Content of the invention
Present invention aims to the deficiencies in the prior art, a kind of scene is provided to store complementary independent micro-capacitance sensor higher-dimension many
Objective optimization collocation method.
The purpose of the present invention is achieved through the following technical solutions:It is many that a kind of scene stores complementary independent micro-capacitance sensor higher-dimension
Objective optimization collocation method, the method comprises the following steps:
(1) read the scene with 1 hour as step-length and store the annual meteorological data that complementary independent micro-grid system implements area
(including wind speed, intensity of illumination, ambient temperature etc.), scene store the complementary each component parameter information of independent micro-grid system and load number
According to (including hourly average DC load and hourly average AC load etc.);Produce reference point, using NBI (Normal-
Boundary intersection) method H reference point of generation, Population Size NP=H is determined according to the reference point number of generation
If (H is even number), NP=H+1 (if H is odd number);
(2) initialize, generate one at random and be uniformly distributed the initial population P={ P that Population Size is NPi, i=1,2 ...,
NP }, wherein i-th individual Pi=(NWGi,NPVi,NBATi), NWGiFor wind-driven generator, number of units, N are installedPViFor photovoltaic battery module
Block number, N are installedBATiFor secondary battery unit group number;
(3) to each of population P individuality Pi, i=1,2 ..., NP, carry out non-uniform mutation, multiple objective function assessment
The multiple-objection optimizations such as calculating, non-dominated ranking operate, and specifically include following sub-step:
(3.1). to Pi={ Pi(j), j=1,2,3 } in each constituent element execute many non-uniform mutation (Multi-non- one by one
Uniform mutation, MNUM), keep other constituent elements constant simultaneously, obtain new individual Pij, j=1,2,3;
Wherein r, r1It is the random number producing in the range of [0,1], t represents current iteration number of times, L (j) represents j-th optimization
The lower limit of variable, U (j) represents the upper limit of j-th optimized variable, and b is the MNUM coefficient of variation, ImaxThe maximum setting for user changes
Generation number;
(3.2) calculate PijCorresponding multiple objective function value, including etc. year value investment cost ACS (Pij), energy surplus multiplying power
Bexc(Pij) and load dead electricity rate LPSP(Pij);
(3.3) to current 3 son individuals PijCarry out non-dominated ranking, thus obtaining its dominated Sorting number rij∈[0,2],j
=1,2,3, the individual record that dominated Sorting number is 0 is Pi0, and by Pi0And corresponding multiple objective function value achieves;
(4) by Pi0As new population at individual, thus producing new population PN={ Pi0, i=1,2 ..., NP };
(5) unconditionally accept P=PN;
(6) repeat step 3-5 is until meeting the maximum iteration time of user's setting;
(7) output Pareto optimal solution and corresponding ACS, energy surplus multiplying power Bexc, load dead electricity rate LPSPEvaluation refers to
Scale value, provides the user scene and stores complementary independent micro-capacitance sensor configuration scheme.
Further, in step (3.2), higher-dimension multiple objective function can set according to the actual requirements, has certain motility
And different effect of optimizations can be reached, the year such as general selecting system is worth investment cost ACS, energy surplus multiplying power Bexc, load short of electricity
Probability LPSPAs object function, that is,:
LmaxFor independent micro-grid system patient maximum load short of electricity probit, BmaxFor patient ceiling capacity mistake
Surplus multiplier value.A the years such as () system are worth investment cost ACS calculating process as follows:
ACS (X)=(C11NWG+C12NPV+C13NBAT)+(C21NWG+C22NPV+C23NBAT)+C3NBAT(16)
In formula:X is optimized variable set, X=(NWG,NPV,NBAT);C11、C12、C13It is respectively blower fan, photovoltaic and accumulator
Each assembly installation cost Average Annual Cost;C21、C22、C23It is respectively blower fan, photovoltaic becomes with accumulator each unit year operation maintenance
This;C3For the average annual replacement cost of accumulator.
Blower fan, photovoltaic and accumulator each assembly installation cost Average Annual Cost are related to the assembly life-span cycle time limit, its pass
It is that expression formula is:
C1i=CPi.CRFi(h,Yproj) (17)
In formula:CPFor installation cost;YprojFor the assembly life-span time limit;CRF is recovery of the capital coefficient (capital
Recovery factor, CRF), its expression formula is:
Wherein, h is discount rate.
Each assembly operation and maintenance cost C of 1 year2iN () is calculated as follows:
C2i(n)=C2i(1).(1+f)n(19)
Wherein C2i(1) it is the operating cost of the 1st year, f represents annual inflation.
In the project time limit, if system component reaches its end-of-life time limit, need assembly to be carried out reset to replace, group
The replacement expense of part is calculated as follows:
C3=Cr.SFE(h,Yr) (20)
In formula:CrFor the replacement cost;YrReset the life-span for assembly;SFE is the compensation fund factor, is counted according to formula (21)
Calculate:
(b) energy surplus multiplying power BexcThe energy wasting by (being usually arranged as 1 year) within the specific period being considered be
The ratio of system load aggregate demand energy, is specifically calculated as follows:
In formula:Pe(t) system excess power;Pl(t) system total load power;T is power supply total time hop count, usually T=
8760;Δ t is simulation step length, usually Δ t=1;
Pl(t)=PD(t)+PA(t)/en(23)
Wherein, PDT () represents total DC load, PAT () represents total AC load, enRepresent inverter efficiency.
C () is to load short of electricity probability LPSPAs reliability evaluation index, represent system short of electricity time and total power-on time
Ratio, is calculated as follows:
In formula:SlossT () is system short of electricity marker character, its value is that 1 expression system short of electricity (is provided that in t system
General power be less than system load demand), its value can meet all workload demands for 0 expression system.、
The invention has the beneficial effects as follows:With wind-driven generator, photovoltaic array output mathematical model and accumulator charge and discharge
Based on electrical characteristics and life cycle mathematical model, consider and wait year value investment cost (to include equipment investment expense, operation is tieed up
Shield expense and equipment replacement expense etc.), energy surplus multiplying power and load dead electricity rate (micro-capacitance sensor whole year operation reliability index) etc.
Many Performance Evaluating Indexes, design higher-dimension multiple target Extremal optimization method, as solver, is realized scene and is stored complementary independent micro-capacitance sensor
Higher-dimension multiple-objection optimization configures.The present invention can achieve that scene stores complementary independent micro-capacitance sensor higher-dimension multiple-objection optimization allocative effect, tool
There is the advantages below not available for traditional single object optimization method and traditional Multipurpose Optimal Method:For micro-capacitance sensor design planning person
The configuration scheme providing is more reasonable, the configuration scheme investment in the case of meeting identical power supply reliability index
Less, optimization method is implemented simply, to adjust without complex cost function weight coefficient, and parameters optimization that need not be complicated is adjusted, and
Optimization efficiency is higher.
Specific embodiment
The present invention is further described below in conjunction with the accompanying drawings, and the purpose of the present invention and effect will be apparent from.
Fig. 1 is independent micro-grid system structure chart, including dual-feed asynchronous wind power generator, photovoltaic array, lead-acid accumulator,
Dc bus, AC/DC changer, DC/DC changer, DC/AC changer, DC load equipment and AC load equipment etc..
Fig. 2 is that a kind of scene proposed by the present invention stores complementary independent micro-capacitance sensor higher-dimension multiple-objection optimization collocation method total figure.
Store as a example complementary independent micro-grid system optimization design engineering by Wenzhou City somewhere 150kW scene, using proposed by the present invention
Light stores complementary independent micro-capacitance sensor higher-dimension multiple-objection optimization collocation method and is designed implementing.
Described scene stores complementary independent micro-capacitance sensor higher-dimension multiple target extremal optimization collocation method, comprises the following steps:
(1) read the Wenzhou City industrial occupancy year meteorological data with 1 hour as step-length and (include wind speed, intensity of illumination, ring
Border temperature etc.), scene store the complementary each component parameter information of independent micro-grid system (system construction drawing is as shown in Figure 1) and Wenzhou City
Certain industrial occupancy load data load data (includes hourly average DC load and hourly average AC load etc.);Produce reference
Point, produces H reference point using NBI (Normal-boundary intersection) method, according to the reference point producing
Number determines Population Size NP=H in higher-dimension multiple target extremal optimization collocation method (if H is even number), and NP=H+1 is (if H is strange
Number);
(2) initialize, generate one at random and be uniformly distributed the initial population P={ P that Population Size is NPi, i=1,2 ...,
NP }, wherein i-th individual Pi=(NWGi,NPVi,NBATi), NWGiFor wind-driven generator, number of units, N are installedPViFor photovoltaic battery module
Block number, N are installedBATiFor secondary battery unit group number;
(3) to each of population P individuality Pi, i=1,2 ..., NP, carry out non-uniform mutation, multiple objective function assessment
The multiple-objection optimizations such as calculating, non-dominated ranking operate, and specifically include following sub-step:
(3.1) to Pi={ Pi(j), j=1,2,3 } in each constituent element execute many non-uniform mutation (Multi-non- one by one
Uniform mutation, MNUM), keep other constituent elements constant simultaneously, obtain new individual Pij, j=1,2,3;
Wherein r, r1It is the random number producing in the range of [0,1], t represents current iteration number of times, L (j) represents j-th optimization
The lower limit of variable, U (j) represents the upper limit of j-th optimized variable, and b is the MNUM coefficient of variation, ImaxThe maximum setting for user changes
Generation number;
(3.2) calculate PijCorresponding multiple objective function value, including etc. year value investment cost ACS (Pij), energy surplus multiplying power
Bexc(Pij) and load dead electricity rate LPSP(Pij), concrete calculating process stores complementary independent micro-capacitance sensor Multi-target evaluation mould referring to scene
Step described by type computational methods;
(3.3) to current 3 son individuals PijCarry out non-dominated ranking, thus obtaining its dominated Sorting number rij∈[0,2],j
=1,2,3, the individual record that dominated Sorting number is 0 is Pi0, and by Pi0And corresponding multiple objective function value achieves;
(4) by Pi0As new population at individual, thus producing new population PN={ Pi0, i=1,2 ..., NP };
(5) unconditionally accept P=PN;
(6) repeat step 3-5 is until meeting the maximum iteration time of user's setting;
(7) output Pareto optimal solution and corresponding ACS, energy surplus multiplying power Bexc, load dead electricity rate LPSPEvaluation refers to
Scale value, provides the user scene and stores complementary independent micro-capacitance sensor configuration scheme.
Fig. 3 gives the complementary independent micro-capacitance sensor multi-object evaluation model computational methods of scene storage in step (3.2) and specifically flows
Cheng Tu, higher-dimension multiple objective function can set according to the actual requirements, have certain motility and can reach different effect of optimizations,
Year value investment cost ACS such as general selecting system, energy surplus multiplying power Bexc, load short of electricity probability LPSPAs object function, that is,:
LmaxFor independent micro-grid system patient maximum load short of electricity probit, BmaxFor patient ceiling capacity mistake
Surplus multiplier value.A the years such as () system are worth investment cost ACS calculating process as follows:
ACS (X)=(C11NWG+C12NPV+C13NBAT)+(C21NWG+C22NPV+C23NBAT)+C3NBAT(28)
In formula:X is optimized variable set, X=(NWG,NPV,NBAT);C11、C12、C13It is respectively blower fan, photovoltaic and accumulator
Each assembly installation cost Average Annual Cost;C21、C22、C23It is respectively blower fan, photovoltaic becomes with accumulator each unit year operation maintenance
This;C3For the average annual replacement cost of accumulator.
Blower fan, photovoltaic and accumulator each assembly installation cost Average Annual Cost are related to the assembly life-span cycle time limit, its pass
It is that expression formula is:
C1i=CPi.CRFi(h,Yproj) (29)
In formula:CPFor installation cost;YprojFor the assembly life-span time limit;CRF is recovery of the capital coefficient (capital
Recovery factor, CRF), its expression formula is:
Wherein, h is discount rate.
Each assembly operation and maintenance cost C of 1 year2iN () is calculated as follows:
C2i(n)=C2i(1).(1+f)n(31)
Wherein C2i(1) it is the operating cost of the 1st year, f represents annual inflation.
In the project time limit, if system component reaches its end-of-life time limit, need assembly to be carried out reset to replace, group
The replacement expense of part is calculated as follows:
C3=Cr.SFE(h,Yr) (32)
In formula:CrFor the replacement cost;YrReset the life-span for assembly;SFE is the compensation fund factor, is counted according to formula (33)
Calculate:
(b) energy surplus multiplying power BexcThe energy wasting by (being usually arranged as 1 year) within the specific period being considered be
The ratio of system load aggregate demand energy, is specifically calculated as follows:
In formula:Pe(t) system excess power;Pl(t) system total load power;T is power supply total time hop count, usually T=
8760;Δ t is simulation step length, usually Δ t=1;
Pl(t)=PD(t)+PA(t)/en(35)
Wherein, PDT () represents total DC load, PAT () represents total AC load, enRepresent inverter efficiency.
C () is to load short of electricity probability LPSPAs reliability evaluation index, represent system short of electricity time and total power-on time
Ratio, is calculated as follows:
In formula:SlossT () is system short of electricity marker character, its value is that 1 expression system short of electricity (is provided that in t system
General power be less than system load demand), its value can meet all workload demands for 0 expression system.
Involved dual-feed asynchronous wind power generator, photovoltaic array output mathematical model and lead in step 1 and step 3
Acid accumulator discharge and recharge is calculated as follows with life cycle mathematical model:
1) wind driven generator output power model
It is possible to the characteristic passed through between the output of wind power generating set and wind speed is bent after the distribution of known wind speed
Line obtains the average output power of blower fan system:
In formula:PrFor rated output power;The wind speed that v highly locates for wind turbine hub, vrFor rated wind speed, vinFor incision
Wind speed;voutFor the wind speed cutting out.Wind speed has the randomness of height, and here adopts two-parameter Weibull (Weibull) model
Produce the air speed data obtaining simulation, its probability density function expression formula is:
In formula:2 parameters that k and c is distributed for Weibull, k is referred to as form parameter, k>0, c is scale parameter, c>1.
2) photovoltaic array output calculates model
Quantity is NPVPhotovoltaic array output power model PPVIt is calculated as follows:
Wherein, G, GmaxIt is respectively actual intensity of illumination within a certain period of time and maximum intensity of illumination;Pp(G) it is single-piece
The output of photovoltaic module,For intensity of illumination probability density function, it is respectively calculated as follows:
Wherein, PSTCRepresent the full test power under standard test condition, G is actual intensity of illumination;K is power temperature
Degree coefficient;TaFor ambient temperature, TNOCFor assembly operating temperature ratings.
Wherein, the form parameter that α, β are distributed for Beta.
3) battery model
State-of-charge (SOC), terminal voltage and life cycle are several important parameters of battery management, its computation model
As follows:
Storage battery charge state SOC be reaction accumulator dump energy account for its total capacity ratio parameter, in front and back two when
Between quarter, accumulator SOC is represented by following relational expression:
SOC (t+1)=SOC (t) (1- δ (t))+IBAT(t).Δt.η(t)/CBAT(43)
In formula:IBATT () is t charging and discharging currents, be that just electric discharge is to be negative during charging;PBATT () is filling of accumulator
Discharge power, is just charged as, it is negative for discharging;VBATT () is accumulator voltage;δ is nature discharge rate;Δ t is two moment in front and back
Time interval;CBATFor accumulator capacity on time;η (t) is efficiency for charge-discharge, and during electric discharge, its value is 1, with SOC and filling during charging
Electricity is current related, and computation model is as follows:
The terminal voltage of accumulator can be by its open-circuit voltage and the internal resistance voltage drop meter producing in its internal resistance because of charging and discharging currents
Show, computing formula is as follows:
VBAT(t)=Eoc(t)+IBAT(t)RBAT(t) (46)
Eoc(t)=VF+b.log (SOC (t)) (47)
RBAT(t)=Relectrode(t)+Relectrlyte(t) (48)
Relectrode(t)=r1+r2.SOC(t) (49)
Relectrolyte(t)=[r3+r4.SOC(t)]-1(50)
In formula:EocT () is the open-circuit voltage of accumulator;IBAT(t) be battery current (its value be more than 0 represent charge, its
Value represents electric discharge less than 0);RBATT () is accumulator internal resistance, including bath resistance Relectrode(t) and bath resistance
Relectrolyte(t) two parts;b,r1,r2,r3,r4For empirical coefficient, its value has different values in both charge and discharge modes:
The life of storage battery damages phase TBATIt is calculated as follows:
nS=T/ Δ t represents the total fixed number of emulation, SOC1,t、SOC2,tIt is respectively t-th emulation period beginning and end
When SOC numerical value, SBAT,tRepresent the charging and discharging state of t-th emulation period accumulator reality,
DOD,tRepresent the battery discharging depth detecting at the end of t-th emulation period, DOD,t=1-SOC2,t.
The reset cycle Y of accumulatorBIt is calculated as follows:
TBR=min { TBAT,TFL} (53)
Wherein, TFLRepresent the float life reference value that storage battery production producer provides.
In system operation, accumulator is limited scope (SOC by its state-of-chargemin≤SOC≤SOCmax) and electric power storage
The impact of pond technical limitations itself, its maximum charge-discharge electric power is calculated as follows:
Pcm(t)=NBAT.max{0,min{(SOCmax-SOC(t)).CBAT/Δt,Icm}.VBAT(t)} (54)
Pdm(t)=NBAT.max{0,min{(SOC(t)-SOCmin).CBAT/Δt,Idm}.VBAT(t)} (55)
In formula:SOCmax,SOCminIt is respectively the bound of storage battery charge state;CBATFor accumulator capacity;VBATT () is
Accumulator voltage;Δ t is unit simulation time interval;Pcm(t),PdmT () is respectively accumulator within t emulation period
Maximum chargeable electric current and maximum discharge current;Icm,IdmIt is respectively maximum charging current and the maximum electric discharge electricity that accumulator allows
Stream, in the unit interval, maximum charging and discharging currents are the 20% of the specified ampere-hour capacity of accumulator, i.e.
Icm=Idm=0.2CBAT/Δt (56)
Complementary independent micro-grid system is stored to Wenzhou Area 150kW scene using the present invention and is optimized design, result table
Bright:The present invention can achieve that scene stores complementary independent micro-capacitance sensor higher-dimension multiple-objection optimization allocative effect, and meeting, identical power supply is reliable
Property index in the case of configuration scheme output investment ratio tradition single object optimization method (as genetic algorithm, particle cluster algorithm) and
Traditional Multipurpose Optimal Method (as NSGA-II algorithm) at least saves more than 20%, and the present invention only has the MNUM coefficient of variation
Need to adjust with two parameters of iteration optimization number of times, adjust without complex cost function weight coefficient, optimization method is also without kind
The operation links such as flock-mate fork, implement more simple, the calculating time of whole method is shorter.
In sum, can achieve that scene stores complementary independent micro-capacitance sensor higher-dimension multiple-objection optimization allocative effect using the present invention,
There is the advantages below not available for traditional single object optimization method and traditional Multipurpose Optimal Method:For micro-capacitance sensor design planning
The configuration scheme that person provides is more reasonable, and the configuration scheme in the case of meeting identical power supply reliability index is thrown
Money is less, and optimization method is implemented simply, to adjust without complex cost function weight coefficient, and parameters optimization that need not be complicated is adjusted,
And optimization efficiency is higher.