CN106327011A - Micro-power-grid programming and designing method based on dynamic adaptive particle swarm algorithm - Google Patents

Abstract

A micro-power-grid programming and designing method based on a dynamic adaptive particle swarm algorithm comprises the following steps of (1), performing modeling on the micro-power-grid, each micro power supply and a controllable load thereof; 2), determining an optimizing configuration objective function of the micro-power-grid; 3), determining an optimizing configuration constraint condition of the micro-power-grid; 4), performing a control strategy; and 5), performing particle swarm searching optimization. According to the micro-power-grid programming and designing method based on the dynamic adaptive particle swarm algorithm, excitation type requirement side response measures and direct load controlling are combined in the micro-power-grid for improving IRP, and load recombination is performed through direct load controlling, thereby realizing a peak value reduction effect.

Description

A kind of micro-capacitance sensor planning and designing method based on dynamic self-adapting particle cluster algorithm

Technical field

A kind of micro-capacitance sensor planning and designing method based on dynamic self-adapting particle cluster algorithm of the present invention, relates to micro-capacitance sensor planning Technical field.

Background technology

Micro-capacitance sensor planning is the basis that micro-capacitance sensor is built, and the height of its planning level not only directly influences micro-capacitance sensor self Safety, reliability and economy, and being continuously increased along with micro-capacitance sensor permeability, also the operation of bulk power grid can be produced Profound influence.Therefore, micro-capacitance sensor planning design work is the highly important link of micro-capacitance sensor early construction.Owing to microgrid is local Region energy supplying system, therefore its optimization design problem key factor is capacity configuration problem.

In order to make the operation of micro-capacitance sensor more stability and high efficiency, the planning of its early stage is inseparable.We are usual It is according to local distributed power source and load condition, using economy, power supply reliability, Environmental etc. as object function, root Capacity and the position of distributed power source (DG) is determined according to experience.But load side is but the most too much considered, according only to It is modeled distributing rationally to supply side to the result of load prediction or measurement, does not relate to the need being likely encountered when running Side is asked to respond (demand response, DR).Additionally, on optimization method, existing relate to software, in HOMER and HOGA, Processing for all simplification of this problem, other a lot of optimized algorithms may have a convergence rate the lowest.But, too early convergence Or it is absorbed in local optimum and all can produce negative impact.At present, direct load is controlled seldom, unless generated electricity in peak period In the case of amount is inadequate, peak load may be lowered past Demand Side Response.Therefore, a reliable optimized algorithm is selected Often particular importance.

Summary of the invention

For the deficiencies in the prior art, the present invention provides a kind of micro-capacitance sensor based on dynamic self-adapting particle cluster algorithm to plan Method for designing, combines stimulable type Demand Side Response measure (direct load control) in micro-capacitance sensor and improves integral resource planning (Integrated Resources Planning,IRP).Control load is recombinated by direct load thus reach peak The effect that value reduces.Integral resource planning refer to supply of electric power side become with the various forms of resource comprehensives of Demand-side one whole Body carries out power planning, by efficient, economical, reasonably utilize supply and demand side resource potential, before keeping energy services level Putting, the overall society cost making whole planning system is minimum.

The technical solution adopted in the present invention is:

A kind of micro-capacitance sensor planning and designing method based on dynamic self-adapting particle cluster algorithm, comprises the steps:

Step 1: micro-capacitance sensor, each micro battery and controllable burden thereof are modeled, determine whole micro-capacitance sensor model and Each micro battery and the model of controllable burden, wherein the model of controllable burden is:

P (t)=P_load(t)-P_{C_load}(t) (7)

In formula: P (t) is again the load after t direct load controls, Pload (t) is bearing in t load prediction Lotus, PC_load (t) is the load reduced due to direct load control in t；

Step 2: determine the object function that micro-capacitance sensor is distributed rationally:

Step 2.1: this object function be the investment of micro-capacitance sensor and operating cost minimum, be embodied as:

$C_{total}=\underset{i=1}{\overset{4}{\Σ}}{C}_i+C_{DLC}---\left(8\right)$

In formula: i represents photovoltaic cell, blower fan, accumulator and diesel-driven generator, Ctotal represents totle drilling cost；Ci represents every Plant the totle drilling cost of micro battery；CDLC represents the cost that direct load controls,

Wherein Ci can be expressed as following formula:

$C_i=N_i(e_i{P}_i\frac{r{(1+r)}^m}{{(1+r)}^m-1}+u(P_i\left)\right)---\left(9\right)$

In formula: Ni represents the quantity of i-th kind of micro battery；Ei represents the unit price of i-th kind of micro battery；Pi represents i-th kind of micro-electricity The output in source；M represents its depreciation life-span；R represents discount rate；U represents its operating cost；

Step 3: determine that micro-capacitance sensor distributes constraints rationally:

1), power-balance condition:

P_PV(t)+P_WT(t)+P_DG(t)+P_ESS(t)+P_Cload(t)=P_load(t) (10)

In formula: PPV (t), PWT (t), PDG (t), PESS (t) represent photovoltaic cell, blower fan, diesel-driven generator, storage respectively Battery is at the power of t；PCload (t) represents at the load that t is cut off；Pload (t) represents the merit at t load Rate.

2), micro battery capacity constraints, including controllable burden capacity:

X_k≤X_kmax (11)

In formula: Xk is the capacity of kth micro battery, Xkmax is the capability value that kth micro battery is maximum allowable；

3), storage battery charge state constraint:

SOC_min≤SOC(t)≤SOC_max (12)

In formula: SOC (t) is the accumulator state-of-charge in t, and SOCmin, SOCmax represent state-of-charge respectively Maximum and minima；

Step 4: control strategy:

As a example by 24 hours every days, during 0-16 every day, load is at a low ebb and the stage of stable development；T is in 16-24 moment, load Be in peak period, due to accumulator can not the performance of frequent discharge and recharge, so in the 0-16 moment, and photovoltaic cell is exerted oneself and blower fan When exerting oneself more than load, i.e. PPV+PWT ＞ Pload, and during the state-of-charge SOC ＜ 80% of accumulator, just allow distributed electrical Source charges a battery；Due to the unstability of distributed power source, each moment is in change, so the charging merit of accumulator Rate is:

P_charge(t)=min{P₁,P₂,P₃} (13)

P₁=P_PV+P_WT-P_load

P₂=0.9*P_rate (14)

P₃=(SOC_max*E_rate-E_t-1)/Δt

Wherein Pcharge (t) is the accumulator charge power in t；P1, P2, P3 are that accumulator is possible in t Charge power, Prate is the rated power of accumulator；Erate is the rated capacity of accumulator；

As PPV+PWT ＜ Pload, accumulator is not charged and is not discharged, diesel-driven generator meets workload demand, this Time Pcharge (t)=0, when accumulator is fully charged, will not recharge；

When the 16-24 moment, load has welcome peak period, regulates and controls controllable burden, as PPV+PWT+PDG ＜ During Pload-PC_load, SOC ＞ 20%, just allowing battery discharging, its discharge power is:

P_dis(t)=min{P₄,P₅,P₆} (15)

P₄=P_load-P_{C_load}-(P_PV+P_WT+P_DG)

P₅=P_rate

P₆=(E_t-1-SOC_min*E_rate)/Δt (16)

Wherein Pdis (t) is the accumulator discharge power in t；P4, P5, P6 are the electric discharge that t accumulator is possible Power；Prate is the rated power of accumulator；Erate is the rated capacity of accumulator；

As PPV+PWT+PDG ＞ Pload-PC_load, accumulator is not charged and is not discharged, diesel-driven generator expires Foot workload demand, now Pdis (t)=0, when accumulator is discharged, will no longer discharge；

Step 5: population optimizing:

Step 5.1: initialize Fe coatings, population density, iterations and restriction；

Step 5.2: speed and the position of each particle are set；

Step 5.3: calculate fitness fitness value, individual optimum pbest, global optimum gbest；

Step 5.4: the speed of the parameter arranged according to each particle, more new particle；

Step 5.5: update the position of each particle；

Step 5.6: calculate the fitness value of particle after updating, and update；

Step 5.7: if not seeking obtaining optimal solution, returns step 5.3；If seeking optimal solution, end loop.

A kind of micro-capacitance sensor planning and designing method based on dynamic self-adapting particle cluster algorithm of the present invention, has the beneficial effect that:

Relative to traditional micro-capacitance sensor planing method, Demand-side is processed into one by the present invention can be actively engaged in microgrid rule The power supply (equivalent action that load reduces) drawn and run, and it is not only the load passively accepting electric energy.Demand Side Response Measure modeling incorporates in the integral resource planning model of microgrid, Demand-side resource is put on an equal footing with supply side resource, reaches to cut Subtract peak load and realize the purpose that micro-grid system overall society cost is optimum.In order to realize above-mentioned target, use dynamic self-adapting population Algorithm, its fast convergence rate, convergence is good, and convergence precision is high.

Accompanying drawing explanation

Fig. 1 is the particle swarm optimization algorithm flow chart of the present invention.

Fig. 2 is the control strategy figure of the present invention.

Fig. 3 is tradition micro-capacitance sensor plan optimization cost curve figure.

Fig. 4 is that integral resource planning optimizes cost curve chart.

Detailed description of the invention

Owing to parameter is few, it is achieved simple, the excellent feature such as fast convergence rate, particle cluster algorithm (PSO) has become as one Plant most popular optimized algorithm, and be the most successfully applied in all kinds of problems of power system, and be proved to be one Well optimization tool.Wherein the most significant feature of PSO is exactly its fast convergence rate.Without Premature Convergence, from And causing whole population to be absorbed in local optimum and cannot therefrom jump out, its fast convergence rate would is that an extraordinary spy Levy.Therefore, how avoiding being absorbed in local optimum to improve ability of searching optimum is a major issue that must solve.

In PSO, each potential optimal result is seen as motion particle in optimizing space, each grain Sub-i and its speed v i=[vi1, vi2, vid, viD] T and position xi=[xi1, xi2 ..., xid ..., xiD] T association, its Middle D is the dimension in optimizing space.In searching process, the more new formula of its speed and position is expressed as

v_id(t+1)=wv_id(t)+c₁r₁[p_bid-x_id(t)]+c₂r₂[g_bd-x_id(t)] (1)

x_id(t+1)=x_id(t)+v_id(t+1) (2)

In formula, t is current iteration number of times；W is inertia weight；R1 and r2 is generally evenly distributed in the random number in 0-1；Pb and Gb represents individual optimum and global optimum respectively.

Formula (1) and formula (2) show when particle rapidity is 0, and population is stagnated, thus cause the too early convergence of optimum results and And cannot jump out from local optimum, especially prominent in complicated multi-objective optimization question.Therefore, in order to control its convergence Property, the multiformity properly increasing PSO is very important link.Generally, adding mutation operation is exactly that one effectively increases kind The multifarious method of group, thus jump out from local optimum, and improve the ability of searching optimum of PSO.Although mutation operation Highly advantageous for improving population diversity, and can be good at avoiding Premature Convergence, but his randomness limits PSO's Search precision, even results in it and degenerates.

Therefore, a kind of micro-capacitance sensor planning and designing method based on dynamic self-adapting particle cluster algorithm of the present invention, one is proposed TSP question strategy.(less than certain ultimate value) when each particle rapidity is fairly small when, according to subscribe aberration rate with Some particles chosen by machine, then make the particle chosen carry out mutation operation according to formula (3).These variation particles are stored in Then these variation particles are estimated finding the variation particle gb_mut of optimum by Pmut.If its fitness value is little In gb_mut's, then just replace gb.This special mutation operation can be jumped out local optimum and can keep very well by particle The multiformity of population.

x_id=x_id+sign(2(r₃-0.5))βVmax_d (3)

Wherein sign is sign function；R3 is generally evenly distributed in the random number in 0-1；β ∈ [0,1] is the degree of variation； Vmaxd is the maximal rate in d dimension.

Additionally, according to formula (1) it can be seen that inertia weight controls convergence behavior, and make global search and Local Search Balance.Rather than a constant or the amount of linear change, a dynamic inertia weight formula (4) can be to its dynamic tune Whole.

W (t)=w₀+r₄(1-w₀) (4)

Random number during r4 is generally evenly distributed in 0-1 in formula, w0 ∈ [0,0.5] is a constant.Formula (4) makes inertia weigh Focus in w0 to 1 and change, and in global and local is searched for, keep a dynamic equilibrium.

In searching process, c1 changes from 2.5-0.5, and c2 can provide one from 0.5-2.5 change under most of benchmark Individual more optimal result.Therefore, in order to improve the c1 being gradually reduced that ability of searching optimum one represents by expression formula (5) and One c2 being gradually increased represented by expression formula (6) is suggested.

c₁=2.5-t/M_t (5)

c₂=0.5+t/M_t (6)

In formula, Mt refers to maximum iteration time.

Coefficient c1 and c2 is the amount controlling PSO " tension force ", to guide each particle to be respectively facing pi and pg.Such as formula (5) and shown in (6), c1 is set to a bigger value and c2 is set to a less value, in each iteration at the beginning During, its value reduces the most respectively and increases.This mechanism provide not only one in early days global optimizing process just have many The search volume of sample, and the accuracy of optimum results is also can guarantee that in optimizing final stage.

A kind of micro-capacitance sensor planning and designing method based on dynamic self-adapting particle cluster algorithm, comprises the steps:

Step 1: micro-capacitance sensor, each micro battery and controllable burden thereof are modeled.Determine whole micro-capacitance sensor model and Each micro battery (photovoltaic cell, blower fan, accumulator and diesel-driven generator) and the model of controllable burden.The wherein mould of controllable burden Type is:

P (t)=P_load(t)-P_{C_load}(t) (7)

In formula: P (t) is again the load after t direct load controls；Pload (t) is bearing in t load prediction Lotus；PC_load (t) is the load reduced due to direct load control in t.

Step 2: determine the object function that micro-capacitance sensor is distributed rationally.

Step 2.1: this object function be the investment of micro-capacitance sensor and operating cost minimum.

It is embodied as:

$C_{total}=\underset{i=1}{\overset{4}{\Σ}}{C}_i+C_{DLC}---\left(8\right)$

In formula: i represents photovoltaic cell, blower fan, accumulator and diesel-driven generator.Ctotal represents totle drilling cost；Ci represents every Plant the totle drilling cost of micro battery；CDLC represents the cost that direct load controls.

Wherein Ci can be expressed as following formula:

$C_i=N_i(e_i{P}_i\frac{r{(1+r)}^m}{{(1+r)}^m-1}+u(P_i\left)\right)---\left(9\right)$

In formula: Ni represents the quantity of i-th kind of micro battery；Ei represents the unit price of i-th kind of micro battery；Pi represents i-th kind of micro-electricity The output in source；M represents its depreciation life-span；R represents discount rate；U represents its operating cost.

Step 3: determine that micro-capacitance sensor distributes constraints rationally.

1), power-balance condition.

P_PV(t)+P_WT(t)+P_DG(t)+P_ESS(t)+P_Cload(t)=P_load(t) (10)

In formula: PPV (t), PWT (t), PDG (t), PESS (t) represent photovoltaic cell, blower fan, diesel-driven generator, storage respectively Battery is at the power of t；PCload (t) represents at the load that t is cut off；Pload (t) represents the merit at t load Rate.

2), micro battery capacity constraints (including controllable burden capacity).

X_k≤X_kmax (11)

In formula: Xk is the capacity of kth micro battery；Xkmax is the capability value that kth micro battery is maximum allowable.

3), storage battery charge state constraint.

SOC_min≤SOC(t)≤SOC_max (12)

In formula: SOC (t) is the accumulator state-of-charge in t.SOCmin, SOCmax represent state-of-charge respectively Maximum and minima.

Step 4: control strategy.

As a example by 24 hours every days.During 0-16 every day, load is at a low ebb and the stage of stable development；T is in 16-24 moment, load It is in peak period.Due to accumulator can not the performance of frequent discharge and recharge, so in the 0-16 moment, and photovoltaic cell is exerted oneself and blower fan Exert oneself more than (PPV+PWT ＞ Pload) during load, and during the state-of-charge SOC ＜ 80% of accumulator, just allow distributed electrical Source charges a battery.Due to the unstability of distributed power source, each moment is in change, so the charging merit of accumulator Rate is:

P_charge(t)=min{P₁,P₂,P₃} (13)

P₁=P_PV+P_WT-P_load

P₂=0.9*P_rate

P₃=(SOC_max*E_rate-E_t-1)/Δt (14)

Wherein Pcharge (t) is the accumulator charge power in t；P1, P2, P3 are that accumulator is possible in t Charge power.Prate is the rated power of accumulator；Erate is the rated capacity of accumulator.

As PPV+PWT ＜ Pload, accumulator is not charged and is not discharged, diesel-driven generator meets workload demand.This Time Pcharge (t)=0.When accumulator is fully charged, will not recharge.

When the 16-24 moment, load has welcome peak period, regulates and controls controllable burden, as PPV+PWT+PDG ＜ During Pload-PC_load, SOC ＞ 20%, just allow battery discharging.Its discharge power is:

P_dis(t)=min{P₄,P₅,P₆} (15)

P₄=P_load-P_{C_load}-(P_PV+P_WT+P_DG)

P₅=P_rate

P₆=(E_t-1-SOC_min*E_rate)/Δt (16)

Wherein Pdis (t) is the accumulator discharge power in t；P4, P5, P6 are the electric discharge that t accumulator is possible Power；Prate is the rated power of accumulator；Erate is the rated capacity of accumulator.

As PPV+PWT+PDG ＞ Pload-PC_load, accumulator is not charged and is not discharged, diesel-driven generator expires Foot workload demand.Now Pdis (t)=0.When accumulator is discharged, will no longer discharge.

Step 5: population optimizing.

Step 5.1: initialize Fe coatings, population density, iterations and restriction.

Step 5.2: speed and the position of each particle are set.

Step 5.3: calculate fitness fitness value, individual optimum pbest, global optimum gbest.

Step 5.4: the speed of the parameter arranged according to each particle, more new particle.

Step 5.5: update the position of each particle.

Step 5.6: calculate the fitness value of particle after updating, and update.

Step 5.7: if not seeking obtaining optimal solution, returns step 5.3；If seeking optimal solution, end loop.

Application DAPSO solves the distributed power source comprising photovoltaic array, blower fan, accumulator, diesel-driven generator of optimum Installed capacity.Micro-grid operation mode of the present invention is island mode, except when when electric automobile utilizes paddy, electricity price is charged night Connect bulk power grid charging.Typical daily load curve, local renewable energy source data, distributed electrical source dates is all given.Wherein The maximum size of photovoltaic array is 100kW, single-machine capacity 2kW；The fan capacity upper limit is 33kW, single-machine capacity 3kW；Accumulator is held The amount upper limit is 120kW*h, and monoblock battery capacity is 1.2kW*h；Diesel-driven generator maximum size is 210kW, single-machine capacity 21kW； Controllable burden maximum size is 50kW.This micro-grid system is a residential building, and payload is about 221kW；Planning is limited in year 20 years；The price 1.1804 $/L of diesel oil is；The installation cost of controllable burden is 761.9 $/kW；

The parameter of each distributed power source of table 1

Table 2 each scheme program results:

Scheme 1 is tradition micro-capacitance sensor planing method, and scheme 2 is integral resource planning method.Scheme 2 is compared with scheme 1 Relatively, the controllable burden capacity of 50kW is added.Diesel-driven generator number decreases 2.General planning cost reduces 8.83%. It can thus be seen that the direct load of integral resource planning controls to reduce investment and operating cost.

Claims (1)

1. a micro-capacitance sensor planning and designing method based on dynamic self-adapting particle cluster algorithm, it is characterised in that include walking as follows Rapid:

Step 1: be modeled micro-capacitance sensor, each micro battery and controllable burden thereof, determines the model of whole micro-capacitance sensor and each Micro battery and the model of controllable burden, wherein the model of controllable burden is:

P (t)=P_load(t)-P_{C_load}(t) (7)

In formula: P (t) is again the load after t direct load controls, and Pload (t) is the load in t load prediction, PC_load (t) is the load reduced due to direct load control in t；

Step 2: determine the object function that micro-capacitance sensor is distributed rationally:

Step 2.1: this object function be the investment of micro-capacitance sensor and operating cost minimum, be embodied as:

$C_{total}=\underset{i=1}{\overset{4}{\Σ}}{C}_i+C_{DLC}---\left(8\right)$

In formula: i represents photovoltaic cell, blower fan, accumulator and diesel-driven generator, Ctotal represents totle drilling cost；Ci represent every kind micro- The totle drilling cost of power supply；CDLC represents the cost that direct load controls,

Wherein Ci can be expressed as following formula:

$C_i=N_i(e_i{P}_i\frac{r{(1+r)}^m}{{(1+r)}^m-1}+u(P_i\left)\right)---\left(9\right)$

In formula: Ni represents the quantity of i-th kind of micro battery；Ei represents the unit price of i-th kind of micro battery；Pi represents i-th kind of micro battery Output；M represents its depreciation life-span；R represents discount rate；U represents its operating cost；

Step 3: determine that micro-capacitance sensor distributes constraints rationally:

1), power-balance condition:

P_PV(t)+P_WT(t)+P_DG(t)+P_ESS(t)+P_Cload(t)=P_load(t) (10)

In formula: PPV (t), PWT (t), PDG (t), PESS (t) represent photovoltaic cell, blower fan, diesel-driven generator, accumulator respectively Power in t；PCload (t) represents at the load that t is cut off；Pload (t) represents the power at t load,

2), micro battery capacity constraints, including controllable burden capacity:

X_k≤X_kmax (11)

In formula: Xk is the capacity of kth micro battery, Xkmax is the capability value that kth micro battery is maximum allowable；

3), storage battery charge state constraint:

SOC_min≤SOC(t)≤SOC_max (12)

In formula: SOC (t) is the accumulator state-of-charge in t, and SOCmin, SOCmax represent the maximum of state-of-charge respectively Value and minima；

Step 4: control strategy:

As a example by 24 hours every days, during 0-16 every day, load is at a low ebb and the stage of stable development；T is in the 16-24 moment, and load is in Peak period, due to accumulator can not the performance of frequent discharge and recharge, so in the 0-16 moment, and photovoltaic cell is exerted oneself and blower fan is exerted oneself During more than load, i.e. PPV+PWT ＞ Pload, and during the state-of-charge SOC ＜ 80% of accumulator, just allow distributed power source to give Accumulator is charged；Due to the unstability of distributed power source, each moment is in change, so the charge power of accumulator For:

P_charge(t)=min{P₁,P₂,P₃} (13)

$\begin{array}{c}{P}_1=P_{PV}+P_{WT}-P_{load}\\ P_2=0.9*P_{rate}\\ P_3=({\mathrm{SOD}}_{\max }*E_{rate}-E_{t-1})/\mathrm{\Δ}t\end{array}---\left(14\right)$

Wherein Pcharge (t) is the accumulator charge power in t；P1, P2, P3 are accumulator in the possible charging of t Power, Prate is the rated power of accumulator；Erate is the rated capacity of accumulator；

As PPV+PWT ＜ Pload, accumulator is not charged and is not discharged, diesel-driven generator meets workload demand, now Pcharge (t)=0, when accumulator is fully charged, will not recharge；

When the 16-24 moment, load has welcome peak period, regulates and controls controllable burden, as PPV+PWT+PDG ＜ Pload- During PC_load, SOC ＞ 20%, just allowing battery discharging, its discharge power is:

$\begin{array}{c}{P}_{dis}\left(t\right)=min\{P_4,P_5,P_6\}\\ P_4=P_{load}-P_{C\_load}-(P_{PV}+P_{WT}+P_{DG})\end{array}---\left(15\right)$

P₅=P_rate

P₆=(E_t-1-SOC_min*E_rate)/△t (16)

Wherein Pdis (t) is the accumulator discharge power in t；P4, P5, P6 are the discharge power that t accumulator is possible； Prate is the rated power of accumulator；Erate is the rated capacity of accumulator；

As PPV+PWT+PDG ＞ Pload-PC_load, accumulator is not charged and is not discharged, diesel-driven generator meet negative Lotus demand, now Pdis (t)=0, when accumulator is discharged, will no longer discharge；

Step 5: population optimizing:

Step 5.1: initialize Fe coatings, population density, iterations and restriction；

Step 5.2: speed and the position of each particle are set；

Step 5.3: calculate fitness fitness value, individual optimum pbest, global optimum gbest；

Step 5.4: the speed of the parameter arranged according to each particle, more new particle；

Step 5.5: update the position of each particle；

Step 5.6: calculate the fitness value of particle after updating, and update；

Step 5.7: if not seeking obtaining optimal solution, returns step 5.3；If seeking optimal solution, end loop.