot make the operating cost of whole day minimum.And the utility model adopts particle swarm optimization algorithm, when optimization, take into full account the impact of time-of-use tariffs, by minimum whole day cost as target function, therefore its optimum results approaches optimization truly more.

3. electrical network " peak load shifting ": economical operation optimization of the present utility model, from electrical network power purchase or to charge in batteries by the paddy phase, the peak phase to electrical network sale of electricity or preferentially utilize storage battery electric energy realize, so just, realized " peak load shifting " of electrical network, contribute to electrical network steadily, efficiently operation.

4. prolonging service life of battery: the utility model is when by PSO Algorithm target function, can add the restrictive condition of accumulator cell charging and discharging power and storage battery charge state, therefore can limit the maximum depth of discharge of storage battery, thereby extend service lifetime of accumulator.

Brief description of the drawings

Fig. 1 is system architecture diagram of the present utility model;

Fig. 2 is shutdown mode schematic diagram;

Fig. 3 is from net storage battery power supply pattern diagram;

Fig. 4 is under net photovoltaic-battery-operated pattern, battery discharging situation schematic diagram;

Fig. 5 is under net photovoltaic-battery-operated pattern, charge in batteries situation schematic diagram;

Fig. 6 is the independent powering mode schematic diagram of electrical network;

Fig. 7 mains supply-charge mode schematic diagram;

Fig. 8 is photovoltaic power supply-grid-connected-charge mode schematic diagram;

Fig. 9 is mains supply-grid-connected-charge mode schematic diagram;

Figure 10 is mains supply-grid-connected-discharge mode schematic diagram;

Figure 11 is the process chart of decision making algorithm.

Embodiment

In order to make the purpose of this utility model, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the utility model is described in further detail.Specific embodiment described herein is only in order to explain the utility model, and is not used in restriction utility model.

Fig. 1 is system architecture diagram of the present utility model.The utility model has been constructed a kind of system that based on tou power price, microgrid is optimized control, it is by photovoltaic array, and combining inverter, batteries, two-way inverter, load, wattmeter, ac bus, public electric wire net, demand response controller, supervisory computer and switches set form.

Demand response controller can be controlled tetra-diverter switches of S1-S4, and each diverter switch has closed and disconnected two states, by changing the state of S1-S4, can make system works 8 kinds of different operational modes:

1) shutdown mode.All switches are off-state, and whole system is operated in shutdown mode, there is no energy flow, as shown in Figure 2.

2) from net storage battery power supply pattern.S1, S2, S3 disconnect, S4 closure, and load is powered separately by storage battery, and energy flows to load by storage battery, as shown in Figure 3.

3) from net photovoltaic-battery-operated pattern.This pattern is again because the charging and discharging state difference of storage battery is divided into two kinds of situations: when S1, S2 disconnect, S3, S4 closure, when photovoltaic array underpower is when meeting loading demand, supplemented by storage battery, common is load supplying, and energy flows to load by photovoltaic array and storage battery, as shown in Figure 4; S1, S2 disconnect, and S3, S4 closure, in the time that photovoltaic array power is enough to meet loading demand, charge to storage battery, and energy flows to storage battery and load by photovoltaic array, as shown in Figure 5.

4) the independent powering mode of electrical network.S1, S3, S4 disconnect, S2 closure, and public electric wire net is load supplying separately, energy flows to load by public electric wire net, as shown in Figure 6.

5) mains supply-charge mode.S1, S3 disconnect, S2, S4 closure, and public electric wire net is to load supplying and to charge in batteries, and energy flows to storage battery and load by public electric wire net, as shown in Figure 7.

6) photovoltaic power supply-grid-connected-charge mode.S2 disconnects, S1, S3, S4 closure, and photovoltaic array power is enough to meet loading demand, and after charge in batteries demand, unnecessary power delivery is to public electric wire net.Energy flows to load from photovoltaic array, storage battery and public electric wire net, as shown in Figure 8.

7) mains supply-grid-connected-charge mode.S3 disconnects, S1, S2, S4 closure, and photovoltaic array power output is to public electric wire net on the one hand, and public electric wire net is to load supplying, to charge in batteries, as shown in Figure 9 on the other hand.

8) mains supply-grid-connected-discharge mode.S4 disconnects, S1, S2, S3 closure, and photovoltaic array and storage battery power output are to public electric wire net, and public electric wire net is to load supplying, as shown in figure 10.

When system operation, Management Calculation chance is predicted photovoltaic power output and the power load on the same day by markovian method.Therefore before system commencement of commercial operation, need the historical data of statistics photovoltaic power output and power load, and be stored in historical information database.The concrete statistical method of photovoltaic power output is:

1) add up under various weather, in the research period, the photovoltaic power output of every day.

2) power output is evenly divided into some intervals from 0 to maximum, the power in same interval is as a state.

3) be divided into multiple time periods (at least 1 hour time period) by one day, taking 1 hour as minimum interval, in the statistical research period, the transfer number of each time period photovoltaic power output, obtains state-transition matrix corresponding to this time period.

The statistical method of the statistical method of power load and photovoltaic power output is basic identical, just in the time of statistics, does not need to consider weather conditions.

After system commencement of commercial operation, power prediction software on supervisory computer, first obtains the Weather information that RSMC provides, and finds the statistics under corresponding weather in database, then by markovian method, daylong photovoltaic power output is predicted.

While carrying out power prediction, first obtain the initial condition probability mass function of the first unit interval, i.e. initial distribution p ₁, making the state probability corresponding to measured power of initial time is 1, all the other state probabilities are 0.Then utilize state-transition matrix to calculate the distribution probability of each state of moment, formula is as follows:

p _m+1＝p _mP _m

Wherein p _mand p _m+1represent respectively the distribution probability row vector of m time period and m+1 time period, P _mbe m state-transition matrix corresponding to each time period.Obtain the distribution probability p of m+1 time period _m+1after, the method for recycling mathematic expectaion obtains the predicted value in m+1 moment, and computing formula is as follows:

F _m+1＝p _m+1P _EXP

Wherein P _eXPfor mathematic expectaion matrix, F _m+1it is the power prediction value of m+1 time period.

Obtain F _m+1after, repeat said process, until obtain the power prediction value of all periods of whole day.

Then by Ethernet, power output is sent to demand response controller.Demand response controller is according to photovoltaic power generation quantity and the power load of prediction, taking operating cost minimum as optimization aim, with the accumulator cell charging and discharging power in per hour, with the exchange power of electrical network and the operational mode of system be optimised variable, with electric energy balance condition, be restricted to constraints with bound, accumulator cell charging and discharging Power Limitation and the state-of-charge of electrical network exchange power, carry out scheduling decision by particle swarm optimization algorithm, as shown in figure 11, detailed process is the handling process of decision making algorithm:

(1) obtain the every operational factor of predefined system, cost parameters, electric price parameter and photovoltaic power and power load.

(2) obtain predefined particle cluster algorithm parameter, mainly comprise population scale, maximum iteration time, study factor c ₁, c ₂with inertia weight coefficient ω etc., and Offered target function and containing the confidence level of the constraints of stochastic variable.

Wherein target function is:

$\left\{\begin{array}{c}\mathrm{Min}\stackrel{\‾}{f}\\ s.t.\mathrm{Pr}\{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{C}_i\≤\stackrel{\‾}{f}\}\≥\mathrm{\β}\end{array}\right.$

In formula, be target function, β is given confidence level, C _ithe operating cost of i period.Wherein C _i=T[J _{buy, i}p _{buy, i}+ P _{pv, i}c _{pv_m}+ | P _{bt, i}| C _{bt_m}-J _{sel, i}p _{sel, i}], T is the time interval of unit period, n is the period sum in dispatching cycle, P _{buy, i}be the electrical power of buying from public electric wire net the i period, P _{sel, i}be the electrical power that the i period exports public electric wire net to, P _{pv, i}be the generated output of i period photovoltaic array, P _{bt, i}be the power that discharges and recharges of i period storage battery, discharge for just, be charged as negative, J _{buy, i}be the electricity price of i period from public electric wire net power purchase, J _{sel, i}be the i period to sell the electricity price to public electric wire net, C _{pv_m}for the unit operating cost of photovoltaic array, C _{bt_m}for the maintenance cost of storage battery.

(3) initialization population.Discharging and recharging power and buying the power from public electric wire net of the each scheduling slot storage battery of random generation, forms a particulate, and utilizes the feasibility of constraints inspection particulate, until all particulate initialization is complete.Meanwhile, generate at random the initial velocity of each particulate.

Wherein constraints comprises electrical power Constraints of Equilibrium, retrains with electrical network exchange power constraint and storage battery.

Electrical power Constraints of Equilibrium formula is:

$P_{\mathrm{buy},i}+P_{\mathrm{pv},i}{\mathrm{\η}}_{\mathrm{inv}}+\frac{P_{\mathrm{bt},i}}{{\mathrm{\η}}_{\mathrm{ch}}}-P_{\mathrm{sel},i}-P_{\mathrm{ld},i}=0$

P _buy,i+P _pv,iη _inv+P _bt,iη _dis-P _sel,i-P _ld,i＝0

In formula, P _{pv, i}be the photovoltaic generation power of i period, P _{buy, i}and P _{sel, i}be respectively the power that the i period buys in and sell from electrical network, P _{bt, i}be the power that discharges and recharges of i period storage battery, P _{ld, i}be the load of i period, η _invfor the efficiency of inverter, η _chfor the charge efficiency of storage battery, η _disfor the discharging efficiency of storage battery;

With the formula of electrical network exchange power be:

$0\≤P_{\mathrm{buy},i}\≤P_{\mathrm{buy}}^{\max }$

$0\≤P_{\mathrm{sel},i}\≤P_{\mathrm{sel}}^{\max }$

In formula, it is respectively the maximum power value of power purchase and sale of electricity.

Storage battery constraint formulations is:

$P_{\mathrm{cbt},i}^{\max }\≤P_{\mathrm{bt},i}\≤P_{\mathrm{abt},i}^{\max }$

SOC _min≤SOC _i≤SOC _max

$\underset{i=1}{\overset{n}{\text{\Σ}}}{P}_{\mathrm{bt},i}T=0$

Wherein, be respectively the maximum power discharging and recharging of storage battery i period, SOC _ifor the storage battery state-of-charge of i period, SOC _min, SOC _maxbe respectively minimum and the peak of storage battery charge state, suppose that the capacity of storage battery is constant herein, represent that storage battery equates at the initial time of dispatching cycle and the energy storage capacity of the finish time.

Because photovoltaic power output has randomness, the existence of this stochastic variable makes some constraints no longer have certainty.Therefore adopt Probability Forms to be described the inequality constraints that contains stochastic variable, it can be set up under certain confidence level, thereby realize the processing to these constraintss.

(4) calculate the fitness value of each particulate, and contrast fitness value and the individual extreme value of each particulate, if the former is more excellent, upgrade the individual extreme value of current particulate and personal best particle; Otherwise remain unchanged.

(5) contrast current whole individual extreme value and global extremum, get the superior and upgrade current global extremum and global optimum position thereof.

(6) upgrade speed and the position of each particulate, and check the feasibility of particulate by constraints, until all particulate is feasible.

(7) repeat (4)～(7), until meet end condition (reach maximum iteration time, preferably separate several times continuously unchanged or preferably separate with the difference of average adaptive value be less than a certain setting constant).

(8) output optimal solution, system is in the operational mode of each hour, accumulator cell charging and discharging power, and with the power of electrical network exchange.

After decision-making completes, demand response controller is according to the break-make of decision information control switch group, thereby makes system operate in the pattern of appointment, implemented scheduling scheme.

Claims (1)

1. One kind from net grid-connected mixing photovoltaic power generation control system, it is characterized in that, comprise photovoltaic array and batteries, described photovoltaic array is connected with switches set by combining inverter, described batteries is connected with switches set by two-way inverter, combining inverter is connected with public network by switch S 1, combining inverter is connected with two-way inverter by switch S 3, two-way inverter is connected with load by switch S 4, two-way inverter is successively by switch S 4, S2 is connected with public network, public network is connected with load by switch S 2, switches set is also connected with demand response controller and supervisory computer successively, switch S 1 and switch S 2 are connected with public network by ammeter respectively.