CN105119312A - A photovoltaic energy storage scheduling method based on mixed integer non-linear programming - Google Patents

Abstract

In the method, modeling is realized according to relations between photovoltaic power generation output power, time values, illumination intensity and temperature values, and predicted photovoltaic generation output power and load consumption power are obtained according to the relation between consumption power and time; by means of taking a present battery residual capacity into consideration, a charge and discharge decision sequence and an operation switching decision sequence during the occurrence of an optimal electricity-taking power from a power supply grid are obtained; and according to a calculating result, photovoltaic energy storage scheduling is carried out. According to the invention, the photovoltaic energy storage scheduling method is obtained through calculating of data; according to predicted specific parameters, charge and discharge and system switching controlling are carried out; avoidance of over-charge and over-discharge states of a charging battery module is avoided; the usage lift of the battery is prolonged; cases that photovoltaic power generation exceeds load electricity consumption and thus causing waste of photovoltaic power generation when the battery is fully charged are avoided so as to realize optimal energy utilization. Simultaneously, the safety of the system is raised. As a photovoltaic energy storage scheduling method based on the mixed integer non-linear programming, the invention can be widely applied to the field of photovoltaic power generating and energy storage.

Description

A kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming

Technical field

The present invention relates to photovoltaic generation energy storage field, especially a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming.

Background technology

Energy shortage problem is a 21 century very important problem, and one of best mode addressed this problem uses solar power generation.Solar energy has widely distributed and that near endless is endless feature, but photovoltaic resources also has intermittent and uncertain feature to use the problem of photovoltaic generating system of solar power generation to be, is difficult to load and provides a continual and steady supply of electric power.

At present, conventional method adopts energy-storage system store or discharge electric energy, thus reduce weather to the impact of photovoltaic generating system, is user's stable power-supplying, ensures reliability and the quality of power supply of power supply.In order to improve the energy management efficiency of photovoltaic generating system further, generally can arrange battery management system to manage it, but this mode just controls the charge status of battery module simply, along with the increase of photovoltaic generating system assembly, the expansion of function, this control method more and more shows the defects such as inefficiency, low-response, control accuracy difference, and charge and discharge control cannot be carried out according to the parameter of loading condition and battery module, energy utilization efficiency is low.

Summary of the invention

In order to solve the problems of the technologies described above, the object of the invention is: a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming is provided.

The technical solution adopted in the present invention is: a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming, includes following steps:

The neural network model of A, the neural network model setting up the photovoltaic generation prediction of output and load consumption prediction;

B, train according to the neural net of history data to the photovoltaic generation prediction of output of photovoltaic generating module, the history consumed power data according to AC load are trained the neural net that load consumption is predicted;

The service data of the photovoltaic generating module of C, collection scheduling initial time, and then the photovoltaic generation power output and the load consuming power that calculate the prediction of each time cycle in future time section according to the neural network model of the photovoltaic generation prediction of output and the neural network model of load consumption prediction;

The battery remaining power of D, acquisition current time charging cell module, according to photovoltaic generation power output and the load consuming power of above-mentioned prediction, utilize MIXED INTEGER nonlinear model to calculate to obtain from the discharge and recharge sequence of decisions of charging cell module during the power taking power optimized of power supply grid and run handover decisions sequence.

Further, also include step e: re-execute step B-D every a time cycle, and according to the discharge and recharge sequence of decisions of the charging cell module newly calculated and run handover decisions sequence and control.

Further, in described step D:

The battery remaining power of current time charging cell module is SOC (k), and its constraints is: SOC (min)≤SOC (k)≤SOC (max), SOC (min), SOC (max) represent minimum value and the maximum of battery remaining power SOC (k) respectively, and k is time series;

P _loadk () represents the load consuming power of prediction, P _loadmaxrepresent the load consuming power P of prediction _loadthe maximum of (k), P _pVk () represents the photovoltaic generation power of prediction;

In described mixed-integer nonlinear programming model, target function J is:

$J=\underset{k=1}{\overset{24}{\Σ}}\[P_{grid}\left(k\right)\×\mathrm{\Δ}t\]$

Wherein, P _gridk () represents the power taking power of system from power supply grid, Δ t represents the discharge and recharge time of charging cell module;

Described discharge and recharge decision variable is the charge-discharge electric power P of charging cell module _bat, its constraints is: P _batmin≤ P _bat≤ P _batmax, P _batmin, P _batmaxrepresent charge-discharge electric power P respectively _batthe minimum value of (k) and maximum;

Battery capacity change formula is:

$SOC(k+1)=SOC\left(k\right)+\frac{I\left(k\right)\mathrm{\Δ}t}{C};$

Described operation handover decisions variable is σ (k), two-valued variable σ (k)=0 when system takes alternating current directly to export; σ (k)=1 when system takes inverter to export;

Described power balance equation is:

$P_{PV}\left(k\right)+{\mathrm{\η}}_g{P}_{grid}\left(k\right)=P_{bat}\left(k\right)+(\frac{1}{{\mathrm{\η}}_i}-{\mathrm{\η}}_g)\mathrm{\σ}\left(k\right)P_{load}\left(k\right)+{\mathrm{\η}}_g{P}_{load}\left(k\right)+P_w\left(k\right)+P_{sys}$

Wherein, η _grepresent that the alternating current of power supply grid is converted to galvanic efficiency, η by rectifier _irepresent that direct current is converted to the efficiency of alternating current by inversion module, P _sysfor system is from loss, P _wk () is unnecessary solar energy;

Battery charging and discharging power equation is:

P _bat(k)＝V _bat(k)I(k)

V _bat(k)＝OCV(SOC)+RI(k)

Wherein, V _batk () is cell voltage, I (k) is battery charging and discharging electric current, P _batk () represents the charge-discharge electric power of charging cell module, OCV (SOC) represents the relation of battery open circuit voltage and battery remaining power SOC (k), and R is the internal resistance of cell;

When battery is filled to SOC=1, two-valued variable δ (k)=0; Otherwise δ (k)=1

Then 1-SOC (k)≤δ (k)≤1000 (1-SOC (k))

Unnecessary solar energy P _wk () meets following constraints:

0＜P _w(k)≤P _PVmax(1-δ(k))

P _PV(k)-P _w(k)≥0

Wherein, P _pVmaxfor maximum photovoltaic exports energy;

When system takes to exchange directly output, city's electric output power meets following output condition:

P _grid(k)-P _load(k)≥(P _gridmin-P _loadmax)σ(k)

Wherein, P _gridminfor obtaining least energy from electrical network;

In above-mentioned each formula, δ (k), σ (k) are two-valued variable.

Further, the future time section in described step B is following 24 hours, and each time cycle is 1 hour.

Further, described in described steps A, the history data of photovoltaic generating module includes time value, illumination intensity value and temperature value.

Further, in described steps A, the neural net of the photovoltaic generation prediction of output is trained for setting up photovoltaic generation power output and time value, relation between intensity of illumination and temperature value.

Further, in described steps A, the relation for setting up between load consuming power and time is trained to the neural net of load consumption prediction.

The invention has the beneficial effects as follows: the inventive method is according to photovoltaic generation power output and time value, relationship modeling between intensity of illumination and temperature value, and the relation between load consuming power and time, obtain photovoltaic generation power output and the load consuming power of prediction, and in conjunction with the battery remaining power of current time charging cell module, calculate the discharge and recharge sequence of decisions of charging cell module during the power taking power optimized obtained from power supply grid and run handover decisions sequence, and carrying out photovoltaic energy storage scheduling according to result of calculation.This method obtains photovoltaic energy storage dispatching method by mode data calculated, according to predicting that the design parameter of photovoltaic power generation quantity and load power consumption condition and the charging cell module obtained carries out discharge and recharge and systematic evaluation controls, ensure that charging cell module is not in super-charge super-discharge state, extend the life-span of charging cell module, when avoiding occurring that battery is full of, photovoltaic generation exceedes the situation appearance of load power consumption waste photovoltaic generation, thus reach optimum Energy harvesting, increase security of system simultaneously.

Accompanying drawing explanation

Fig. 1 is the main flow chart of steps of the inventive method;

Fig. 2 is the structure chart of photovoltaic energy storage dispatching patcher of the present invention;

Fig. 3 is neural metwork training block diagram of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described further:

With reference to Fig. 3, algorithm application of the present invention is in a kind of photovoltaic generation energy storage dispatching patcher, comprise photovoltaic generating module, photovoltaic controller, rectifier, inversion module, charging cell module, battery management system, diverter switch and the DC converter for the charging and discharging state that controls charging cell module, described photovoltaic generating module is connected with AC load by photovoltaic controller and inversion module successively, described photovoltaic controller is connected with charging cell module by DC converter with the link between inversion module, this link is also connected with power supply grid by rectifier, described inversion module is also directly connected with power supply grid, described battery management system is connected with charging cell module, the DC inverter that described inversion module is used for charging cell module to provide is output to AC load after alternating current, described diverter switch selects mains supply or inverter output power supply with deciding load.

With reference to Fig. 1, based on the system shown in Fig. 3, a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming of the present invention, includes following steps:

The neural network model of A, the neural network model setting up the photovoltaic generation prediction of output and load consumption prediction;

B, train according to the neural net of history data to the photovoltaic generation prediction of output of photovoltaic generating module, the history consumed power data according to AC load are trained the neural net that load consumption is predicted;

The service data of the photovoltaic generating module of C, collection scheduling initial time, and then the photovoltaic generation power output and the load consuming power that calculate the prediction of each time cycle in future time section according to the neural network model of the photovoltaic generation prediction of output and the neural network model of load consumption prediction;

The battery remaining power of D, acquisition current time charging cell module, according to photovoltaic generation power output and the load consuming power of above-mentioned prediction, utilize MIXED INTEGER nonlinear model to calculate to obtain from the discharge and recharge sequence of decisions of charging cell module during the power taking power optimized of power supply grid and run handover decisions sequence.

Be further used as preferred embodiment, also include step e: re-execute step B-D every a time cycle, and according to the discharge and recharge sequence of decisions of the charging cell module newly calculated and run handover decisions sequence and control; Concrete neural metwork training block diagram is with reference to Fig. 3.

Be further used as preferred embodiment, in described step D:

The battery remaining power of current time charging cell module is SOC (k), and its constraints is: SOC (min)≤SOC (k)≤SOC (max), SOC (min), SOC (max) represent minimum value and the maximum of battery remaining power SOC (k) respectively, and k is time series;

P _loadk () represents the load consuming power of prediction, P _loadmaxrepresent the load consuming power P of prediction _loadthe maximum of (k), P _pVk () represents the photovoltaic generation power of prediction;

In described mixed-integer nonlinear programming model, target function J is:

$J=\underset{k=1}{\overset{24}{\Σ}}\[P_{grid}\left(k\right)\×\mathrm{\Δ}t\]$

Wherein, P _gridk () represents the power taking power of system from power supply grid, Δ t represents the discharge and recharge time of charging cell module;

Described discharge and recharge decision variable is the charge-discharge electric power P of charging cell module _bat, its constraints is: P _batmin≤ P _bat≤ P _batmax, P _batmin, P _batmaxrepresent charge-discharge electric power P respectively _batthe minimum value of (k) and maximum;

Battery capacity change formula is:

$SOC(k+1)=SOC\left(k\right)+\frac{I\left(k\right)\mathrm{\Δ}t}{C};$

Described operation handover decisions variable is σ (k), two-valued variable σ (k)=0 when system takes alternating current directly to export; σ (k)=1 when system takes inverter to export;

Described power balance equation is:

$P_{PV}\left(k\right)+{\mathrm{\η}}_g{P}_{grid}\left(k\right)=P_{bat}\left(k\right)+(\frac{1}{{\mathrm{\η}}_i}-{\mathrm{\η}}_g)\mathrm{\σ}\left(k\right)P_{load}\left(k\right)+{\mathrm{\η}}_g{P}_{load}\left(k\right)+P_w\left(k\right)+P_{sys}$

Wherein, η _grepresent that the alternating current of power supply grid is converted to galvanic efficiency, η by rectifier _irepresent that direct current is converted to the efficiency of alternating current by inversion module, P _sysfor system is from loss, P _wk () is unnecessary solar energy;

Battery charging and discharging power equation is:

P _bat(k)＝V _bat(k)I(k)

V _bat(k)＝OCV(SOC)+RI(k)

Wherein, V _batk () is cell voltage, I (k) is battery charging and discharging electric current, P _batk () represents the charge-discharge electric power of charging cell module, OCV (SOC) represents the relation of battery open circuit voltage and battery remaining power SOC (k), and R is the internal resistance of cell;

When battery is filled to SOC=1, two-valued variable δ (k)=0; Otherwise δ (k)=1

Then 1-SOC (k)≤δ (k)≤1000 (1-SOC (k))

Unnecessary solar energy P _wk () meets following constraints:

0＜P _w(k)≤P _PVmax(1-δ(k))

P _PV(k)-P _w(k)≥0

Wherein, P _pVmaxfor maximum photovoltaic exports energy;

When system takes to exchange directly output, city's electric output power meets following output condition:

P _grid(k)-P _load(k)≥(P _gridmin-P _loadmax)σ(k)

Wherein, P _gridminfor obtaining least energy from electrical network;

In above-mentioned each formula, δ (k), σ (k) are two-valued variable.

According to above-mentioned constraints, by each time cycle, target function is calculated, obtain the discharge and recharge sequence of decisions of charging cell module during power taking power optimized (namely target function is minimum) from power supply grid and run handover decisions sequence.

Be further used as preferred embodiment, the future time section in described step B is following 24 hours, and each time cycle is 1 hour.

Be further used as preferred embodiment, described in described steps A, the history data of photovoltaic generating module includes time value, illumination intensity value and temperature value.

Being further used as preferred embodiment, in described steps A, the neural net of the photovoltaic generation prediction of output being trained for setting up photovoltaic generation power output and time value, relation between intensity of illumination and temperature value.

Being further used as preferred embodiment, in described steps A, the relation for setting up between load consuming power and time being trained to the neural net of load consumption prediction.

More than that better enforcement of the present invention is illustrated, but the invention is not limited to described embodiment, those of ordinary skill in the art can also make all equivalents or replacement under the prerequisite without prejudice to spirit of the present invention, and these equivalent distortion or replacement are all included in the application's claim limited range.

Claims (8)

1., based on a photovoltaic energy storage dispatching method for mixed integer nonlinear programming, it is characterized in that: include following steps:

The neural network model of A, the neural network model setting up the photovoltaic generation prediction of output and load consumption prediction;

B, train according to the neural net of history data to the photovoltaic generation prediction of output of photovoltaic generating module, the history consumed power data according to AC load are trained the neural net that load consumption is predicted;

The service data of the photovoltaic generating module of C, collection scheduling initial time, and then the photovoltaic generation power output and the load consuming power that calculate the prediction of each time cycle in future time section according to the neural network model of the photovoltaic generation prediction of output and the neural network model of load consumption prediction;

The battery remaining power of D, acquisition current time charging cell module, according to photovoltaic generation power output and the load consuming power of above-mentioned prediction, utilize MIXED INTEGER nonlinear model to calculate to obtain from the discharge and recharge sequence of decisions of charging cell module during the power taking power optimized of power supply grid and run handover decisions sequence.

2. a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming according to claim 1, it is characterized in that: also include step e: re-execute step B-D every a time cycle, and according to the discharge and recharge sequence of decisions of the charging cell module newly calculated and run handover decisions sequence and control.

3. a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming according to claim 1, it is characterized in that: described step D is specially: the battery remaining power obtaining current time charging cell module, according to photovoltaic generation power output and the load consuming power of above-mentioned prediction, set up mixed-integer nonlinear programming model and calculate the target function relevant to charging cell module, and then calculate the discharge and recharge sequence of decisions and operation handover decisions sequence that obtain charging cell module when making target function minimum.

4. a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming according to claim 3, is characterized in that: in described step D:

The battery remaining power of current time charging cell module is SOC (k), and its constraints is: SOC (min)≤SOC (k)≤SOC (max), SOC (min), SOC (max) represent minimum value and the maximum of battery remaining power SOC (k) respectively, and k is time series;

P _loadk () represents the load consuming power of prediction, P _loadmaxrepresent the load consuming power P of prediction _loadthe maximum of (k), P _pVk () represents the photovoltaic generation power of prediction;

In described mixed-integer nonlinear programming model, target function J is:

$J=\underset{k=1}{\overset{24}{\Σ}}\[P_{grid}\left(k\right)\×\mathrm{\Δ}t\]$

Wherein, P _gridk () represents the power taking power of system from power supply grid, Δ t represents the discharge and recharge time of charging cell module;

Described discharge and recharge decision variable is the charge-discharge electric power P of charging cell module _bat, its constraints is: P _batmin≤ P _bat≤ P _batmax, P _batmin, P _batmaxrepresent charge-discharge electric power P respectively _batthe minimum value of (k) and maximum;

Battery capacity change formula is:

$SOC(k+1)=SOC\left(k\right)+\frac{I\left(k\right)\mathrm{\Δ}t}{C};$

Described operation handover decisions variable is σ (k), two-valued variable σ (k)=0 when system takes alternating current directly to export; σ (k)=1 when system takes inverter to export;

Described power balance equation is:

$P_{PV}\left(k\right)+{\mathrm{\η}}_g{P}_{grid}\left(k\right)=P_{bat}\left(k\right)+(\frac{1}{{\mathrm{\η}}_i}-{\mathrm{\η}}_g)\mathrm{\σ}\left(k\right)P_{load}\left(k\right)+{\mathrm{\η}}_g{P}_{load}\left(k\right)+P_w\left(k\right)+P_{sys}$

Wherein, η _grepresent that the alternating current of power supply grid is converted to galvanic efficiency, η by rectifier _irepresent that direct current is converted to the efficiency of alternating current by inversion module, P _sysfor system is from loss, P _wk () is unnecessary solar energy;

Battery charging and discharging power equation is:

P _bat(k)＝V _bat(k)I(k)

V _bat(k)＝OCV(SOC)+RI(k)

Wherein, V _batk () is cell voltage, I (k) is battery charging and discharging electric current, P _batk () represents the charge-discharge electric power of charging cell module, OCV (SOC) represents the relation of battery open circuit voltage and battery remaining power SOC (k), and R is the internal resistance of cell;

When battery is filled to SOC=1, two-valued variable δ (k)=0; Otherwise δ (k)=1

Then 1-SOC (k)≤δ (k)≤1000 (1-SOC (k))

Unnecessary solar energy P _wk () meets following constraints:

0＜P _w(k)≤P _PVmax(1-δ(k))

P _PV(k)-P _w(k)≥0

Wherein, P _pVmaxfor maximum photovoltaic exports energy;

When system takes to exchange directly output, city's electric output power meets following output condition:

P _grid(k)-P _load(k)≥(P _gridmin-P _loadmax)σ(k)

Wherein, P _gridminfor obtaining least energy from electrical network;

In above-mentioned each formula, δ (k), σ (k) are two-valued variable.

5. a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming according to claim 1, is characterized in that: the future time section in described step C is following 24 hours, and each time cycle is 1 hour.

6. a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming according to claim 1, is characterized in that: described in described steps A, the history data of photovoltaic generating module includes time value, illumination intensity value and temperature value.

7. a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming according to claim 1, is characterized in that: train for setting up photovoltaic generation power output and time value, relation between intensity of illumination and temperature value the neural net of the photovoltaic generation prediction of output in described steps A.

8. a kind of photovoltaic energy storage dispatching method based on mixed integer nonlinear programming according to claim 1, is characterized in that: train the relation for setting up between load consuming power and time to the neural net of load consumption prediction in described steps A.