CN114336597A - Light storage inverter energy scheduling based on self-adaptive penalty function particle swarm algorithm - Google Patents

Abstract

The invention provides a light storage inverter energy scheduling method based on a self-adaptive penalty function particle swarm algorithm. In order to reduce daily power expenditure of users, the invention researches the energy scheduling problem of the household photovoltaic energy storage inverter system. The method combining the penalty function and the optimization algorithm is a common means for solving related problems, but the traditional penalty function has the defect that the penalty factor is difficult to determine, so the invention provides a self-adaptive penalty function method. The method can automatically adjust the penalty factor according to the size of the searched scheme exceeding the constraint range in each iteration process of the optimization algorithm, and finally search out the output data of each part of the system when the user power expenditure is small. Through comparative analysis, the algorithm provided by the invention effectively reduces the electric power expenditure of users.

Description

Light storage inverter energy scheduling based on self-adaptive penalty function particle swarm algorithm

Technical Field

The invention belongs to the technical field of power electronics, and particularly relates to light storage inverter energy scheduling based on a self-adaptive penalty function particle swarm algorithm.

Background

In recent years, the social industrialization is rapidly developed, and the demand and consumption of human energy are greatly increased. The large-scale utilization of traditional energy sources, while promoting socioeconomic and civilized progress, also puts irreversible stress on the ecological environment. In contrast, the environmental friendliness of renewable energy sources has made its development enthusiasm in modern society escalating. Among them, solar energy is the most abundant energy available to human beings, it is clean and harmless, does not need to transport, and has the universality. Therefore, the utilization of solar energy for power generation is one of the most effective means for solving the energy problem in the modern society.

As an important exploration of photovoltaic practicality, household photovoltaic energy storage inverters have an increasingly huge market. However, to realize stable and efficient operation of the photovoltaic-storage inverter system, energy management must be performed on the photovoltaic-storage inverter system, that is, on the basis of predicting photovoltaic output and power load requirements, energy distribution among units is coordinated by reasonably arranging charging and discharging strategies of an energy storage device in the system and output of a power grid, and finally, on the premise of meeting self supply and demand balance and the like, energy of each part of the system is fully utilized.

Energy scheduling of the optical storage inverter system is essentially a multi-constraint optimization problem, and generally, the multi-constraint problem is converted into an unconstrained problem through a traditional penalty function, and then a search is performed by adopting an optimization algorithm.

The key of the conventional penalty function solution problem is the selection of a penalty factor, however, a fixed method for determining the size of the penalty factor is not available at present. In practical engineering application, a constant is generally selected as a penalty factor according to experience, and since appropriate penalty strength cannot be determined, a better solution is difficult to obtain finally. At present, related researches have provided some improved methods, such as a static penalty function method, a dynamic penalty function method, an annealing penalty function method and the like, but the methods all have the problems of too many parameters, too high calculation intensity and the like.

Disclosure of Invention

Technical problem to be solved

Based on the defects mentioned in the background technology, the invention designs a self-adaptive penalty function method, and then takes a particle swarm algorithm as a search engine to carry out energy scheduling on the household light storage inverter system, so that the energy flow of each part in the system is reasonably controlled, and the daily power expenditure of the user is reduced.

(II) technical scheme

The invention discloses a light storage inverter energy scheduling method based on a self-adaptive penalty function particle swarm algorithm, which comprises the following steps of:

step 1: building a topological structure of the household photovoltaic energy storage inverter;

step 2: and (4) establishing a mathematical model for the energy scheduling problem of the light storage inverter system. The mathematical model of the optimization problem comprises two parts, an objective function and constraint conditions. The objective function represents the daily power expenditure of a user, the constraint condition represents the limiting condition which must be met by the system, and the optimization objective is to solve the minimum value of the mathematical model, namely the minimum value of the daily power expenditure of the user;

and step 3: constructing an expression, namely an adaptive penalty function expression, of which the penalty factor is automatically updated in each iteration of the optimization algorithm;

and 4, step 4: and (4) adjusting a new target function expression in each iteration of the optimization algorithm according to the self-adaptive penalty function expression obtained in the step (3), and optimizing by utilizing a particle swarm optimization algorithm.

Further, the topology structure of the household photovoltaic energy storage inverter in step 1 is as follows:

the photovoltaic power generation unit and the storage battery energy storage unit are connected to the input of the inverter through the preceding stage boosting direct current converter module and the resonance converter module respectively, and the storage battery can receive energy from other parts to be charged and can also discharge and supply energy to a load after being inverted. The output of the inverter is connected to the grid and the household load side. The inverter system has three load power supply modes, which are respectively: the power supply system comprises an inverter individual power supply mode, a power grid individual power supply mode and a power grid common power supply mode.

Further, the objective function obtained in step 2 is calculated as follows:

in 24 hours a day, a user can purchase electricity from the power grid or sell electricity to the power grid, but the prices of electricity purchase and electricity sale are different at different times. Due to planning, installation and maintenance costs, such as purchasing photovoltaic panels, storage batteries and the like, are sinking costs, and are not considered here. The final objective function, i.e. the daily power expenditure of the user, is thus expressed as:

wherein, J_cFor the total cost in period T, where T is 24 hours; e_buyThe amount of power purchased from the grid for the tth hour; f. of_buyA price to purchase electricity for the tth hour; e_sellThe amount of electricity sold to the grid for the tth hour; f. of_sellThe price of electricity sold at the t hour.

Further, the step 2 further includes a constraint condition of the light storage inverter system: system power balance constraints, grid output power constraints and storage battery related constraints.

The expression of the system power balance constraint is:

P_g+P_b+P_pv＝P_load

the expression of the power grid output power constraint is as follows:

-P_{g_max}≤P_g(t)≤P_{g_max}

the expression for the battery-related constraint is:

P_{bc_max}≤P_b(t)≤P_{bd_max}

DoD_min≤DoD(t)≤DoD_max

SoC_min≤SoC(t)≤SoC_max

wherein, P_pvIs the power of the photovoltaic power generation unit, P_bIf the term is positive value, the term indicates that the current state of the storage battery is discharging, and if the term is negative value, the term indicates that the current state of the storage battery is charging, P_gIf the item is a positive value, the current state indicates that the user purchases electricity from the power grid, and if the item is a negative value, the user sells electricity from the power grid. P_loadRepresenting the home load power. P_{g_max}For maximum power that the grid can deliver, P_{bc_max}And P_{dc_max}The maximum charging power and the maximum discharging power of the battery are shown, the DoD represents the discharging depth of the battery, and the SoC represents the state of charge of the battery; σ represents the self-discharge rate of the battery; p_BThe rated power of the battery.

Further, the adaptive penalty function expression in step 3 is:

p(δ)＝e^1000|δ|-1

wherein, δ is the range of the result obtained after each iteration of the optimization algorithm exceeding the constraint condition, and p (δ) represents the penalty factor value when the error is δ.

Further, the new objective function expression in step 4 is:

further, the step 4 further includes a particle swarm optimization algorithm process:

a: initializing a particle swarm including speed and position information of each particle;

b: calculating the fitness of each particle, namely an objective function result;

c: for each particle, its fitness value is compared with the individual optimal position p it has passed through_bestIn comparison, if more optimal, it is taken as the current individual optimal position p_best；

d: for each granuleSub-system for comparing its fitness value with the global optimum g over which it has passed_bestComparing, if more optimal, taking it as the current global optimal position g_best；

e: the velocity and position of the particles are adjusted according to the following two equations:

V_i＝w×V_i+c₁×rand()×(pbest_i-x_i)+c₂×rand()×(gbest_i-x_i)

x_i＝x_i+V_i

wherein w is an inertial weight factor; c. C₁And c₂The maximum step length of the flight of the particles to the individual optimum position and the global optimum position can be respectively adjusted to determine the influence of the individual experience and the group experience on the motion condition of the particles as a learning factor; rand () is a random number between 0 and 1, p_bestiRepresenting the individual optimum position of the current particle, g_bestiRepresenting the current global optimum position.

f: and c, ending the flow when the iteration times or the precision condition is reached, and otherwise, returning to the step b.

(III) advantageous effects

The invention provides an energy scheduling method of a light storage inverter based on a self-adaptive penalty function particle swarm algorithm. Compared with the prior art, the method constructs the self-adaptive penalty function according to the range of the result obtained in each iteration of the optimization algorithm which violates the constraint condition, so that the penalty factor can be completely updated in each iteration of the optimization process. On the one hand, the method avoids the defect that the penalty factor is difficult to determine, and the calculation amount and the required parameters are not complex. On the other hand, the method also improves the global optimizing capability and effectively reduces the daily power expenditure of the user. Therefore, the method provided by the invention is suitable for practical application.

Drawings

To more clearly illustrate the present invention or the technical solutions in the prior art, the following briefly describes the drawings used in the embodiments:

FIG. 1 is a schematic flow chart illustrating steps of energy scheduling of an optical storage inverter based on an adaptive penalty function particle swarm algorithm according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a light-storing inverter system to which the method of the present invention is applied;

FIG. 3 is a chart of prices of electricity sold and purchased at different times during a day according to the present invention;

FIG. 4 is a flow chart of a particle swarm algorithm based on an adaptive penalty function in the present invention;

FIG. 5 is a graph of photovoltaic power generation used in the present invention;

FIG. 6 is a graph of electricity usage for load used in the present invention;

FIG. 7 is a diagram of the simulated scheduling results obtained by the present invention.

FIG. 8 is a waveform diagram of the scheduling result obtained by the present invention.

Fig. 9 is a diagram of the scheduling result obtained when the penalty factor is 50.

Fig. 10 is a diagram of the scheduling result obtained when the penalty factor is 500.

Fig. 11 is a diagram of the scheduling result obtained when the penalty factor is 5000.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.

As one embodiment of the embodiments of the present invention, in this embodiment, an optical storage inverter energy scheduling based on an adaptive penalty function particle swarm algorithm is provided:

as shown in fig. 1, a schematic diagram of a light storage inverter energy scheduling method based on an adaptive penalty function particle swarm algorithm includes the following steps:

step 1, building a topological structure of a household photovoltaic energy storage inverter;

and 2, establishing a mathematical model for the energy scheduling problem of the light storage inverter system.

Step 3, constructing an expression, namely a self-adaptive penalty function expression, in which the penalty factors are automatically updated in each iteration of the optimization algorithm;

and 4, adjusting a new target function expression in each iteration of the optimization algorithm according to the adaptive penalty function expression obtained in the step 3, and optimizing by using a particle swarm optimization algorithm.

The steps of the invention are embodied as follows:

step 1: and constructing a topological structure of the household photovoltaic energy storage inverter.

System topology referring to fig. 2:

in the figure, a photovoltaic Part (PV) directly converts solar energy into electric energy by utilizing a photovoltaic effect, and a preceding stage boost direct current converter module (BDC) realizes Maximum Power Point Tracking (MPPT) on a photovoltaic panel so as to maximize photovoltaic power generation income. The storage battery is connected to the resonant converter module (LLC) for energy management control, and can receive energy from other parts for charging or discharge and supply energy to a load through the inverter module (SPTI). The local load in the figure represents the home subscriber side, connected to the output of the light storage inverter. The inverter system has three load power supply modes, which are respectively as follows: the power supply system comprises an inverter individual power supply mode, a power grid individual power supply mode and a power grid common power supply mode. The grid is connected to the output of the light storage inverter and can either provide power to the load user side or purchase power from the inverter system.

Due to external factors (such as illumination intensity) and randomness of load electricity consumption in the light storage inverter system, intelligent algorithms such as a neural network are generally adopted to predict the power of photovoltaic power generation and load electricity consumption before energy optimization, but the prediction part is not researched here.

Step 2: and (4) establishing a mathematical model for the energy scheduling problem of the light storage inverter system. The mathematical model of the optimization problem comprises two parts, an objective function and constraint conditions. The objective function represents the daily power expenditure of the user, the constraint condition represents the limiting condition which must be met by the system, and the optimization objective is to solve the minimum value of the mathematical model, namely the minimum value of the daily power expenditure of the user.

In 24 hours a day, a user can purchase electricity from the power grid or sell electricity to the power grid, but the prices of electricity purchase and electricity sale are different at different times. Due to planning, installation and maintenance costs, such as purchasing photovoltaic panels, storage batteries and the like, are sinking costs, and are not considered here. The final objective function, i.e. the daily power expenditure of the user, is thus expressed as:

wherein, J_cFor the total cost in period T, where T is 24 hours; e_buyThe amount of power purchased from the grid for the tth hour; f. of_buyA price to purchase electricity for the tth hour; e_sellThe amount of electricity sold to the grid for the tth hour; f. of_sellThe price of electricity sold at the t hour. The time-sharing electricity selling price and the electricity purchasing price refer to the figure 3.

In the process of scheduling system energy, the constraint conditions of the light storage inverter system must be met at any time: system power balance constraints, grid output power constraints and storage battery related constraints.

The expression of the system power balance constraint is:

P_g+P_b+P_pv＝P_load

the expression of the power grid output power constraint is as follows:

-P_{g_max}≤P_g(t)≤P_{g_max}

the expression for the battery-related constraint is:

P_{bc_max}≤P_b(t)≤P_{bd_max}

DoD_min≤DoD(t)≤DoD_max

SoC_min≤SoC(t)≤SoC_max

wherein, P_pvIs the power of the photovoltaic power generation unit, P_bIf the term is positive value, the term indicates that the current state of the storage battery is discharging, and if the term is negative value, the term indicates that the current state of the storage battery is charging, P_gIf the item is a positive value, the current state indicates that the user purchases electricity from the power grid, and if the item is a negative value, the user sells electricity from the power grid. P_loadRepresenting the home load power. P_{g_max}For maximum power that the grid can deliver, P_{bc_max}And P_{dc_max}The maximum charging power and the maximum discharging power of the battery are shown, the DoD represents the discharging depth of the battery, and the SoC represents the state of charge of the battery; σ represents the self-discharge rate of the battery; p_BThe rated power of the battery.

The constraint specific values are shown in table 1 below:

TABLE 1 system constraint value Table

Parameter(s)	Numerical value
		Pg_max/kW	5
Pbc_max/kW	4.5
		Pbd_max/kW	5
SoCmin/％	10
		SoCmax/％	90
DoDmin/％	20
		DoDmax/％	90
σ/％	7

And step 3: and constructing an adaptive penalty function expression.

The conventional penalty function method is explained below.

Assuming that the minimum value of f (x) is solved under the condition that g (x) is less than zero, the objective function can be expressed as

w＝f(x)+p max(0,g(x))

Where p is a penalty factor and is a constant. By analyzing the formula, it can be known that, if the minimum value of f (x) is desired to be obtained, in the searching process, for the scheme which does not satisfy the constraint condition, the competitive advantage with the scheme which satisfies the constraint condition is lost due to the increase of the objective function result. This penalty function method is therefore somewhat amenable to elimination of non-viable solutions. However, in practical application, there is no fixed method for selecting the penalty factor, so it is very difficult to select the proper penalty factor for different optimization problems, if the p value is too small, the proportion of the penalty term is very small, the penalty effect may not be achieved; conversely, if p is too large, it may prematurely converge to a locally optimal solution.

In the searching process, when the searched scheme does not meet the constraint condition, the penalty factor is correspondingly modified. If the range exceeding the constraint condition is larger, the corresponding penalty factor is also particularly large so as to force the search to rapidly enter the vicinity of the feasible region or the feasible region from the infeasible region; when the searched scheme exceeds the constraint range and becomes smaller, the penalty factor is correspondingly reduced, so that the search is slowly approached to a feasible solution from an area with smaller violation of the constraint condition; when the scheme meets the constraint condition, namely the out-of-constraint range is 0, no penalty item is needed, and the penalty factor is modified to be 0. It can be seen that the adjustment of the penalty factor in the optimization process depends on the size of the solution beyond the constraint range at this time, so that the adaptive penalty function expression can be easily constructed:

p(δ)＝e^1000|δ|-1

wherein, δ is the range of the result obtained after each iteration of the optimization algorithm exceeding the constraint condition, and p (δ) represents the penalty factor value when the error is δ.

And 4, step 4: and (4) adjusting a new target function expression in each iteration of the optimization algorithm according to the self-adaptive penalty function expression obtained in the step (3), and optimizing by utilizing a particle swarm optimization algorithm.

Adding the self-adaptive penalty function expression to the original target function expression to obtain:

after a new objective function expression is obtained, explaining a particle swarm optimization algorithm flow:

a: initializing a particle swarm including speed and position information of each particle;

b: calculating the fitness of each particle, namely an objective function result;

c: for each particle, its fitness value is compared with the individual optimal position p it has passed through_bestIn comparison, if more optimal, it is taken as the current individual optimal position p_best；

d: for each particle, its fitness value is compared with the global optimum g it has passed through_bestComparing, if more optimal, taking it as the current global optimal position g_best；

e: the velocity and position of the particles are adjusted according to the following two equations:

V_i＝w×V_i+c₁×rand()×(pbest_i-x_i)+c₂×rand()×(gbest_i-x_i)

x_i＝x_i+V_i

wherein w is an inertial weight factor; c. C₁And c₂The maximum step length of the flight of the particles to the individual optimum position and the global optimum position can be respectively adjusted to determine the influence of the individual experience and the group experience on the motion condition of the particles as a learning factor; rand () is a random number between 0 and 1, p_bestiRepresenting the individual optimum position of the current particle, g_bestiRepresenting the current global optimum position.

f: and c, ending the flow when the iteration times or the precision condition is reached, and otherwise, returning to the step b.

In the method, the optimization algorithm selects the power grid output power as a search quantity, and each particle comprises 24 (one day) power grid output operation data. The algorithm dimension is 24. After initialization, calculating a penalty factor of the particles, adding the obtained penalty item to the initial objective function to obtain the latest fitness, and then updating the global optimal value and the local optimal value according to the steps of the particle swarm algorithm until the cycle reaches the set maximum iteration times. In summary, a specific flow of the particle swarm algorithm based on the adaptive penalty function is shown in fig. 4. See Table 2 for specific parameters of the Algorithm

TABLE 2 Algorithm parameter Table

Parameter(s)	Numerical value
		Number of iterations	600
Size of population	8000
		wmax	0.85
wmin	0.35
		D	24
c1s	0.5
		c2s	2.5
c1l	2.5
		c2l	0.5

To verify the validity of the proposed method, simulation verification was performed in MATLAB simulation tools. Fig. 5 and 6 are a photovoltaic power generation curve of the light storage inverter for 24 hours and a power utilization curve of the same day, respectively. Through the program operation of the algorithm, the final scheduling simulation result is shown in fig. 7, and the result of the objective function (daily power expenditure) obtained at this time is 18.87 yuan. Meanwhile, the waveform obtained by using the RT-LAB semi-physical simulation platform to realize the scheduling scheme is shown in FIG. 8. While the resulting scheduling schemes using a conventional penalty function with penalty factors 50, 500 and 5000 of several common orders of magnitude are shown in fig. 9, 10 and 11, with the resulting objective function results being 21.01, 22.59 and 20.18, respectively.

According to the simulation result, the method provided by the invention reasonably plans the energy flow in the light storage inverter system, and effectively reduces the daily power expenditure of users.

Finally, the description is as follows: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims (7)

1. The method is characterized by comprising the following steps of:

step 1: building a topological structure of the household photovoltaic energy storage inverter;

step 2: and (4) establishing a mathematical model for the energy scheduling problem of the light storage inverter system. The mathematical model of the optimization problem comprises two parts, an objective function and constraint conditions. The objective function represents the daily power expenditure of a user, the constraint condition represents the limiting condition which must be met by the system, and the optimization objective is to solve the minimum value of the mathematical model, namely the minimum value of the daily power expenditure of the user;

and step 3: constructing an expression, namely an adaptive penalty function expression, of which the penalty factor is automatically updated in each iteration of the optimization algorithm;

and 4, step 4: and (4) obtaining a new target function expression in each iteration of the optimization algorithm according to the self-adaptive penalty function expression obtained in the step (3), and optimizing by utilizing a particle swarm optimization algorithm.

2. The method according to claim 1, wherein the topology of the household photovoltaic energy storage inverter in step 1 is as follows:

the photovoltaic power generation unit and the storage battery energy storage unit are connected to the input of the inverter through the preceding stage boosting direct current converter module and the resonance converter module respectively, and the storage battery can receive energy from other parts to be charged and can also discharge and supply energy to a load after being inverted. The output of the inverter is connected to the grid and the household load side.

3. The method of claim 1, wherein the objective function in step 2 is calculated as follows:

in 24 hours a day, a user can purchase electricity from the power grid or sell electricity to the power grid, but the prices of electricity purchase and electricity sale are different at different times. Due to planning, installation and maintenance costs, such as purchasing photovoltaic panels, storage batteries and the like, are sinking costs, and are not considered here. The final objective function, i.e. the daily power expenditure of the user, is thus expressed as:

wherein, J_cFor the total cost in period T, where T is 24 hours; e_buyThe amount of power purchased from the grid for the tth hour; f. of_buyA price to purchase electricity for the tth hour; e_sellThe amount of electricity sold to the grid for the tth hour; f. of_sellThe price of electricity sold at the t hour.

4. The method of claim 1, wherein the step 2 further comprises the constraint conditions of the light storage inverter system: system power balance constraints, grid output power constraints and storage battery related constraints.

The expression of the system power balance constraint is:

P_g+P_b+P_pv＝P_load

the expression of the power grid output power constraint is as follows:

-P_{g_max}≤P_g(t)≤P_{g_max}

the expression for the battery-related constraint is:

P_{bc_max}≤P_b(t)≤P_{bd_max}

DoD_min≤DoD(t)≤DoD_max

SoC_min≤SoC(t)≤SoC_max

wherein, P_pvIs the power of the photovoltaic power generation unit, P_bIf the term is positive value, the term indicates that the current state of the storage battery is discharging, and if the term is negative value, the term indicates that the current state of the storage battery is charging, P_gIf the item is a positive value, the current state indicates that the user purchases electricity from the power grid, and if the item is a negative value, the user sells electricity from the power grid. P_loadRepresenting the home load power. P_{g_max}For maximum power that the grid can deliver, P_{bc_max}And P_{dc_max}The maximum charging power and the maximum discharging power of the battery are shown, the DoD represents the discharging depth of the battery, and the SoC represents the state of charge of the battery; σ represents the self-discharge rate of the battery; p_BThe rated power of the battery.

5. The method according to claim 1, wherein the adaptive penalty function expression in step 3 is:

p(δ)＝e^1000|δ|-1

wherein, δ is the range of the result obtained after each iteration of the optimization algorithm exceeding the constraint condition, and p (δ) represents the penalty factor value when the error is δ.

6. The method of claim 1, wherein the new objective function expression in step 4 is:

7. the method according to claim 1, wherein the step 4 further comprises a particle swarm optimization algorithm process:

a: initializing a particle swarm including speed and position information of each particle;

b: calculating the fitness of each particle, namely an objective function result;

c: for each particle, its fitness value is compared with the individual optimal position p it has passed through_bestIn comparison, if more optimal, it is taken as the current individual optimal position p_best；

d: for each particle, its fitness value is compared with the global optimum g it has passed through_bestComparing, if more optimal, taking it as the current global optimal position g_best；

e: the velocity and position of the particles are adjusted according to the following two equations:

V_i＝w×V_i+c₁×rand()×(pbest_i-x_i)+c₂×rand()×(gbest_i-x_i)

x_i＝x_i+V_i

wherein w is an inertial weight factor; c. C₁And c₂The maximum step length of the flight of the particles to the individual optimum position and the global optimum position can be respectively adjusted to determine the influence of the individual experience and the group experience on the motion condition of the particles as a learning factor; rand () is a random number between 0 and 1, p_bestiRepresenting the individual optimum position of the current particle, g_bestiRepresenting the current global optimum position.

f: and c, ending the flow when the iteration times or the precision condition is reached, and otherwise, returning to the step b.

Images