CN104765967A - Multi-objective optimizing design method of off-grid hybrid renewable energy system - Google Patents

Abstract

The invention discloses a multi-objective optimizing design method of an off-grid hybrid renewable energy system. According to the method, the illumination, the wind speed, the temperature and other meteorological data of a certain region are considered, the annual cost of the hybrid renewable energy system and the power supply defect rate minimization serve as objectives, and a double-objective optimization model with constraints and suitable for the hybrid renewable energy system is built, so that the number of photovoltaic panels and the mounting inclination angle, the number of wind turbines and the mounting height and the number of energy storage devices and diesel generators for forming the optimizing configuration of the system can be obtained. The model is solved through an evolved multi-objective optimizing algorithm NSGAII, and a set of Pareto optimal solutions is obtained; finally the final implementation scheme is determined according to the reference information of the decision maker. HRES planning is conducted through the multi-objective optimizing method, and the method more accords with actual situations and is higher in feasibility.

Description

A kind of multi-objective optimization design of power method from net mixing renewable energy system

Technical field

The present invention relates to a kind of planning and designing method from net mixing renewable energy system, specifically consider the quantity of each assembly in energy resource system and the configuration of associated component mounted angle, height etc., each target of optimization system.

Background technology

The energy is the important foundation that the mankind depend on for existence and development.Enter 21 century, the development rapidly of economic society makes energy demand constantly increase, and fossil energy such as coal, oil etc. also constantly reduce along with consuming fast, and the problem of environmental pollution that a large amount of uses of these fossil energies cause is day by day serious.Energy problem and environmental pollution become the key factor of restriction social sustainable development day by day.In the face of the dual-pressure that energy problem and environmental problem are brought, the one greatly develop new forms of energy, strengthening utilizing regenerative resource to become socio-economic development in the urgent need to.

Mixing renewable energy system (Hybrid Renewable Energy Systems, HRES) be a kind of system combinationally using different types of regenerative resource and fossil energy in an appropriate manner, the shortcomings such as the intermittence of single regenerative resource, instability can be overcome.Specifically comprise photovoltaic generation, wind-power electricity generation, diesel-driven generator generating and the energy resource system of energy storage device.Refer to that this system does not have connecting system electrical network from net.The remote districts on island, mountain area and so on are generally applicable to from net mixing renewable energy system.Usually user is fewer in these areas, and duty factor is lower, and it is more difficult and economical not that present stage is incorporated to bulk power grid.

Planning and design about HRES can be divided into single object optimization and multi-objective Optimization, and at present major part research is all the optimization considering single goal, as minimization system total cost, maximum reliability, minimize greenhouse gas emissions etc.But from actual angle, a planning and design normally multi-objective optimization question of HRES, namely multiple target such as cost, reliability etc. needs to optimize simultaneously, need the quantity, photovoltaic panel mounted angle, assembling height etc. of considering each assembly in commingled system in addition, therefore the planning and design of HRES are a multivariate, discrete, nonlinear complicated optimum problem.

Summary of the invention

Due to the complicacy of multi-objective optimization question, traditional optimized algorithm such as linear programming, gradient method etc. can not be dealt with problems effectively, for the defect that prior art exists, the object of the invention is to propose a kind of multi-objective optimization design of power method from net mixing renewable energy system.

Technical scheme of the present invention is:

The first step, the optimization aim of design HRES and constraint function, set up the multi-objective optimization design of power model of HRES;

Supply miss rate for target with the annualized cost of minimization system and system power, build two objective optimisation problems of belt restraining, as follows:

min{F _cost,F _reliablility}

s.t.(N _pv,N _wg,N _bat,N _dg)≥0 (1)

H _low≤H _wg≤H _high

0°≤β≤90°

Wherein F _costthe annual cost that expression system is total, comprising: the Operation and Maintenance cost in system in the initial outlay cost of various kinds of equipment, use procedure and the alternative costs of equipment component; F _reliabilityrepresent system power supply miss rate, the ratio that namely system time that can not meet load is shared in simulation time, it is higher to be worth less expression system reliability; N _pv, N _wg, N _bat, N _dgrepresent the quantity of photovoltaic panel, blower fan, electric battery and diesel-driven generator to be optimized in HRES system respectively; H _wgrepresent the setting height(from bottom) of blower fan, β represents the mounted angle of solar energy photovoltaic panel;

Suppose that the model of each equipment forming HRES system is known, namely the character of photovoltaic panel, wind energy conversion system, electric battery and diesel-driven generator and correlation parameter are determined, wherein photovoltaic panel parameter comprises open-circuit voltage, short-circuit current, maximum working voltage, maximum operating currenbt, operating temperature ratings (NCOT), initial outlay cost, operating maintenance cost and serviceable life, fan parameter comprises rated power, initial outlay cost, operating maintenance cost and serviceable life, energy storage device and battery parameter comprise rated capacity, voltage, maximum depth of discharge (DOD), initial outlay cost, operating maintenance cost, alternative costs and serviceable life, diesel-driven generator parameter comprises rated power, initial outlay cost, operating maintenance cost per hour and serviceable life,

According to above-mentioned known each equipment-related data, set up for off-grid HRES multi-objective optimization design of power model;

HRES planning and design model relates to the variable of multiple type, comprise photovoltaic panel quantity, blower fan quantity, electric battery quantity, diesel engine quantity, assembling height and photovoltaic panel mounted angle, above-mentioned variable is encoded, coding rule is: if coding is shown as (22,8,28,3,23.55,59.76), so use the photovoltaic panel that 22 given, 8 aerogenerators, 28 electric battery, 3 diesel-driven generators, assembling height is 23.55 meters simultaneously, and photovoltaic panel mounted angle is 59.76 °;

In conjunction with the life-span of each equipment, with 25 years for the project cycle, HRES system annual cost objective function is:

ACS＝C _ainv(PV+WG+Tower+BAT+DG)

+C _aom(PV+WG+Tower+BAT+DG) (2)

+C _arep(BAT)

Wherein C _ainvannual initial outlay cost, C _aomannual Operation and maintenance cost, C _arepbe year alternative costs, calculate with following formula respectively:

C _ainv＝ΣC _inv·CRF(i,L _com) (3)

$\mathrm{CRF}(i,L_{\mathrm{com}})=\frac{i\·{(1+i)}^{L_{\mathrm{com}}}}{{(1+i)}^{L_{\mathrm{com}}}-1}---\left(4\right)$

$i=\frac{i_{\mathrm{nom}}-f}{1+f}---\left(5\right)$

Wherein C _invbe the initial outlay cost of each element, CRF is the capital recovery factor, L _com(year) be component life, i is a year true rate of interest, i _nombe norminal interest rate, f is a year inflation;

C _arep＝C _rep·SFF(i,L _rep) (6)

$\mathrm{SFF}(i,L_{\mathrm{rep}})=\frac{i}{{(1+i)}^{L_{\mathrm{rep}}}-1}---\left(7\right)$

C _aom(n)＝C _aom(1)·(1+f) ⁿ(8)

Wherein C _repbe the alternative costs of each element, SFF is the reticent fund factor, L _repthat element replaces the life-span, C _aomn () is the Operation and Maintenance cost of 1 year;

Power supply disappearance probabilistic goal function in HRES one-year age is:

$\mathrm{LPSP}=\frac{{\mathrm{\&Sigma;}}_{t=0}^{T}T(P_{\mathrm{avail}}\left(t\right)<P_{\mathrm{load}}\left(t\right))}{T}---\left(9\right)$

Wherein T is 1 year total hourage that is 8760, P _avail(t) and P _load(t) be respectively each simulation time step-length can power supply and loading demand; Load is obtained by user side, and power supply comprises the supply of photovoltaic generation and wind-power electricity generation, energy storage supply and diesel-driven generator and supplies four parts; Power supply process is: first by photovoltaic generation and the direct supply load of wind-power electricity generation to satisfy the demands, when the power of photovoltaic generation and solar electrical energy generation is greater than load, unnecessary electricity just charges to electric battery; Contrary to generated energy can not meet load, first discharged by energy storage device, electric battery reaches maximum depth of discharge when still can not meet load, and diesel-driven generator uses as standby power supply, and remaining still unappeasable load can be cut off with protection system;

The power of system represents for application following formula:

P _avail(t)＝P _pv(t)+P _wg(t)+P _bat(t)+P _dg(t) (10)

Wherein the power of photovoltaic generation and blower fan generating is calculated by following formula respectively:

$T_C\left(t\right)=T_A\left(t\right)+\frac{\mathrm{NCOT}-20}{800}{S}_p(t,\mathrm{\&beta;})---\left(11\right)$

V _OC(t)＝V _OC,STC-K _V·T _C(t) (13)

P _pv(t,β)＝N _S·N _P·V _OC(t)·I _SC(t,β)·FF(t)

＝N _pv·V _OC(t)·I _SC(t,β)·FF(t) (14)

Wherein T _ct () is the photovoltaic battery temperature at time t, T _aenvironment temperature when () is time t t, NCOT is the nominal cell working temperature that manufacturer provides, S _p(t, β) is perpendicular to the solar radiation on photovoltaic panel inclined surface, I _{sC, STC}and V _{oC, STC}(temperature 25, the solar radiation 1kW/m of module under standard test condition ²) short-circuit current and open-circuit voltage, K _iand K _vit is corresponding temperature coefficient; P _pv(t, β) comprises N by one _sindividual series connection, N _pthe output power of the photovoltaic array of individual module in parallel, FF (t) is fill factor, curve factor;

$P_{\mathrm{wg}}\left(t\right)=\left\{\begin{array}{c}0,v<V_c\\ \frac{1}{2}{C}_P\mathrm{\&rho;}{A}_{\mathrm{WG}}{v}^3,V_c\&le;v<V_r\\ P_{\mathrm{WGR}},V_r\&le;v<V_f\\ 0,v\&GreaterEqual;V_f\end{array}\right.---\left(15\right)$

Wherein v is the wind speed in each moment, C _pbe fan performance coefficient, ρ is atmospheric density, A _wGthe inswept area of rotor, P _wGRthe rated power of blower fan; V _cthe cutting speed of blower fan, V _rthe wind rating of blower fan, V _fit is the cut-out wind speed of blower fan;

The solar radiation of using in HRES planning, wind speed profile and temperature data carry out according to the historical data of research area the average data that namely respective handling got over 10 years, or according to distribution function generation simulated data;

Second step, utilize NSGAII Algorithm for Solving HRES plan model, idiographic flow is:

(1) algorithm parameter is arranged: mainly comprise population scale and end condition, population N is set to 100 here, and end condition adopts maximum operation algebraically, is set to maxGen=100;

(2) initialization population, stochastic generation N=100 initial parent population S; Each individual x has 6 codings, i.e. x=(N _pv, N _wg, N _bat, N _dg, H _wg, β), wherein 0≤N _pv≤ 30,0≤N _wg≤ 20,0≤N _bat≤ 30,0≤N _dg≤ 10,5≤H _wg≤ 30,0≤β≤90;

(3) if meet end condition, then stop calculating, export current non-dominant disaggregation; Otherwise based on current population S, produce progeny population Sc by Genetic Recombination operator, scale is also N.Concrete operation step:

A () is for each individual x in current population S _i, in conjunction with two other individual x of Stochastic choice _mand x _n, through type (17) produces new individual x _new.Wherein represent a new individual kth variate-value, here k=[1,2 ..., 6]; F and CR is respectively two parameters of this operation, is set to 0.9 and 0.05 here; Rand represents the random number being positioned at interval (0,1); k _randwhat represent a random generation is positioned at interval [1,6] integer; Floor () represents downward bracket function.

$x_{\mathrm{temp}}^k=\left\{\begin{array}{cc}{x}_i^k+F\&times;(x_m^k-x_n^k)& \mathrm{ifrand}<\mathrm{CRork}=k_{\mathrm{rand}}\\ x_i^k& \end{array}\right.---\left(16\right)$

$x_{\mathrm{new}}^k=\left\{\begin{array}{cc}\mathrm{floor}\left(x_{\mathrm{temp}}^k\right)& \mathrm{ifk}<5\\ x_{\mathrm{temp}}^k& \end{array}\right.$

If b new individuality that () produces is infeasible solutions, namely variate-value is beyond the bound of definition, then take following measures to be modified to feasible solution.Wherein ub ^kand lb ^krepresent the bound of a kth variable respectively.All x _newconstitute progeny population Sc.

$x_{\mathrm{new}}^k=\frac{({\mathrm{ub}}^k+{\mathrm{lb}}^k)}{2}---\left(17\right)$

(4) merged by parent S and filial generation Sc, the scale of obtaining is the conjunction population S of 2N _all=S ∪ Sc, carries out non-bad layering to the individuality in R, then calculates the individual local congestion distance of each non-bad layer, last according to the non-bad layer residing for individuality and crowding distance, sorts to all individualities.

Wherein: the non-bad layered approach of described individuality is specially:

A () is by the individual normalization in population: obtain each objective function f _mmaximal value, max (f _m) and minimum value, min (f _m), then according to following formula, each individual goal functional value is transformed into interval [0,1].

${\stackrel{\&OverBar;}{f}}_i\left(x\right)=\frac{f_i\left(x\right)-\min \left(f_i\right)}{\max \left(f_i\right)-\min \left(f_i\right)}i=\mathrm{1,2},...,M---\left(18\right)$

B () finds out synthesis population S _allin by the individuality of any individuality constraint Pareto domination, and be kept at set A ₁in, as the first non-bad layer; We claim individual x constrain domination individual y, when one of them condition following meets: (i) individual x and y all meets constraint condition and x < y; (ii) individual x meets constraint condition, and y does not meet constraint condition; (iii) individual x and y does not all meet constraint condition, and the degree that individual x violates constraint condition is less than the degree that individual y violates constraint condition.X < y represents that individual x arranges individual y, and and if only if $\&ForAll;i\&Element;\mathrm{1,2},...,M,f_i\left(x\right)\&le;f_i\left(y\right)\&cap;\&Exists;j\&Element;\mathrm{1,2},...,M,$ F _j(x) < f _j(y), M represents objective function number; Namely individual x is not worse than individual y on all objective functions, and x is at least better than y on an objective function.

C () is from S _allmiddle removing is all in set A ₁in individuality, residue synthesis population be designated as S _alla ₁, repeat (a), find out population S _alla ₁in by the individuality of any individuality constraint Pareto domination, and be kept at set A ₂in, as the second non-bad layer.

(d) the like, until whole population is layered complete.

Described crowding distance (being designated as dist), can be seen as individual x intuitively _icomprise individual x around _ibut do not comprise the minimum rectangle of other individualities; Crowding distance is less, illustrates individual denser around.Its circular is as follows:

A () is for each objective function f _m, the individuality in population is sorted;

B () (namely has minimum f for border individuality _mthe individuality of value), definition crowding distance dist is infinitely great;

Flash trimming other individual x out-of-bounds in (c) same non-bad layer _icrowding distance be

$\mathrm{dist}\left(i\right)=\underset{m=1}{\overset{2}{\mathrm{\&Sigma;}}}\frac{f_m^{i-1}-f_m^{i+1}}{f_m^{\max }-f_m^{\min }}---\left(19\right)$

Wherein with represent objective function f in current population _mmaximum (little) value; with i-th-1 and i+1 individual target function value.

Described according to the non-bad layer residing for individuality and crowding distance sort method, specifically refer to: the individuality that (i) is in the i-th non-bad layer is better than the individuality of jth (j > i) non-bad layer; (ii) be in the individuality of same non-bad layer, the individuality that crowding distance is large is more excellent.

(5) according to ranking results choose preferably N number of body as new parent population S.

(6) repeat (3) to (5) step, until meet end condition, namely reach maximum operation algebraically, the non-dominant exported in S is individual as solved solution;

3rd step, in conjunction with the preference information of decision maker, from multiple Pareto optimum solution, selects an embodiment as last HRES;

When decision maker payes attention to cost compare, then, when meeting cost restriction, select the scheme that reliability is the highest, the scheme that namely power supply miss rate is minimum; When the heavier viewing system power supply reliability of decision maker, then under the requirement meeting power supply miss rate, the scheme of selective system cost minimization.

In the present invention: refer to from the planning of net mixing renewable energy system: how to determine that the photovoltaic panel of suitable quantity, blower fan, diesel-driven generator, energy storage device and mounting means are to make whole energy resource system As soon as possible Promising Policy area load demand, simultaneously most economical, most environmental protection.

Multi-objective optimization question refers to: be optimized multiple target simultaneously, owing to normally linking together between each target and conditioning each other, compete mutually, namely the improvement of certain target may cause the degeneration of other targets, be difficult to find an optimum solution truly to make each optimization aim reach optimum simultaneously, therefore the set of the optimum solution of a multi-objective optimization question normally noninferior solution, i.e. Pareto optimal solution set.The core solving multi-objective optimization question finds the uniform Pareto optimum solution of a component cloth.

Evolutionary Multiobjective Optimization is the optimized algorithm based on population come by simulating or disclose some spontaneous phenomenon or process development, and its thought and content relate to mathematics, biology and Computer Subject etc.Such algorithm does not rely on gradient information, once runs and can find one group of Pareto optimum solution, has the features such as the overall situation, parallel, efficient, robust and highly versatile.The effective ways solving complex nonlinear multi-objective optimization question.It is different from the method for traditional process multi-objective optimization question, as weighted method, leash law, Objective Programming etc., these classic methods are by structure evaluation function, multi-objective problem is converted into single-object problem, then utilizes general method for solving to calculate a solution of problem.NSGAII is a classic algorithm in Evolutionary Multiobjective Optimization.

Advantageous Effects of the present invention is:

(1) the present invention adopts Multipurpose Optimal Method to carry out the planning of HRES, more tallies with the actual situation, feasibility is stronger.

(2) adopt Evolutionary Multiobjective Optimization during Solve problems of the present invention, one group of each advantageous Pareto optimum solution can be found simultaneously, for decision maker according to different situations, select to implement.

Accompanying drawing explanation

Fig. 1 typical case is from net mixing renewable energy system schematic diagram

Fig. 2 HRES design proposal is encoded

Fig. 3 Evolutionary Multiobjective Optimization NSGAII solves the process flow diagram of HRES planning problem

The Pareto non-dominant disaggregation that Fig. 4 HRES plans

Embodiment

Multi-objective optimization design of power method from net mixing renewable energy system provided by the invention, concrete steps are as follows:

The first step, the optimization aim of design HRES and constraint function, set up the multi-objective optimization design of power model of HRES.

One typical from net wind-solar-diesel storage energy resource system as shown in Figure 1.For this kind of canonical system, supply miss rate for target with the annualized cost of minimization system and system power, consider the constraint condition of relevant Decision variable, build two objective optimisation problems of belt restraining, as follows

min{F _cost,F _reliablility}

s.t.(N _pv,N _wg,N _bat,N _dg)≥0 (20)

H _low≤H _wg≤H _high

0°≤β≤90°

Wherein F _costthe annual cost that expression system is total, comprising: the Operation and Maintenance cost in system in the initial outlay cost of various kinds of equipment, use procedure and the alternative costs of equipment component; F _reliabilityrepresent system power supply miss rate, namely system can not meet the ratio of the time of load in simulation time shared by (for 1 year), and it is higher to be worth less expression system reliability.N _pvdeng the quantity of the various kinds of equipment such as photovoltaic in expression HRES system, the quantity of photovoltaic panel namely to be optimized, blower fan, electric battery and diesel-driven generator; H _wgrepresent the setting height(from bottom) of blower fan, β represents the mounted angle of solar energy photovoltaic panel.

Suppose that the model of each equipment forming HRES system is known, namely the character of photovoltaic panel, wind energy conversion system, electric battery and diesel-driven generator and correlation parameter are determined.The simulated data of these equipment is respectively as shown in table 1-4.Wherein photovoltaic panel parameter comprises open-circuit voltage, short-circuit current, maximum working voltage, maximum operating currenbt, operating temperature ratings (NCOT), initial outlay cost and operating maintenance cost etc., fan parameter comprises rated power, initial outlay cost, operating maintenance cost, serviceable life etc., energy storage device (electric battery) parameter comprises rated capacity, voltage, maximum depth of discharge (DOD), initial outlay cost, operating maintenance cost and alternative costs etc., diesel-driven generator correlation parameter comprises rated power, initial outlay cost, operating maintenance cost per hour and serviceable life etc.

Table 1 photovoltaic panel correlation parameter (supposes that every block photovoltaic panel area is 1 meter ²)

Table 2 wind energy conversion system correlation parameter

Table 3 energy storage device correlation parameter

Table 4 diesel engine correlation parameter

According to the equipment-related data provided in table 1 to 4, set up for off-grid HRES multi-objective optimization design of power model, specifically, refer to the optimum composition method and photovoltaic panel mounted angle, assembling height etc. of finding out various kinds of equipment quantity, make HRES not only economy but also have higher reliability.

HRES planning and design model relates to the variable of multiple type, as: the quantity (integer variable) using certain type equipment, photovoltaic panel mounted angle, assembling height (real variable).The coding rule of a concrete embodiment is as Fig. 2.

Scheme code considers 6 variablees, comprises photovoltaic panel quantity, blower fan quantity, electric battery quantity, diesel engine quantity, assembling height and photovoltaic panel mounted angle.Following face code is example, if coding is shown as (22,8,28,3,23.55,59.76), the photovoltaic panel that 22 given is so used, 8 aerogenerators, 28 electric battery, 3 diesel-driven generators, assembling height is 23.55 meters simultaneously, and photovoltaic panel mounted angle is 59.76 °.

In conjunction with the life-span of each equipment, with 25 years for the project cycle, only have electric battery to need to consider alternative costs thus in simulation process, HRES system annual cost objective function is:

ACS＝C _ainv(PV+WG+Tower+BAT+DG)

+C _aom(PV+WG+Tower+BAT+DG) (21)

+C _arep(BAT)

Wherein C _ainvannual initial outlay cost, C _aomannual Operation and maintenance cost, C _arepbe year alternative costs, can calculate with following formula respectively:

C _ainv＝ΣC _inv·CRF(i,L _com) (22)

$\mathrm{CRF}(i,L_{\mathrm{com}})=\frac{i\&CenterDot;{(1+i)}^{L_{\mathrm{com}}}}{{(1+i)}^{L_{\mathrm{com}}}-1}---\left(23\right)$

$i=\frac{i_{\mathrm{nom}}-f}{1+f}---\left(24\right)$

Wherein C _invbe the initial outlay cost of each element, CRF is the capital recovery factor, L _com(year) be component life, i is a year true rate of interest, i _nombe norminal interest rate, f is a year inflation.

C _arep＝C _rep·SFF(i,L _rep) (25)

$\mathrm{SFF}(i,L_{\mathrm{rep}})=\frac{i}{{(1+i)}^{L_{\mathrm{rep}}}-1}---\left(26\right)$

C _aom(n)＝C _aom(1)·(1+f) ⁿ(27)

Wherein C _repbe the alternative costs of each element, SFF is the reticent fund factor, L _repthat element replaces the life-span, C _aomn () is the Operation and Maintenance cost of 1 year.

Power supply disappearance probabilistic goal function in HRES one-year age is:

$\mathrm{LPSP}=\frac{{\mathrm{\&Sigma;}}_{t=0}^{T}T(P_{\mathrm{avail}}\left(t\right)<P_{\mathrm{load}}\left(t\right))}{T}---\left(28\right)$

Wherein T is 1 year total hourage considering that is 8760, P _avail(t) and P _load(t) be respectively each simulation time step-length can power supply and loading demand.Load can be obtained by user side, and power supply comprises supply, four parts such as energy storage supply and diesel-driven generator supply of photovoltaic generation and wind-power electricity generation.Power supply process is: first by photovoltaic generation and the direct supply load of wind-power electricity generation to satisfy the demands, when the power of photovoltaic generation and solar electrical energy generation is greater than load, unnecessary electricity just charges to electric battery.Contrary to generated energy can not meet load, first discharged by energy storage device, electric battery reaches maximum depth of discharge when still can not meet load, and diesel-driven generator uses as standby power supply, and remaining still unappeasable load can be cut off with protection system.

The power supply of system can be represented by the formula:

P _avail(t)＝P _pv(t)+P _wg(t)+P _bat(t)+P _dg(t) (29)

Wherein the power of photovoltaic generation and blower fan generating can be calculated by following formula respectively:

$T_C\left(t\right)=T_A\left(t\right)+\frac{\mathrm{NCOT}-20}{800}{S}_p(t,\mathrm{\&beta;})---\left(30\right)$

V _OC(t)＝V _OC, _STC-K _V·T _C(t) (32)

P _pv(t,β)＝N _S·N _P·V _OC(t)·I _SC(t,β)·FF(t)

＝N _pv·V _OC(t)·I _SC(t,β)·FF(t) (33)

Wherein T _ct () is the photovoltaic battery temperature at time t, T _aenvironment temperature when () is time t t, NCOT is the nominal cell working temperature that manufacturer provides, S _p(t, β) is perpendicular to the solar radiation on photovoltaic panel inclined surface, I _{sC, STC}and V _{oC, STC}(temperature 25, the solar radiation 1kW/m of module under standard test condition ²) short-circuit current and open-circuit voltage, K _iand K _vit is corresponding temperature coefficient; P _pv(t, β) comprises N by one _sindividual series connection, N _pthe output power of the photovoltaic array of individual module in parallel, FF (t) is fill factor, curve factor.

$P_{\mathrm{wg}}\left(t\right)=\left\{\begin{array}{c}0,v<V_c\\ \frac{1}{2}{C}_P\mathrm{\&rho;}{A}_{\mathrm{WG}}{v}^3,V_c\&le;v<V_r\\ P_{\mathrm{WGR}},V_r\&le;v<V_f\\ 0,v\&GreaterEqual;V_f\end{array}\right.---\left(34\right)$

Wherein v is the wind speed in each moment, C _pbe fan performance coefficient, ρ is atmospheric density, A _wGthe inswept area of rotor, P _wGRthe rated power of blower fan.V _cthe cutting speed of blower fan, V _rthe wind rating of blower fan, V _fit is the cut-out wind speed of blower fan.

The solar radiation of using in HRES planning, wind speed profile and temperature data can be carried out respective handling according to the historical data of research area and obtain, and as got over the average data of 10 years, also can produce simulated data according to distribution function.

Second step, utilize NSGAII Algorithm for Solving HRES Combinatorial Optimization Model, as shown in Figure 3, idiographic flow is:

(1) algorithm parameter is arranged: mainly comprise population scale and end condition, population N is set to 100 here, and end condition adopts maximum operation algebraically, is set to maxGen=100.Certainly for different problems and problem scale, different parameters value can be selected.

(2) initialization population, stochastic generation N=100 initial parent population S.Each individual x has 6 codings, i.e. x=(N _pv, N _wg, N _bat, N _dg, H _wg, β), wherein 0≤N _pv≤ 30,0≤N _wg≤ 20,0≤N _bat≤ 30,0≤N _dg≤ 10,5≤H _wg≤ 30,0≤β≤90;

(3) if meet end condition, then stop calculating, export current non-dominant disaggregation; Otherwise based on current population S, produce progeny population Sc by Genetic Recombination operator, scale is also N.Concrete operation step:

A () is for each individual x in current population S _i, in conjunction with two other individual x of Stochastic choice _mand x _n, produce new individual x by following formula _new.Wherein represent a new individual kth variate-value, here k=[1,2 ..., 6]; F and CR is respectively two parameters of this operation, is set to 0.9 and 0.05 here; Rand represents the random number being positioned at interval (0,1); k _randwhat represent a random generation is positioned at interval [1,6] integer; Floor () represents downward bracket function.

$x_{\mathrm{temp}}^k=\left\{\begin{array}{cc}{x}_i^k+F\&times;(x_m^k-x_n^k)& \mathrm{ifrand}<\mathrm{CRork}=k_{\mathrm{rand}}\\ x_i^k& \end{array}\right.---\left(35\right)$

$x_{\mathrm{new}}^k=\left\{\begin{array}{cc}\mathrm{floor}\left(x_{\mathrm{temp}}^k\right)& \mathrm{ifk}<5\\ x_{\mathrm{temp}}^k& \end{array}\right.$

If b new individual infeasible solutions that () produces, then following measures is taked to be modified to feasible solution.Wherein ub ^kand lb ^krepresent the bound of a kth variable respectively.All x _newconstitute progeny population Sc.

$x_{\mathrm{new}}^k=\frac{({\mathrm{ub}}^k+{\mathrm{lb}}^k)}{2}---\left(36\right)$

(4) merged by parent S and filial generation Sc, the scale of obtaining is the conjunction population S of 2N _all=S ∪ Sc, to S _allin individuality carry out non-bad layering, then calculate the individual local congestion distance of each non-bad layer, last according to the non-bad layer residing for individuality and crowding distance, all individualities are sorted.

The non-bad layered approach of described individuality is specially:

A () is by the individual normalization in population: obtain each objective function f _mmaximal value, max (f _m) and minimum value, min (f _m), then according to following formula, each individual goal functional value is transformed into interval [0,1].

${\stackrel{\&OverBar;}{f}}_i\left(x\right)=\frac{f_i\left(x\right)-\min \left(f_i\right)}{\max \left(f_i\right)-\min \left(f_i\right)}i=\mathrm{1,2},...,M---\left(37\right)$

B () finds out synthesis population S _allin by the individuality of any individuality constraint Pareto domination, and be kept at set A ₁in, as the first non-bad layer; We claim individual x constrain domination individual y, when one of them condition following meets: (i) individual x and y all meets constraint condition and x < y; (ii) individual x meets constraint condition, and y does not meet constraint condition; (iii) individual x and y does not all meet constraint condition, and the degree that individual x violates constraint condition is less than the degree that individual y violates constraint condition.X < y represents that individual x arranges individual y, and and if only if $\&ForAll;i\&Element;\mathrm{1,2},...,M,f_i\left(x\right)\&le;f_i\left(y\right)\&cap;\&Exists;j\&Element;\mathrm{1,2},...,M,$ F _j(x) < f _j(y), M represents objective function number; Namely individual x is not worse than individual y on all objective functions, and x is at least better than y on an objective function.

C () is from S _allmiddle removing is all in set A ₁in individuality, residue synthesis population be designated as S _alla ₁, repeat (a), find out population S _alla ₁in by the individuality of any individuality constraint Pareto domination, and be kept at set A ₂in, as the second non-bad layer.

(d) the like, until whole population is layered complete.

Described crowding distance (being designated as dist), can be seen as individual x intuitively _icomprise individual x around _ibut do not comprise the minimum rectangle of other individualities; Crowding distance is less, illustrates individual denser around.Its circular is as follows:

A () is for each objective function f _m, the individuality in population is sorted;

B () (namely has minimum f for border individuality _mthe individuality of value), definition crowding distance dist is infinitely great;

Flash trimming other individual x out-of-bounds in (c) same non-bad layer _icrowding distance be

$\mathrm{dist}\left(i\right)=\underset{m=1}{\overset{2}{\mathrm{\&Sigma;}}}\frac{f_m^{i-1}-f_m^{i+1}}{f_m^{\max }-f_m^{\min }}---\left(38\right)$

Wherein with represent objective function f in current population _mmaximum (little) value; with i-th-1 and i+1 individual target function value.

Described according to the non-bad layer residing for individuality and crowding distance sort method, specifically refer to: the individuality that (i) is in the i-th non-bad layer is better than the individuality of jth (j > i) non-bad layer; (ii) be in the individuality of same non-bad layer, the individuality that crowding distance is large is more excellent.

(5) according to ranking results choose preferably N number of body as new parent population S.

(6) repeat (3) to (5) step, until meet end condition, namely reach maximum operation algebraically, the non-dominant exported in S is individual as solved solution;

According to the data in table 1 to 4, and other the known weather data in model is as illumination, wind speed and temperature etc., and load data also can be obtained by user side.With 1 year for emulation cycle, within one hour, be simulation step length, finally solve the most non-dominant collection of the Pareto obtained as shown in Figure 4.

3rd step, in conjunction with the preference information of decision maker, from multiple Pareto optimum solution, selects an embodiment as last HRES.Can consider from two kinds of situations: when decision maker payes attention to cost compare, then, when meeting cost restriction, select the scheme that reliability is the highest, the scheme that namely power supply miss rate is minimum; When the heavier viewing system power supply reliability of decision maker, then under the requirement meeting power supply miss rate, the scheme of selective system cost minimization.

In sum; although the present invention discloses as above with better enforcement; so itself and be not used to limit the present invention; any those of ordinary skill in the art; without departing from the spirit and scope of the present invention; when doing various change and retouching, the scope that therefore protection scope of the present invention ought define depending on claims is as the criterion.

Claims (4)

1., from a multi-objective optimization design of power method for net mixing renewable energy system, it is characterized in that comprising the following steps:

The first step, the optimization aim of design HRES and constraint function, set up the multi-objective optimization design of power model of HRES;

Supply miss rate for target with the annualized cost of minimization system and system power, build two objective optimisation problems of belt restraining, as follows:

min{F _cost,F _reliablility}

s.t.(N _pv,N _wg,N _bat,N _dg)≥0 (1)

H _low≤H _wg≤H _high

0°≤β≤90°

Wherein F _costthe annual cost that expression system is total, comprising: the Operation and Maintenance cost in system in the initial outlay cost of various kinds of equipment, use procedure and the alternative costs of equipment component; F _reliabilityrepresent system power supply miss rate, the ratio that namely system time that can not meet load is shared in simulation time, it is higher to be worth less expression system reliability; N _pv, N _wg, N _bat, N _dgrepresent the quantity of photovoltaic panel, blower fan, electric battery and diesel-driven generator to be optimized in HRES system respectively; H _wgrepresent the setting height(from bottom) of blower fan, β represents the mounted angle of solar energy photovoltaic panel;

Suppose that the model of each equipment forming HRES system is known, namely the character of photovoltaic panel, wind energy conversion system, electric battery and diesel-driven generator and correlation parameter are determined, wherein photovoltaic panel parameter comprises open-circuit voltage, short-circuit current, maximum working voltage, maximum operating currenbt, operating temperature ratings (NCOT), initial outlay cost, operating maintenance cost and serviceable life, fan parameter comprises rated power, initial outlay cost, operating maintenance cost and serviceable life, energy storage device and battery parameter comprise rated capacity, voltage, maximum depth of discharge (DOD), initial outlay cost, operating maintenance cost, alternative costs and serviceable life, diesel-driven generator parameter comprises rated power, initial outlay cost, operating maintenance cost per hour and serviceable life,

According to above-mentioned known each equipment-related data, set up for off-grid HRES multi-objective optimization design of power model;

HRES planning and design model relates to the variable of multiple type, comprise photovoltaic panel quantity, blower fan quantity, electric battery quantity, diesel engine quantity, assembling height and photovoltaic panel mounted angle, above-mentioned variable is encoded, coding rule is: if coding is shown as (22,8,28,3,23.55,59.76), so use the photovoltaic panel that 22 given, 8 aerogenerators, 28 electric battery, 3 diesel-driven generators, assembling height is 23.55 meters simultaneously, and photovoltaic panel mounted angle is 59.76 °;

In conjunction with the life-span of each equipment, with 25 years for the project cycle, HRES system annual cost objective function is:

ACS＝C _ainv(PV+WG+Tower+BAT+DG)

+C _aom(PV+WG+Tower+BAT+DG) (2)

+C _arep(BAT)

Wherein C _ainvannual initial outlay cost, C _aomannual Operation and maintenance cost, C _arepbe year alternative costs, calculate with following formula respectively:

C _ainv＝∑C _inv·CRF(i,L _com) (3)

$\mathrm{CRF}(i,L_{\mathrm{com}})=\frac{i\&CenterDot;{(1+i)}^{L_{\mathrm{com}}}}{{(1+i)}^{L_{\mathrm{com}}}-1}---\left(4\right)$

$i=\frac{i_{\mathrm{nom}}-f}{1+f}---\left(5\right)$

Wherein C _invbe the initial outlay cost of each element, CRF is the capital recovery factor, L _com(year) be component life, i is a year true rate of interest, i _nombe norminal interest rate, f is a year inflation;

C _arep＝C _rep·SFF(i,L _rep) (6)

$\mathrm{SFF}(i,L_{\mathrm{rep}})=\frac{i}{{(1+i)}^{L_{\mathrm{rep}}}-1}---\left(7\right)$

C _aom(n)＝C _aom(1)·(1+f) ⁿ(8)

Wherein C _repbe the alternative costs of each element, SFF is the reticent fund factor, L _repthat element replaces the life-span, C _aomn () is the Operation and Maintenance cost of 1 year;

Power supply disappearance probabilistic goal function in HRES one-year age is:

$\mathrm{LPSP}=\frac{{\mathrm{\&Sigma;}}_{t=0}^{T}T(P_{\mathrm{avail}}\left(t\right)<P_{\mathrm{load}}\left(t\right))}{T}---\left(9\right)$

Wherein T is 1 year total hourage that is 8760, P _avail(t) and P _load(t) be respectively each simulation time step-length can power supply and loading demand; Load is obtained by user side, and power supply comprises the supply of photovoltaic generation and wind-power electricity generation, energy storage supply and diesel-driven generator and supplies four parts; Power supply process is: first by photovoltaic generation and the direct supply load of wind-power electricity generation to satisfy the demands, when the power of photovoltaic generation and solar electrical energy generation is greater than load, unnecessary electricity just charges to electric battery; Contrary to generated energy can not meet load, first discharged by energy storage device, electric battery reaches maximum depth of discharge when still can not meet load, and diesel-driven generator uses as standby power supply, and remaining still unappeasable load can be cut off with protection system;

The power of system represents for application following formula:

P _avail(t)＝P _pv(t)+P _wg(t)+P _bat(t)+P _dg(t) (10)

Wherein the power of photovoltaic generation and blower fan generating is calculated by following formula respectively:

$T_C\left(t\right)=T_A\left(t\right)+\frac{\mathrm{NCOT}-20}{800}{S}_p(t,\mathrm{\&beta;})---\left(11\right)$

V _OC(t)＝V _OC,STC-K _V·T _C(t) (13)

P _pv(t,β)＝N _S·N _P·V _OC(t)·I _SC(t,β)·FF(t) (14)

＝N _pv·V _OC(t)·I _SC(t,β)·FF(t)

Wherein T _ct () is the photovoltaic battery temperature at time t, T _aenvironment temperature when () is time t t, NCOT is the nominal cell working temperature that manufacturer provides, S _p(t, β) is perpendicular to the solar radiation on photovoltaic panel inclined surface, I _{sC, STC}and V _{oC, STC}(temperature 25, the solar radiation 1kW/m of module under standard test condition ²) short-circuit current and open-circuit voltage, K _iand K _vit is corresponding temperature coefficient; P _pv(t, β) comprises N by one _sindividual series connection, N _pthe output power of the photovoltaic array of individual module in parallel, FF (t) is fill factor, curve factor;

$P_{\mathrm{wg}}\left(t\right)=\left\{\begin{array}{cc}0,& v<V_c\\ \frac{1}{2}{C}_P{\mathrm{\&rho;A}}_{\mathrm{WG}}{v}^3,& V_c\&le;v<V_r\\ P_{\mathrm{WGR}},& V_r,\&le;v<V_f\\ 0,& v\&GreaterEqual;V_f\end{array}\right.---\left(15\right)$

Wherein v is the wind speed in each moment, C _pbe fan performance coefficient, ρ is atmospheric density, A _wGthe inswept area of rotor, P _wGRthe rated power of blower fan; V _cthe cutting speed of blower fan, V _rthe wind rating of blower fan, V _fit is the cut-out wind speed of blower fan;

The solar radiation of using in HRES planning, wind speed profile and temperature data carry out according to the historical data of research area the average data that namely respective handling got over 10 years, or according to distribution function generation simulated data;

Second step, utilize NSGAII Algorithm for Solving HRES plan model, idiographic flow is:

(1) algorithm parameter is arranged: mainly comprise population scale and end condition, population N is set to 100 here, and end condition adopts maximum operation algebraically, is set to maxGen=100;

(2) initialization population, stochastic generation N=100 initial parent population S; Each individual x has 6 codings, i.e. x=(N _pv, N _wg, N _bat, N _dg, H _wg, β), wherein 0≤N _pv≤ 30,0≤N _wg≤ 20,0≤N _bat≤ 30,0≤N _dg≤ 10,5≤H _wg≤ 30,0≤β≤90;

(3) if meet end condition, then stop calculating, export current non-dominant disaggregation; Otherwise based on current population S, produce progeny population Sc by Genetic Recombination operator, scale is also N; Concrete operation step:

A () is for each individual x in current population S _i, in conjunction with two other individual x of Stochastic choice _mand x _n, through type (17) produces new individual x _new; Wherein represent a new individual kth variate-value, here k=[1,2 ..., 6]; F and CR is respectively two parameters of this operation, is set to 0.9 and 0.05 here; Rand represents the random number being positioned at interval (0,1); k _randwhat represent a random generation is positioned at interval [1,6] integer; Floor () represents downward bracket function;

$\begin{array}{c}{x}_{\mathrm{temp}}^k=\left\{\begin{array}{c}\begin{array}{ccc}{x}_i^k+F\&times;(x_m^k-x_n^k)& \mathrm{if}& \mathrm{rand}<\mathrm{CR\; or\; k}=k_{\mathrm{rand}}\end{array}\\ x_i^k\end{array}\right.\\ x_{\mathrm{new}}^k=\left\{\begin{array}{c}\begin{array}{ccc}\mathrm{floor}\left(x_{\mathrm{temp}}^k\right)& \mathrm{if}& k<5\end{array}\\ x_{\mathrm{temp}}^k\end{array}\right.\end{array}---\left(17\right)$

If b new individuality that () produces is infeasible solutions, namely variate-value is beyond the bound of definition, then take following measures to be modified to feasible solution; Wherein ub ^kand lb ^krepresent the bound of a kth variable respectively; All x _newconstitute progeny population Sc;

$x_{\mathrm{new}}^k=\frac{({\mathrm{ub}}^k+{\mathrm{lb}}^k)}{2}---\left(18\right)$

(4) merged by parent S and filial generation Sc, the scale of obtaining is the conjunction population S of 2N _all=S ∪ Sc, carries out the non-bad layering of individuality to the individuality in R, then calculates the individual local congestion distance of each non-bad layer, last according to the non-bad layer residing for individuality and crowding distance, sorts to all individualities;

(5) according to ranking results choose preferably N number of body as new parent population S;

(6) repeat (3) to (5) step, until meet end condition, namely reach maximum operation algebraically, the non-dominant exported in S is individual as solved solution;

3rd step, in conjunction with the preference information of decision maker, from multiple Pareto optimum solution, selects an embodiment as last HRES;

When decision maker payes attention to cost compare, then, when meeting cost restriction, select the scheme that reliability is the highest, the scheme that namely power supply miss rate is minimum; When the heavier viewing system power supply reliability of decision maker, then under the requirement meeting power supply miss rate, the scheme of selective system cost minimization.

2. the multi-objective optimization design of power method from net mixing renewable energy system according to claim 1, is characterized in that in (4) step by step of described second step, and individual non-bad layered approach is specially:

A () is by the individual normalization in population: obtain each objective function f _mmaximal value, max (f _m) and minimum value, min (f _m), then according to following formula, each individual goal functional value is transformed into interval [0,1];

${\stackrel{\&OverBar;}{f}}_i\left(x\right)=\frac{f_i\left(x\right)-\min \left(f_i\right)}{\max \left(f_i\right)-\min \left(f_i\right)},i=\mathrm{1,2},...,M---\left(16\right)$

B () finds out synthesis population S _allin by the individuality of any individuality constraint Pareto domination, and be kept at set A ₁in, as the first non-bad layer; The individual y of individual x constrain domination, when one of them condition following meets: (i) individual x and y all meet constraint condition and (ii) individual x meets constraint condition, and y does not meet constraint condition; (iii) individual x and y does not all meet constraint condition, and the degree that individual x violates constraint condition is less than the degree that individual y violates constraint condition; represent that individual x arranges individual y, and if only if $\&ForAll;i\&Element;\mathrm{1,2},...,M,f_i\left(x\right)\&le;f_i\left(y\right)\&cap;\&Exists;j\&Element;\mathrm{1,2},...,M,f_j\left(x\right)<f_j\left(y\right),$ M represents objective function number; Namely individual x is not worse than individual y on all objective functions, and x is at least better than y on an objective function;

C () is from S _allmiddle removing is all in set A ₁in individuality, residue synthesis population be designated as S _alla ₁, repeat (a), find out population S _alla ₁in by the individuality of any individuality constraint Pareto domination, and be kept at set A ₂in, as the second non-bad layer;

(d) the like, until whole population is layered complete.

3. the multi-objective optimization design of power method from net mixing renewable energy system according to claim 2, is characterized in that in (4) step by step of described second step, described crowding distance can be seen as individual x intuitively _icomprise individual x around _ibut do not comprise the minimum rectangle of other individualities; Crowding distance is less, illustrates individual denser around; Its circular is as follows:

A () is for each objective function f _m, the individuality in population is sorted;

B () namely has minimum f for border individuality _mthe individuality of value, definition crowding distance dist is infinitely great;

Flash trimming other individual x out-of-bounds in (c) same non-bad layer _icrowding distance be

$\mathrm{dist}\left(i\right)=\underset{m=1}{\overset{2}{\mathrm{\&Sigma;}}}\frac{f_m^{i-1}-f_m^{i+1}}{f_m^{\max }-f_m^{\min }}---\left(19\right)$

Wherein with represent objective function f in current population _mmaximum (little) value; with i-th-1 and i+1 individual target function value.

4. the multi-objective optimization design of power method from net mixing renewable energy system according to claim 3, it is characterized in that in (4) step by step of described second step, described according to the non-bad layer residing for individuality and crowding distance, all individualities are sorted, specifically refers to: the individuality that (i) is in the i-th non-bad layer is better than the individuality of jth (j > i) non-bad layer; (ii) be in the individuality of same non-bad layer, the individuality that crowding distance is large is more excellent.