CN105205245B - Direct-driven permanent-magnetic wind power generator multi-work-condition global efficiency optimum design method - Google Patents

Abstract

The invention discloses a direct-driven permanent-magnetic wind power generator multi-work-condition global efficiency optimum design method, which comprises the following steps of determining a system design parameter of a direct-driven permanent-magnetic wind power generator; building a design mathematical model by using the system constraint of a current converter and the control algorithm constraint as the constraint conditions and using the highest weighted average efficiency or smallest weighted average power loss of a plurality of work condition points as a target function on the basis of the system design parameter; setting a group of initial work condition values by aiming at the system design parameter; on the basis of the initial work condition values, using a simulated annealing algorithm for performing multi-work-condition iterative solving on the design mathematical model to obtain a plurality of groups of resolving results; selecting a group of optimum results from all solving results to be used as the optimum system design parameter obtained through design to be output. The method has the advantage that the multi-work-condition efficient operation of the direct-driven permanent-magnetic wind power generator can be realized under the conditions of variable speed, variable frequency, change of stator voltage along with the load and rotating speed change, and influence of operation characteristics by a current converter and a control strategy.

Description

A kind of direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method

Technical field

The present invention relates to the designing technique of direct-drive permanent magnet wind power generator is and in particular to a kind of direct-drive permanent magnet wind power generator Multi-state global efficiency optimal-design method.

Background technology

The method that is typically designed of direct-drive permanent magnet wind power generator is on traditional synchronous generator method for designing basis at present The distinctive magnetic circuit design method of upper combination magneto, for single rated condition, is run with voltage constant and constant-speed and constant-frequency Precondition designs, therefore exist can not meet direct-drive permanent magnet wind power generator speed changing, frequency converting run, stator voltage with Load and rotation speed change and change, the shortcomings of operation characteristic is affected by current transformer and control strategy.

Content of the invention

The technical problem to be solved in the present invention：For the problems referred to above of prior art, one kind is provided to be capable of directly driving forever Magnetism type wind driven generator is in speed changing, frequency converting, stator voltage changes with load and rotation speed change, operation characteristic is subject to current transformer and control The direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method of multi-state Effec-tive Function in the case of strategy impact.

In order to solve above-mentioned technical problem, the technical solution used in the present invention is：

A kind of direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method, step includes：

1) system design parameterses of direct-drive permanent magnet wind power generator are determined；

2) constrained with the current/voltage of current transformer and control algolithm is constrained to constraints, is put down with the weighting of multiple operating points All efficiency highest or the minimum object function of weighted average power loss, set up design mathematic mould based on described system design parameterses Type；

3) one group of initial operating mode value of system design parameterses is obtained based on rated point design, based on described initial operating mode value, Multi-state iterative is carried out to described design mathematic model using simulated annealing, obtains multigroup solving result；

4) select the optimal system design parameter output that one group of optimal result obtains as design from all solving results.

Preferably, described step 1) in the system design parameterses of direct-drive permanent magnet wind power generator that determine include stator core Internal diameter D_i1, stator core length L_s, close amplitude B of stator yoke magnetic_ysm, the high h of stator slot_s, stator groove width b_s, permanent magnet height h_m, forever Magnet width b_m, to pole span τ_p, MgO-ZrO_2 brick q.

Preferably, described step 2) in current/voltage constrain as shown in formula (1)；

$\left\{\begin{array}{c}u\≤U_{\lim }\\ I\≤I_{\lim }\end{array}\right.---\left(1\right)$

In formula (1), u is current transformer voltage, i is current transformer electric current, U_limFor voltage binding occurrence, I_limFor restriction of current value.

Preferably, described step 2) in control algolithm constraint as shown in formula (2) three constraint in any one；

$\left\{\begin{array}{c}i\→\min \\ u=u_c\\ Q_s=0\end{array}\right.---\left(2\right)$

In formula (2), i is current transformer electric current, and → min represents optimal current control, and u is current transformer voltage, u_cFor constant electricity Pressure binding occurrence, Q_sFor unit power factor.

Preferably, described step 2) in object function such as formula (3) or formula (4) shown in；

$\stackrel{\‾}{\mathrm{\η}}=\frac{{\mathrm{\Σ\η}}_i\left(x\right)W_i}{{\mathrm{\ΣW}}_i}\→max---\left(3\right)$

In formula (3),For the weighted average efficiency of all operating points, η_iX () is the average efficiency of i-th operating point, i is big In equal to 3, W_iFor the weight factor of i-th operating point, → max represents weighted average efficiency highest constraints；

$\stackrel{\‾}{p}=\frac{{\mathrm{\Σp}}_i\left(x\right)W_i}{{\mathrm{\ΣW}}_i}\→min---\left(4\right)$

In formula (4),For the weighted average power loss of all operating points, p_iX () is the power loss of i-th operating point, I is more than or equal to 3, W_iFor the weight factor of i-th operating point, → min represents the minimum constraints of weighted average power loss.

Preferably, described step 2) shown in the design mathematic model such as formula (5) set up；

$\left\{\begin{array}{c}\underset{X\∈D}{max}:{\mathrm{\η}}_i\left(X\right)/\underset{X\∈D}{min}:p_i\left(X\right)\\ D=\left\{X\right|g_i\left(X\right)\≤0,i=1,2,...,m\}\\ X={\[D_{i1},L_s,B_{ysm},h_s,b_s,h_m,b_m,{\mathrm{\τ}}_p,q\]}^T\end{array}\right.---\left(5\right)$

In formula (5), D is the total quantity of system design parameterses to be asked, η_i(X) be i-th operating point average efficiency, p_i(X) For the power loss of i-th operating point, i is more than or equal to 3,For i-th operating point average efficiency Big or i-th operating point power loss is minimum, g_i(X) it is constraints, m is the quantity of constraints, X is system to be asked The matrix that design parameter is constituted, D_i1For frame bore, L_sFor stator core length, B_ysmFor the close amplitude of stator yoke magnetic, h_sFor Stator slot is high, b_sFor stator groove width, h_mFor permanent magnet height, b_mFor permanent magnet width, τ_pIt is to be every extremely every phase groove to pole span, q Number.

Preferably, described step 3) in detailed step as follows：

3.1) one group of initial operating mode value of system design parameterses is obtained based on rated point design；

3.2) it is based on described initial operating mode value, using simulated annealing, one is randomly generated to described design mathematic model Initial solution x₀, calculate initial solution x₀Target function valueInitialization iterationses L_k, initialize the first of simulated annealing Beginning temperature T₀, temperature change number of times k and tempering before accept solution number of times m；Redirect execution step 3.3) it is iterated；

3.3) random disturbance is made to current optimal solution, generate the new explanation of current iteration by function expression shown in formula (6)；

$\left\{\begin{array}{c}{x^j}_{k+1}=x_k^j+y^j(B_j-A_j)\\ y^j=\mathrm{sgn}(u^j-\frac{1}{2})T_k\[{(1+1/T_k)}^{\left|2u^j-1\right|}-1\]\\ T_k=T_0\exp (-{\mathrm{ck}}^{1/D})\end{array}\right.---\left(6\right)$

In formula (6), x^j _k+1For the new explanation of current iteration, x^j _kFor the new explanation of last iteration, y^jSet for j-th system to be asked The disturbance of meter parameter, 1≤j≤D, D are the total quantity of system design parameterses to be asked, B^jCan for j-th system design parameters to be asked The maximum that can take, A^jThe minima that may take for j-th system design parameters to be asked, u^jIt is between the random value between 0 and 1, T_kFor the annealing temperature of current iteration, T₀For the initial temperature of simulated annealing, c is the temperature control system of simulated annealing Number, k is current temperature change time numerical value；

3.4) calculate the new explanation x of current iteration^j _k+1Corresponding target function valueCalculating target function value Relatively last target function valueBetween target function value increment

3.5) judge target function value incrementWhether setting up less than 0, if set up, accepting the new optimal solution producing For current optimal solution；If be false, to judge whether that with specified probability P the optimal solution accepting newly to produce is currently optimum Solution；If accepting the new optimal solution producing is current optimal solution, before updating tempering, accept solution number of times m；

3.6) accepting before judging to update tempering whether solution number of times m reaches predetermined threshold value, if reaching predetermined threshold value, carrying out The temper of simulated annealing, and reset acceptance solution number of times m before tempering after temper finishes；

3.7) update iterationses L_kValue, judge new iterationses L_kWhether reach predetermined threshold value, if not yet reached Predetermined threshold value then redirects execution step 3.3), otherwise redirect execution step 3.8)；

3.8) if the continuous N number of new explanation of specified quantity is not all accepted or the cost function value of simulated annealing is less than Set-point between one 0 to 1, then judge surely to reach the end condition of simulated annealing, redirect execution step 4)；Otherwise, Reset iterationses L_kIf annealing process slowly, is simulated the Quenching Treatment of annealing algorithm, and redirects execution step 3.3).

Preferably, described step 3.4) in judge whether that the optimal solution accepting newly to produce is current with specified probability P Optimal solution specifically refers to：Randomly generate an equally distributed random number ζ on [0,1] interval first, then always with formula (7) institute Show Metropolis criterion to calculate probability P, if probability P is more than random number ζ, accepting the new optimal solution producing is currently Excellent solution, the optimal solution otherwise abandoning accepting newly to produce is current optimal solution；

$P=\exp \[-\frac{\mathrm{\Δ}\stackrel{\‾}{\mathrm{\η}}}{T_k}\]---\left(7\right)$

In formula (7), P is probability to be calculated,For target function value increment, T_kAnnealing temperature for current iteration.

Preferably, described step 3.6) in be simulated the detailed step of temper of annealing algorithm and include：

3.6.1) according to formula (8), finite difference calculus calculates the gradient of each system design parameters in vector x；

$s_j=\left|\frac{\stackrel{\‾}{\mathrm{\η}}(x_{opt}+e_j\mathrm{\δ})-\stackrel{\‾}{\mathrm{\η}}\left(x_{opt}\right)}{\mathrm{\δ}}\right|---\left(8\right)$

In formula (8), s_jRepresent the gradient of j-th system design parameters in vector x,Weighted average for all operating points Efficiency, x_optFor the optimal solution in the current all solutions having found, e_jIt is the vector of D for a dimension, D is system design parameterses Quantity, vectorial e_jIn j-th component be 1, remaining component be 0, δ be step-length；

3.6.2) it is directed to each system design parameters, if the gradient of this system design parameters is equal to default gradient threshold Value s_max, then redirect execution step 3.6.3)；Otherwise exit temper；

3.6.3) annealing temperature T is updated according to formula (9)_kValue with temperature change number of times k；

$\left\{\begin{array}{c}{T}_{k^{\′}}^{\′}=\frac{s_{max}}{s_j}{T}_k\\ k^{\′}={\[ln(T_0/T_{k^{\′}}^{\′})/c\]}^D\end{array}\right.---\left(9\right)$

In formula (9),For the annealing temperature after updating, s_maxDefault Grads threshold, s_jRepresent j-th system in vector x The gradient of design parameter, s_jRepresent the gradient of j-th system design parameters in vector x, T_kFor the annealing temperature before updating；K' is Temperature change number of times after renewal, T₀For the initial temperature of simulated annealing, c is the temperature control system of simulated annealing Number, D is the quantity of system design parameterses.

Preferably, described step 3.8) in Quenching Treatment specifically refer to according to formula (10) update annealing temperature T_kValue；

$T_k=T_0\exp (-{\mathrm{ck}}^{q_j/D})---\left(10\right)$

In formula (10), T_kFor the annealing temperature after updating, T₀For the initial temperature of simulated annealing, c is simulated annealing calculation The temperature control coefficient of method, k is temperature change number of times, q_jFor the quenching factor of simulated annealing, D is system design parameterses Quantity.

Direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method tool of the present invention has the advantage that：The present invention Direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method is with rated point design result as initial condition, with unsteady flow Device voltage x current and control algolithm be constraints, with the weighted average efficiency highest of multiple operating points (or weighted average power Loss reduction) it is target, based on global optimization approach, direct-drive permanent magnet wind power generator multi-state load design problem is asked Solution, thus obtaining one group of optimum system design parameters, is capable of direct-drive permanent magnet wind power generator in speed changing, frequency converting, stator electricity In the case that pressure changes with load and rotation speed change, operation characteristic is affected by current transformer and control strategy, multi-state is efficiently transported OK.

Brief description

Fig. 1 is the basic procedure schematic diagram of present invention method.

Fig. 2 be embodiment of the present invention step 3) schematic flow sheet.

Specific embodiment

As shown in figure 1, the step of the present embodiment direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method Including：

1) system design parameterses of direct-drive permanent magnet wind power generator are determined；

2) constrained with the current/voltage of current transformer and control algolithm is constrained to constraints, (refers to generate electricity with multiple operating points Running parameter under a certain specified conditions for the machine) weighted average efficiency highest or weighted average power loss minimum target letter Number, sets up design mathematic model based on system design parameterses；

3) obtain one group of initial operating mode value of system design parameterses based on rated point design, based on initial operating mode value, adopt Simulated annealing carries out multi-state iterative to design mathematic model, obtains multigroup solving result；

4) select the optimal system design parameter output that one group of optimal result obtains as design from all solving results.

In the present embodiment, step 1) in the system design parameterses of direct-drive permanent magnet wind power generator that determine include stator core Internal diameter D_i1, stator core length L_s, close amplitude B of stator yoke magnetic_ysm, the high h of stator slot_s, stator groove width b_s, permanent magnet height h_m, forever Magnet width b_m, to pole span τ_p, MgO-ZrO_2 brick q.

In the present embodiment, step 2) in current/voltage constrain as shown in formula (1)；

$\left\{\begin{array}{c}u\≤U_{\lim }\\ I\≤I_{\lim }\end{array}\right.---\left(1\right)$

In formula (1), u is current transformer voltage, i is current transformer electric current, U_limFor voltage binding occurrence, I_limFor restriction of current value.

In the present embodiment, step 2) in control algolithm constraint three constraints as shown in formula (2) in any one；

$\left\{\begin{array}{c}i\→\min \\ u=u_c\\ Q_s=0\end{array}\right.---\left(2\right)$

In formula (2), i is current transformer electric current, and → min represents optimal current control, and u is current transformer voltage, u_cFor constant electricity Pressure binding occurrence, Q_sFor unit power factor.

In the present embodiment, step 2) in object function such as formula (3) or formula (4) shown in；

$\stackrel{\‾}{\mathrm{\η}}=\frac{{\mathrm{\Σ\η}}_i\left(x\right)W_i}{{\mathrm{\ΣW}}_i}\→max---\left(3\right)$

In formula (3),For the weighted average efficiency of all operating points, η_iX () is the average efficiency of i-th operating point, i is big In equal to 3, W_iFor the weight factor of i-th operating point, → max represents weighted average efficiency highest constraints；Due to straight Drive permanent magnet wind power generator global efficiency optimal design Problem is a max problem, therefore can adopt formula as needed (3) object function shown in, converts thereof into(i.e.) solved.

$\stackrel{\‾}{p}=\frac{{\mathrm{\Σp}}_i\left(x\right)W_i}{{\mathrm{\ΣW}}_i}\→min---\left(4\right)$

In formula (4),For the weighted average power loss of all operating points, p_iX () is the power loss of i-th operating point, I is more than or equal to 3, W_iFor the weight factor of i-th operating point, → min represents the minimum constraints of weighted average power loss. Certainly it is also possible to adopt object function shown in formula (4) as needed, convert thereof intoSolved.

Direct-drive permanent magnet wind power generator global efficiency optimal design Problem belongs to non-linear, multiple target Solve problems, its solution Not unique.In the present embodiment, multi-state iterative, convergence rate are carried out to design mathematic model using simulated annealing Comparatively fast, and the solving result that finally gives is globally optimal solution.For step 3) initial for one group of system design parameterses setting Operating mode value, based on initial operating mode value, carries out multi-state iterative using simulated annealing to design mathematic model, obtains many The executable portion of group solving result, is written as direct-drive permanent magnet wind power generator energy using Visual C++ in the present embodiment Performance calculation program QE.exe, execution step 3) then call execution QE.exe, can greatly simplify design procedure.

In the present embodiment, step 2) shown in the design mathematic model such as formula (5) set up；

$\left\{\begin{array}{c}\underset{X\∈D}{max}:{\mathrm{\η}}_i\left(X\right)/\underset{X\∈D}{min}:p_i\left(X\right)\\ D=\left\{X\right|g_i\left(X\right)\≤0,i=1,2,...,m\}\\ X={\[D_{i1},L_s,B_{ysm},h_s,b_s,h_m,b_m,{\mathrm{\τ}}_p,q\]}^T\end{array}\right.---\left(5\right)$

In formula (5), D is the total quantity of system design parameterses to be asked, η_i(X) be i-th operating point average efficiency, p_i(X) For the power loss of i-th operating point, i is more than or equal to 3,Average efficiency for i-th operating point Maximum or i-th operating point power loss is minimum, g_i(X) it is constraints, m is the quantity of constraints, X is to wait to ask to be The matrix that system design parameter is constituted, D_i1For frame bore, L_sFor stator core length, B_ysmFor the close amplitude of stator yoke magnetic, h_s For stator slot height, b_sFor stator groove width, h_mFor permanent magnet height, b_mFor permanent magnet width, τ_pIt is to be every extremely every phase groove to pole span, q Number.Design mathematic model shown in formula (5) comprises the mapping relations between system design parameterses and energy characteristics under different operating modes.

As shown in Fig. 2 step 3) in detailed step as follows：

3.1) one group of initial operating mode value of system design parameterses is obtained based on rated point design；

3.2) it is based on initial operating mode value, an initial solution is randomly generated using simulated annealing to design mathematic model x₀, calculate initial solution x₀Target function valueInitialization iterationses L_k, the initial temperature of initialization simulated annealing T₀, temperature change number of times k and tempering before accept solution number of times m；Redirect execution step 3.3) it is iterated；In the present embodiment, simulation Initial temperature T of annealing algorithm₀Initialization value be 1.0, the initialization value of temperature change number of times k is 1；

3.3) random disturbance is made to current optimal solution, generate the new explanation of current iteration by function expression shown in formula (6)；

$\left\{\begin{array}{c}{x^j}_{k+1}=x_k^j+y^j(B_j-A_j)\\ y^j=\mathrm{sgn}(u^j-\frac{1}{2})T_k\[{(1+1/T_k)}^{\left|2u^j-1\right|}-1\]\\ T_k=T_0\exp (-{\mathrm{ck}}^{1/D})\end{array}\right.---\left(6\right)$

In formula (6), x^j _k+1For the new explanation of current iteration, x^j _kFor the new explanation of last iteration, y^jSet for j-th system to be asked The disturbance of meter parameter, 1≤j≤D, D are the total quantity of system design parameterses to be asked, B^jCan for j-th system design parameters to be asked The maximum that can take, A^jThe minima that may take for j-th system design parameters to be asked, u^jIt is between the random value between 0 and 1, T_kFor the annealing temperature of current iteration, T₀For the initial temperature of simulated annealing, c is the temperature control system of simulated annealing Number, k is current temperature change time numerical value；

3.4) calculate the new explanation x of current iteration^j _k+1Corresponding target function valueCalculating target function value Relatively last target function valueBetween target function value increment

3.5) judge target function value incrementWhether setting up less than 0, if set up, accepting the new optimal solution producing For current optimal solution；If be false, to judge whether that with specified probability P the optimal solution accepting newly to produce is currently optimum Solution；If accepting the new optimal solution producing is current optimal solution, before updating tempering, accept solution number of times m；

3.6) accepting before judging to update tempering whether solution number of times m reaches predetermined threshold value, if reaching predetermined threshold value, carrying out The temper of simulated annealing, and reset acceptance solution number of times m before tempering after temper finishes；

3.7) update iterationses L_kValue, judge new iterationses L_kWhether reach predetermined threshold value, if not yet reached Predetermined threshold value then redirects execution step 3.3), otherwise redirect execution step 3.8)；

3.8) if the continuous N number of new explanation of specified quantity is not all accepted or the cost function value of simulated annealing is less than Set-point ε (0 between one 0 to 1<ε<1), then judge surely to reach the end condition of simulated annealing, redirect execution step 4)；Otherwise, reset iterationses L_kIf annealing process slowly, is simulated the Quenching Treatment of annealing algorithm, and redirects and hold Row step 3.3).

In the present embodiment, step 3.4) in judge whether that the optimal solution accepting newly to produce is current with specified probability P Optimal solution specifically refers to：Randomly generate an equally distributed random number ζ on [0,1] interval first, then always with formula (7) institute Show Metropolis criterion to calculate probability P, if probability P is more than random number ζ, accepting the new optimal solution producing is currently Excellent solution, the optimal solution otherwise abandoning accepting newly to produce is current optimal solution；

$P=\exp \&lsqb;-\frac{\mathrm{\&Delta;}\stackrel{\&OverBar;}{\mathrm{\&eta;}}}{T_k}\&rsqb;---\left(7\right)$

In formula (7), P is probability to be calculated,For target function value increment, T_kAnnealing temperature for current iteration.

In the present embodiment, step 3.6) in be simulated annealing algorithm the detailed step of temper include：

3.6.1) according to formula (8), finite difference calculus calculates the gradient of each system design parameters in vector x；

$s_j=\left|\frac{\stackrel{\&OverBar;}{\mathrm{\&eta;}}(x_{opt}+e_j\mathrm{\&delta;})-\stackrel{\&OverBar;}{\mathrm{\&eta;}}\left(x_{opt}\right)}{\mathrm{\&delta;}}\right|---\left(8\right)$

In formula (8), s_jRepresent the gradient of j-th system design parameters in vector x,Weighted average for all operating points Efficiency, x_optFor the optimal solution in the current all solutions having found, e_jIt is the vector of D for a dimension, D is system design parameterses Quantity, vectorial e_jIn j-th component be 1, remaining component be 0, δ be step-length；

3.6.2) it is directed to each system design parameters, if the gradient of this system design parameters is equal to default gradient threshold Value s_max, then redirect execution step 3.6.3)；Otherwise exit temper；

3.6.3) annealing temperature T is updated according to formula (9)_kValue with temperature change number of times k；

$\left\{\begin{array}{c}{T}_{k^{\&prime;}}^{\&prime;}=\frac{s_{max}}{s_j}{T}_k\\ k^{\&prime;}={\&lsqb;ln(T_0/T_{k^{\&prime;}}^{\&prime;})/c\&rsqb;}^D\end{array}\right.---\left(9\right)$

In formula (9),For the annealing temperature after updating, s_maxDefault Grads threshold, s_jRepresent j-th system in vector x The gradient of design parameter, s_jRepresent the gradient of j-th system design parameters in vector x, T_kFor the annealing temperature before updating；K' is Temperature change number of times after renewal, T₀For the initial temperature of simulated annealing, c is the temperature control system of simulated annealing Number, D is the quantity of system design parameterses.

In the present embodiment, step 3.8) in Quenching Treatment specifically refer to update annealing temperature T according to formula (10)_kValue；

$T_k=T_0\exp (-{\mathrm{ck}}^{q_j/D})---\left(10\right)$

In formula (10), T_kFor the annealing temperature after updating, T₀For the initial temperature of simulated annealing, c is simulated annealing calculation The temperature control coefficient of method, k is temperature change number of times, q_jFor the quenching factor of simulated annealing, D is system design parameterses Quantity.

The above is only the preferred embodiment of the present invention, and protection scope of the present invention is not limited merely to above-mentioned enforcement Example, all technical schemes belonging under thinking of the present invention belong to protection scope of the present invention.It should be pointed out that for the art Those of ordinary skill for, some improvements and modifications without departing from the principles of the present invention, these improvements and modifications Should be regarded as protection scope of the present invention.

Claims (9)

1. a kind of direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method is it is characterised in that step includes：

1) system design parameterses of direct-drive permanent magnet wind power generator are determined；

2) constrained with the current/voltage of current transformer and control algolithm is constrained to constraints, with the weighted average effect of multiple operating points Rate highest or the minimum object function of weighted average power loss, set up design mathematic model based on described system design parameterses；

3) obtain one group of initial operating mode value of system design parameterses based on rated point design, based on described initial operating mode value, adopt Simulated annealing carries out multi-state iterative to described design mathematic model, obtains multigroup solving result；

4) select the optimal system design parameter output that one group of optimal result obtains as design from all solving results；

Described step 3) in detailed step as follows：

3.1) one group of initial operating mode value of system design parameterses is obtained based on rated point design；

3.2) it is based on described initial operating mode value, described design mathematic model is randomly generated with one initially using simulated annealing Solution x₀, calculate initial solution x₀Target function valueInitialization iterationses L_k, the initial temperature of initialization simulated annealing Degree T₀, temperature change number of times k and tempering before accept solution number of times m；Redirect execution step 3.3) it is iterated；

3.3) random disturbance is made to current optimal solution, generate the new explanation of current iteration by function expression shown in formula (6)；

$\left\{\begin{array}{c}{x^j}_{k+1}=x_k^j+y^j(B_j-A_j)\\ y^j=sgn(u^j-\frac{1}{2})T_k\&lsqb;{(1+1/T_k)}^{|2u^j-1|}-1\&rsqb;\\ T_k=T_0\exp (-{\mathrm{ck}}^{1/D})\end{array}\right.---\left(6\right)$

In formula (6), x^j _k+1For the new explanation of current iteration, x^j _kFor the new explanation of last iteration, y^jFor j-th system design ginseng to be asked The disturbance of number, 1≤j≤D, D are the total quantity of system design parameterses to be asked, B^jMay take for j-th system design parameters to be asked Maximum, A^jThe minima that may take for j-th system design parameters to be asked, u^jIt is between the random value between 0 and 1, T_kFor The annealing temperature of current iteration, T₀For the initial temperature of simulated annealing, c is the temperature control coefficient of simulated annealing, k For current temperature change time numerical value；

3.4) calculate the new explanation x of current iteration^j _k+1Corresponding target function valueCalculating target function valueRelatively go up Target function value onceBetween target function value increment

3.5) judge target function value incrementWhether set up less than 0, if set up, accepting the new optimal solution producing is to work as Front optimal solution；If be false, to judge whether with specified probability P to accept the optimal solution newly producing for current optimal solution； If accepting the new optimal solution producing is current optimal solution, before updating tempering, accept solution number of times m；

3.6) accepting before judging to update tempering whether solution number of times m reaches predetermined threshold value, if reaching predetermined threshold value, being simulated The temper of annealing algorithm, and reset acceptance solution number of times m before tempering after temper finishes；

3.7) update iterationses L_kValue, judge new iterationses L_kWhether reach predetermined threshold value, if not yet reaching default Threshold value then redirects execution step 3.3), otherwise redirect execution step 3.8)；

3.8) if the continuous N number of new explanation of specified quantity is not all accepted or the cost function value of simulated annealing is less than one Set-point between 0 to 1, then judge surely to reach the end condition of simulated annealing, redirect execution step 4)；Otherwise, reset Iterationses L_kIf annealing process slowly, is simulated the Quenching Treatment of annealing algorithm, and redirects execution step 3.3).

2. direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method according to claim 1, its feature Be, described step 1) in the system design parameterses of direct-drive permanent magnet wind power generator that determine include frame bore D_i1, fixed Sub- core length L_s, close amplitude B of stator yoke magnetic_ysm, the high h of stator slot_s, stator groove width b_s, permanent magnet height h_m, permanent magnet width b_m、 To pole span τ_p, MgO-ZrO_2 brick q.

3. direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method according to claim 2, its feature Be, described step 2) in current/voltage constrain as shown in formula (1)；

$\left\{\begin{array}{c}u\&le;U_{\lim }\\ i\&le;I_{\lim }\end{array}\right.---\left(1\right)$

In formula (1), u is current transformer voltage, i is current transformer electric current, U_limFor voltage binding occurrence, I_limFor restriction of current value.

4. direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method according to claim 3, its feature Be, described step 2) in control algolithm constraint as shown in formula (2) three constraint in any one；

$\left\{\begin{array}{c}i\&RightArrow;\min \\ u=u_c\\ Q_s=0\end{array}\right.---\left(2\right)$

In formula (2), i is current transformer electric current, and → min represents optimal current control, and u is current transformer voltage, u_cFor constant voltage constraint Value, Q_sFor unit power factor.

5. direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method according to claim 4, its feature Be, described step 2) in object function such as formula (3) or formula (4) shown in；

$\stackrel{\&OverBar;}{\mathrm{\&eta;}}=\frac{{\mathrm{\&Sigma;\&eta;}}_i\left(x\right)W_i}{{\mathrm{\&Sigma;W}}_i}\&RightArrow;max---\left(3\right)$

In formula (3),For the weighted average efficiency of all operating points, η_iX () is the average efficiency of i-th operating point, i is more than or equal to 3, W_iFor the weight factor of i-th operating point, → max represents weighted average efficiency highest constraints；

$\stackrel{\&OverBar;}{p}=\frac{{\mathrm{\&Sigma;p}}_i\left(x\right)W_i}{{\mathrm{\&Sigma;W}}_i}\&RightArrow;min---\left(4\right)$

In formula (4),For the weighted average power loss of all operating points, p_iX () is the power loss of i-th operating point, i is big In equal to 3, W_iFor the weight factor of i-th operating point, → min represents the minimum constraints of weighted average power loss.

6. direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method according to claim 5, its feature It is, described step 2) shown in the design mathematic model such as formula (5) set up；

$\left\{\begin{array}{c}\underset{X\&Element;D}{max}:{\mathrm{\&eta;}}_i\left(X\right)/\underset{X\&Element;D}{min}:p_i\left(X\right)\\ D=\left\{X\right|g_i\left(X\right)\&le;0,i=1,2,...,m\}\\ X={\&lsqb;D_{i1},L_s,B_{ysm},h_s,b_s,h_m,b_m,{\mathrm{\&tau;}}_p,q\&rsqb;}^T\end{array}\right.---\left(5\right)$

In formula (5), D is the total quantity of system design parameterses to be asked, η_i(X) be i-th operating point average efficiency, p_i(X) it is the The power loss of i operating point, i is more than or equal to 3,For the average efficiency of i-th operating point maximum or The power loss of i-th operating point of person is minimum, g_i(X) it is constraints, m is the quantity of constraints, X is system design to be asked The matrix that parameter is constituted, D_i1For frame bore, L_sFor stator core length, B_ysmFor the close amplitude of stator yoke magnetic, h_sFor stator Groove is high, b_sFor stator groove width, h_mFor permanent magnet height, b_mFor permanent magnet width, τ_pBe to pole span, q be MgO-ZrO_2 brick.

7. direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method according to claim 1, its feature Be, described step 3.4) in judge whether that the optimal solution accepting newly to produce is concrete for current optimal solution with specified probability P Refer to：Randomly generate an equally distributed random number ζ on [0,1] interval first, then always with formula (7) Suo Shi Metropolis criterion, to calculate probability P, if probability P is more than random number ζ, accepts the new optimal solution producing for currently optimum Solution, the optimal solution otherwise abandoning accepting newly to produce is current optimal solution；

$P=\exp \&lsqb;-\frac{\mathrm{\&Delta;}\stackrel{\&OverBar;}{\mathrm{\&eta;}}}{T_k}\&rsqb;---\left(7\right)$

In formula (7), P is probability to be calculated,For target function value increment, T_kAnnealing temperature for current iteration.

8. direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method according to claim 7, its feature Be, described step 3.6) in be simulated the detailed step of temper of annealing algorithm and include：

3.6.1) according to formula (8), finite difference calculus calculates the gradient of each system design parameters in vector x；

$s_j=\left|\frac{\stackrel{\&OverBar;}{\mathrm{\&eta;}}(x_{opt}+e_j\mathrm{\&delta;})-\stackrel{\&OverBar;}{\mathrm{\&eta;}}\left(x_{opt}\right)}{\mathrm{\&delta;}}\right|---\left(8\right)$

In formula (8), s_jRepresent the gradient of j-th system design parameters in vector x,For the weighted average efficiency of all operating points, x_optFor the optimal solution in the current all solutions having found, e_jIt is the vector of D for a dimension, D is the number of system design parameterses Amount, vectorial e_jIn j-th component be 1, remaining component be 0, δ be step-length；

3.6.2) it is directed to each system design parameters, if the gradient of this system design parameters is equal to default Grads threshold s_max, then redirect execution step 3.6.3)；Otherwise exit temper；

3.6.3) annealing temperature T is updated according to formula (9)_kValue with temperature change number of times k；

$\{\begin{array}{c}{T}_{k^{\&prime;}}^{\&prime;}=\frac{s_{max}}{s_j}{T}_k\\ k^{\&prime;}={\&lsqb;ln(T_0/T_{k^{\&prime;}}^{\&prime;})/c\&rsqb;}^D\end{array}---\left(9\right)$

In formula (9), T '_k′For the annealing temperature after updating, s_maxDefault Grads threshold, s_jRepresent that in vector x, j-th system sets The gradient of meter parameter, s_jRepresent the gradient of j-th system design parameters in vector x, T_kFor the annealing temperature before updating；K ' is more Temperature change number of times after new, T₀For the initial temperature of simulated annealing, c is the temperature control coefficient of simulated annealing, D Quantity for system design parameterses.

9. direct-drive permanent magnet wind power generator multi-state global efficiency optimal-design method according to claim 8, its feature Be, described step 3.8) in Quenching Treatment specifically refer to according to formula (10) update annealing temperature T_kValue；

$T_k=T_0\exp (-{\mathrm{ck}}^{q_j/D})---\left(10\right)$

In formula (10), T_kFor the annealing temperature after updating, T₀For the initial temperature of simulated annealing, c is simulated annealing Temperature control coefficient, k is temperature change number of times, q_jFor the quenching factor of simulated annealing, D is the number of system design parameterses Amount.