CN103437955B - Minitype permanent magnetism directly-driving wind power generation system maximum power tracking device and controlling method - Google Patents

Abstract

The invention discloses a kind of minitype permanent magnetism directly-driving wind power generation system maximum power tracking device and the controlling method of technical field of wind power generation.Wherein, device comprises blower fan, MPPT maximum power point tracking MPPT controller, rectifier, a n air velocity transducer, speed probe, voltage transducer, current sensor, DC-DC converter, driver module, the first electric capacity, the second electric capacity and load; Obtain wind velocity vector by the multiple air velocity transducers being installed on diverse location, and gather the actual sample of a large amount of wind velocity vector-optimum speed, utilize support vector machine to set up wind speed-optimum speed forecasting model.Maximal power tracing is carried out by being combined with little step-length disturbance observation method by forecasting model.Invention increases tracking velocity, effectively reduce the power loss of perturbation process; And after the characteristic of blower fan changes, can, by again collecting sample, train new forecasting model to ensure precision of prediction.

Description

Minitype permanent magnetism directly-driving wind power generation system maximum power tracking device and controlling method

Technical field

The invention belongs to technical field of wind power generation, particularly relate to a kind of minitype permanent magnetism directly-driving wind power generation system maximum power tracking device and controlling method.

Background technique

Along with the day of Energy situation is becoming tight, the distributed power generation being representative with small-size wind power-generating etc. and power-saving technology more and more come into one's own, and become current study hotspot.

Wind energy is the energy that a kind of randomness is very large, ensure to capture wind energy to greatest extent, and variable speed wind sends out the control strategy that system generally adopts MPPT maximum power point tracking (MPPT) at present.For minitype permanent magnetism directly-driving wind power generation system, conventional MPPT method has the methods such as optimized rotating speed is given, disturbance observation.

The given ratio juris of optimized rotating speed is: the power that under certain wind speed, blower fan absorbs has a maximum power point (MPP), and output power value is P _max, a corresponding optimized rotating speed is ω _max, according to power-rotation speed characteristic that blower fan producer provides, the optimized rotating speed under certain wind speed can be determined easily, its rotating speed of target as blower fan is carried out controlling.This method has two shortcomings: one is the Measurement accuracy being difficult to realize wind speed.Because the area of draught fan impeller is comparatively large, in whole impeller area, be not that the wind speed of each position is all consistent, the error of wind speed measurement is larger; Two is the impacts along with external conditions such as blower fan wearing and tearing, and the power-rotation speed characteristic of blower fan will change, and is thus difficult to ensure to follow the tracks of MPP accurately.

The principle of disturbance observation method (P & O) is: constantly apply a fixing disturbing quantity to the rotating speed of blower fan, and the direction of disturbing quantity is next time determined according to the change direction that power caught by blower fan, real work point can be made constantly to move towards MPP.The realization of P & O is relatively easy, but the operation point found can only near MPP oscillatory operation, cause the loss of Partial Power.In addition, initial value and disturbance step-length have larger impact to the precision of following the tracks of and speed, sometimes misjudgment phenomenon occur.

Summary of the invention

Accurately determine the deficiencies such as poor, easy generation erroneous judgement for the existing peak power output tracking method mentioned in above-mentioned background technology, the present invention proposes a kind of minitype permanent magnetism directly-driving wind power generation system maximum power tracking device and controlling method.

A kind of minitype permanent magnetism directly-driving wind power generation system maximum power tracking device, it is characterized in that, described device comprises blower fan, MPPT maximum power point tracking MPPT controller, rectifier, a n air velocity transducer, speed probe, voltage transducer, current sensor, DC-DC converter, driver module, the first electric capacity, the second electric capacity and load;

Wherein, the three-phase input end of described rectifier is connected with the three-phase output end of blower fan, and the single-phase output plus terminal of rectifier is connected with the positive pole of the first electric capacity, the single-phase output negativing ending grounding of rectifier; First electric capacity minus earth;

The voltage input end to be measured of described voltage transducer is connected with the first capacitance cathode, voltage transducer voltage output end ground connection to be measured; The measurement signal output terminal of voltage transducer is connected with MPPT maximum power point tracking MPPT controller;

The current input terminal to be measured of described current sensor is connected with voltage transducer positive pole, and the current output terminal to be measured of current sensor is connected with the input end of DC-DC converter; The measurement signal output terminal of current sensor is connected with MPPT maximum power point tracking MPPT controller;

The pulse-width signal input end of described DC-DC converter is connected with driver module one end, the other one end of driver module is connected with MPPT controller; The output terminal of DC-DC converter is connected with the second capacitance cathode; Second electric capacity minus earth;

The measurement signal output terminal of a described n air velocity transducer is connected with MPPT maximum power point tracking MPPT controller respectively;

Described speed probe two input ends are connected with the wherein two ends in blower fan three-phase output end, and the measurement signal output terminal of speed probe is connected with MPPT controller;

Described load one end is connected with the second capacitance cathode, load other one end ground connection.

Described n air velocity transducer collection is installed on front side of draught fan impeller, coaxial with impeller circular area, parallel and diverse location in equal-sized plane.

Described DC-DC converter adopts Boost circuit.

Described speed probe adopts voltage zero-cross to detect formula frequency meter.

A kind of minitype permanent magnetism directly-driving wind power generation system maximum power tracking method, it is characterized in that, described method specifically comprises step:

Step 1: the air speed data of the diverse location adopting n air velocity transducer to export forms wind velocity vector V=[V ₁, V ₂..., V _n] ^t;

Step 2: with wind velocity vector V=[V ₁, V ₂..., V _n] ^tas input, SVM prediction model is utilized to obtain optimized rotating speed predicted value ω corresponding to maximum power point _ref;

Step 3: passing ratio integral control method regulates the rotating speed of blower fan, makes blower fan reach optimized rotating speed predicted value ω corresponding to maximum power point _ref;

Step 4: with the optimized rotating speed predicted value ω that maximum power point is corresponding _reffor initial value, disturbance observation method is adopted to follow the tracks of the peak output of blower fan with the disturbance step delta ω of setting;

Step 5: the power difference before and after the disturbance that disturbance observation method is tried to achieve is more than or equal to setting threshold value T _rtime, illustrate that wind speed there occurs sudden change, repeat step 1 to step 4; Otherwise, continue to adopt disturbance observation method to follow the tracks of the peak output of blower fan with the disturbance step-length of setting.

In step 2, SVM prediction model is utilized to obtain optimized rotating speed predicted value ω corresponding to maximum power point _refprocess be:

The effect of support vector machines forecasting model is the wind velocity vector V=[V according to being recorded by multiple air velocity transducer ₁, V ₂..., V _n] ^tto blowing machine maximum power point optimum speed ω _optpredicted value ω _ref;

Step 201: collect training sample;

Remember that the wind velocity vector under a certain wind speed environments is V (i)=[V ₁(i), V ₂(i) ..., V _n(i)] ^t, corresponding blower fan maximum power point rotating speed is ω _opti (), then can form pair of sample (V (i), ω _opt(i)); By collecting the sample pair under various different wind speed environments, form sample set { (V (i), ω _opt(i)) },

Under a certain wind speed environments, the acquisition of training sample adopts method of trial to gather; Gatherer process is:

Step 2011: the pulse duty factor D of the pwm signal of initialization DC-DC converter is with less initial value D ₀, make it constantly increase with fixed increment Δ D at every turn, for kth time, have

D(k)=D ₀+k·ΔD （1）

Wherein: D (k) is kth sub-pulse duty cycle;

D ₀for dutycycle initial value;

Δ D is fixed increment;

Step 2012: gather the VD V of blower fan after rectifier by voltage transducer and current sensor _dc(k) and average anode current I _dck (), calculates the output power P (k) of current blower fan:

P(k)=V _dc(k)·I _dc(k) （2）

Step 2013: compare with the output power P (k-1) of front primary air fan, when there is P (k) <P (k-1), then thinks that the working state of now blower fan is close to maximum power point; Order:

D(k)=D ₀+(k-0.5)·ΔD （3）

Record rotation speed of fan is now as maximum power point rotational speed omega _opt(i), and wind velocity vector V (i)=[V ₁, V ₂..., V _n] ^t, complete and once gather, namely obtain pair of sample V (i), ω _opt(i));

Step 202: Training Support Vector Machines SVM model; Detailed process is:

Step 2021: given sample set wherein, X _i∈ R ⁿfor input vector, y _i∈ R is corresponding output value, and N is number of samples, and n is input vector dimension;

Step 2022: setting support vector machines linear regression function used is:

y _i=f(X _i)=Wφ(X _i)+b （4）

Wherein: y _ifor linear regression function exports;

φ (X _i) be Nonlinear Mapping from the input space to high-dimensional feature space;

X _ifor input vector;

W is weight vector;

B is biased;

Weight vector W and biased b calculates by minimizing formula (5):

$\frac{1}{2}{\left|\right|W\left|\right|}^2+C\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&xi;}}_i$

$s.t.\left\{\begin{array}{c}{y}_i-\mathrm{W\&phi;}\left(X_i\right)-b\&le;\mathrm{\&epsiv;}+{\mathrm{\&xi;}}_i\\ {\mathrm{\&xi;}}_i\&GreaterEqual;0\end{array}\right.---\left(5\right)$

Wherein: W is weight vector, the 1st determine the generalization ability of regression function; C is penalty factor (C > 0), for controlling the punishment degree to the sample exceeded; N is number of samples; ξ i is the slack variable introduced; ε is error;

Step 2023: set up Lagrange's equation according to minimizing formula (5), solving linear regression function is:

$\left(X_i\right)=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_{j}K(X_i,X_j)+b---\left(6\right)$

Wherein: K (X _i, X _j) be kernel function, kernel function is Gaussian function δ ²for the width parameter of gaussian kernel function; α _jfor Lagrange coefficient; X _jfor sample vector, and non-vanishing α _jcorresponding vectorial X _jbe called support vector; After obtaining support vector, regression function y=f (X can be tried to achieve _i);

Step 2024: adopt statistic average relative error Δ _mREthe performance of valuation prediction models; Its representation is:

${\mathrm{\&Delta;}}_{\mathrm{MRE}}=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}\left|\frac{Y-\hat{Y}}{Y}\right|\&times;100\%---\left(7\right)$

In formula:

Δ _mREfor statistic average relative error;

Y is the true value of sample;

for the estimated value of Y;

Step 2025: evenly extract 3/5ths in total sample as training sample, all the other get different C and δ respectively 2/5ths as test samples ², utilize training sample to learn, and calculate the Δ on test samples _mRE, select minimum Δ _mREcorresponding model is as final forecasting model;

Step 203: by wind velocity vector V=[V ₁, V ₂..., V _n] ^tobtaining blower fan maximum power point rotor speed forecast value by final forecasting model is ω _ref.

Beneficial effect of the present invention is: working speed, when external environment changes, directly can be adjusted to optimized rotating speed predicted value ω by means of forecasting model by (1) _refnear, avoid the process that disturbance observation method P & O progressively sounds out, thus improve tracking velocity; (2) with optimized rotating speed predicted value ω _reffor initial value carry out disturbance observe process time, due to optimized rotating speed predicted value ω _refvery close to actual optimum rotating speed, therefore less disturbance step-length can be set, thus effectively reduce the power loss of perturbation process; (3) after the characteristic of blower fan changes, can, by again collecting sample, train new forecasting model to ensure precision of prediction.

Accompanying drawing explanation

Fig. 1 is hardware structure diagram of the present invention;

Fig. 2 is disturbance observation method P & O algorithm flow chart;

Fig. 3 is the flow chart of a kind of minitype permanent magnetism directly-driving wind power generation system maximum power tracking method provided by the invention;

Wherein, 1-air velocity transducer; 2-permanent magnet direct-driving aerogenerator; 3-speed probe; 4-rectifier; 5-voltage transducer; 6-first electric capacity; 7-second electric capacity; 8-current sensor.

Embodiment

Below in conjunction with accompanying drawing, preferred embodiment is elaborated.It should be emphasized that following explanation is only exemplary, instead of in order to limit the scope of the invention and apply.

Fig. 1 is the hardware structure diagram of the embodiment of the present invention, and wherein, minitype permanent magnetism direct wind-driven generator 2 major parameter is: rotor diameter is 1.2m, rated power is 300W, and voltage rating is 24V, rated speed 800r/min, threshold wind velocity 1m/s, rated wind speed 10m/s, survival wind speed 25m/s, the major parameter of maximum power tracking device is: MPPT controller adopts dsPIC33FJ06GS101 single-chip microcomputer, DC-DC converter adopts Boost circuit, driver module selects MCP14E3, voltage transducer selecting and purchasing LV28-P, LA25-NP selected by current sensor 8, air velocity transducer 1 adopts JL-FS2, totally 5: to be installed on front side of draught fan impeller 0.5 meter respectively, coaxial with impeller circular area, parallel and circle centre position in equal-sized circular flat and the distance center of circle 1/4 radius, 1/2 radius, 3/4 radius, 1 times of radius, speed probe 3 adopts voltage zero-cross to detect formula frequency meter, first electric capacity C1=10uF, second electric capacity C2=100uF.

The collection of training sample is carried out according to step 201.In embodiment, utilize blower, frequency variator and straight length to form small test wind-tunnel, control blower by regulating the frequency setting value of frequency variator and exert oneself, and then create different wind speed environments.Control frequency converter frequency from 10Hz, be that interval rises to 60Hz with 0.2Hz, 250 kinds of wind speed environments can be provided, i.e. { V (i)=[V ₁(i), V ₂(i), V ₃(i), V ₄(i), V ₅(i)] ^t, i=1 ..., 250}, thus 250 pairs of samples can be constructed.

The training of SVM model is carried out according to step 202.150 couple in the total sample of even extraction is as training sample, and all the other are 100 to as test samples.For preventing study phenomenon or owing study phenomenon, got different C=10 respectively ^-1, 10 ⁰, 10 ¹, 10 ², 10 ³, δ ²=10 ^-2, 10 ^-1, 10 ⁰, 10 ¹, 10 ², utilize training sample to learn, and calculate the Δ on test samples _mRE, select minimum Δ _mREcorresponding model, as final forecasting model, wherein comprises 56 support vectors altogether, by the storage of these support vectors write single-chip microcomputer for calling.

Fig. 2 is the flow chart of disturbance observation method P & O, its principle is the rotating speed (ω+Δ ω) of periodically disturbance blower fan, compare the changed power before and after its disturbance again, if output power increases, then represent that perturbation direction is correct, continues (+Δ ω) disturbance in the same direction; If output power reduces, then towards contrary (-Δ ω) direction disturbance.A/D module due to dsPIC33FJ06GS101 completes once to sample only needs 0.5us, when measuring VD, the average anode current of blower fan, in order to eliminate the signal jitter that the copped wave of DC-DC converter medium-high frequency causes, making A/D module change 10 times continuously and averaging as measured value.

Fig. 3 is the flow chart of a kind of minitype permanent magnetism directly-driving wind power generation system maximum power tracking method provided by the invention.During beginning, first gather the output being installed on 5 air velocity transducers of diverse location, form wind velocity vector V=[V ₁, V ₂, V ₃, V ₄, V ₅] ^t, sent into SVM prediction model, obtained the optimized rotating speed predicted value ω that maximum power point is corresponding _ref; Proportional integral PI controller is according to blower fan actual speed ω and ω _refthe PWM dutycycle of deviation adjustment DC-DC converter, make real work voltage track to ω fast _ref.Then with ω _reffor initial value, less disturbance step delta ω is utilized to start P & O process.In P & O process, the difference power Δ P before and after disturbance is each time compared with a certain threshold value Tr, when | Δ P|<T _rtime, continue P & O process, otherwise think that now wind speed environments there occurs sudden change, again go out ω by support vector machines model prediction _ref, and repeat said process.For the blower fan of 300W, the present embodiment gets T _r=10W.

By above-mentioned maximal power tracing MPPT method by C programmer write control chip dsPIC33FJ06GS101, export PWM square wave and drive DC-DC converter, maximal power tracing function can be realized.

The correctness of extracting method in order to verify, itself and conventional disturbance observation method P & O compare by hardware platform of the present invention.Be specially: under same wind speed environments (frequency converter frequency arranging wind-tunnel is 40Hz), controlling blower fan initial speed is ω ₀=600r/min, compares the tracking velocity of two kinds of methods and the average power of steady-state process respectively.

The tracing process of conventional disturbance observation method P & O method (adjustment cycle T=5s, step delta ω=20r/min) is: after the continuous 15 postive direction disturbances of clapping, start at maximum power point MPP(ω _opt=900r/min) left and right vibration, namely enter steady-state process, needed time is about 75 seconds altogether.

The tracing process of the method for the invention is first measure wind velocity vector, then provides maximum power point rotor speed forecast value ω through support vector machines model _refafter=915r/min, directly working speed ω is adjusted to 915r/min by PI controller, then starts to carry out little step-length disturbance and observe P & O(adjustment cycle T=5s, step delta ω=5r/min), due to ω _refitself is maximum power point MPP(ω closely _opt=900r/min), after the continuous 3 negative direction disturbances of clapping, enter steady-state process, needed time is about 15 seconds altogether.It can thus be appreciated that the tracking velocity of institute of the present invention extracting method will apparently higher than conventional disturbance observation method P & O method.

After entering steady-state process, calculate the average power of 60 seconds steady-state processs respectively, show that conventional disturbance observation method P & O method is 210 watts, and the method for the invention is 236 watts, this illustrates and adopts the method for the invention to effectively reduce power loss.

The above; be only the present invention's preferably embodiment, but protection scope of the present invention is not limited thereto, is anyly familiar with those skilled in the art in the technical scope that the present invention discloses; the change that can expect easily or replacement, all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of claim.

Claims (2)

1. a minitype permanent magnetism directly-driving wind power generation system maximum power tracking method, its minitype permanent magnetism directly-driving wind power generation system maximum power tracking device comprises blower fan, MPPT maximum power point tracking MPPT controller, rectifier, a n air velocity transducer, speed probe, voltage transducer, current sensor, DC-DC converter, driver module, the first electric capacity, the second electric capacity and load;

Wherein, the three-phase input end of described rectifier is connected with the three-phase output end of blower fan, and the single-phase output plus terminal of rectifier is connected with the positive pole of the first electric capacity, the single-phase output negativing ending grounding of rectifier; First electric capacity minus earth;

The voltage input end to be measured of described voltage transducer is connected with the first capacitance cathode, voltage transducer voltage output end ground connection to be measured; The measurement signal output terminal of voltage transducer is connected with MPPT maximum power point tracking MPPT controller;

The current input terminal to be measured of described current sensor is connected with voltage transducer positive pole, and the current output terminal to be measured of current sensor is connected with the input end of DC-DC converter; The measurement signal output terminal of current sensor is connected with MPPT maximum power point tracking MPPT controller;

The pulse-width signal input end of described DC-DC converter is connected with driver module one end, the other one end of driver module is connected with MPPT controller; The output terminal of DC-DC converter is connected with the second capacitance cathode; Second electric capacity minus earth;

The measurement signal output terminal of a described n air velocity transducer is connected with MPPT maximum power point tracking MPPT controller respectively;

Described speed probe two input ends are connected with the wherein two ends in blower fan three-phase output end, and the measurement signal output terminal of speed probe is connected with MPPT controller;

Described load one end is connected with the second capacitance cathode, load other one end ground connection;

Described n air velocity transducer collection is installed on front side of draught fan impeller, coaxial with impeller circular area, parallel and diverse location in equal-sized plane;

Described DC-DC converter adopts Boost circuit;

Described speed probe adopts voltage zero-cross to detect formula frequency meter; It is characterized in that, described method specifically comprises step:

Step 1: the air speed data of the diverse location adopting n air velocity transducer to export forms wind velocity vector V=[V ₁, V ₂..., V _n] ^t;

Step 2: with wind velocity vector V=[V ₁, V ₂..., V _n] ^tas input, SVM prediction model is utilized to obtain optimized rotating speed predicted value ω corresponding to maximum power point _ref;

Step 3: passing ratio integral control method regulates the rotating speed of blower fan, makes blower fan reach optimized rotating speed predicted value ω corresponding to maximum power point _ref;

Step 4: with the optimized rotating speed predicted value ω that maximum power point is corresponding _reffor initial value, disturbance observation method is adopted to follow the tracks of the peak output of blower fan with the disturbance step delta ω of setting;

Step 5: the power difference before and after the disturbance that disturbance observation method is tried to achieve is more than or equal to setting threshold value T _rtime, illustrate that wind speed there occurs sudden change, repeat step 1 to step 4; Otherwise, continue to adopt disturbance observation method to follow the tracks of the peak output of blower fan with the disturbance step-length of setting.

2. method according to claim 1, is characterized in that, in described step 2, utilizes SVM prediction model to obtain optimized rotating speed predicted value ω corresponding to maximum power point _refprocess be:

Step 201: collect training sample;

Remember that the wind velocity vector under a certain wind speed environments is V (i)=[V ₁(i), V ₂(i) ..., V _n(i)] ^t, corresponding blower fan maximum power point rotating speed is ω _opti (), then can form pair of sample (V (i), ω _opt(i)); By collecting the sample pair under various different wind speed environments, form sample set { (V (i), ω _opt(i)) }; Gatherer process is:

Step 2011: the pulse duty factor D of the pwm signal of initialization DC-DC converter is with less initial value D ₀, make it constantly increase with fixed increment Δ D at every turn, for kth time, have:

D(k)＝D ₀+k·ΔD

Wherein: D (k) is kth sub-pulse duty cycle;

D ₀for dutycycle initial value;

Δ D is fixed increment;

Step 2012: gather the VD V of blower fan after rectifier by voltage transducer and current sensor _dc(k) and average anode current I _dck (), calculates the output power P (k) of current blower fan:

P(k)＝V _dc(k)·I _dc(k)

Step 2013: compare with the output power P (k-1) of front primary air fan, work as appearance

Time P (k) <P (k-1), then think that the working state of now blower fan is close to maximum power point; Order:

D(k)＝D ₀+(k-0.5)·ΔD

Record rotation speed of fan is now as maximum power point rotational speed omega _opt(i), and wind velocity vector V (i)=[V ₁, V ₂..., V _n] ^t, complete and once gather, namely obtain pair of sample V (i), ω _opt(i));

Step 202: Training Support Vector Machines SVM model; Detailed process is:

Step 2021: given sample set wherein, X _i∈ R ⁿfor input vector, y _i∈ R is corresponding output value, and N is number of samples, and n is input vector dimension;

Step 2022: setting support vector machines linear regression function used is:

y _i＝f(X _i)＝Wφ(X _i)+b

Wherein: y _ifor linear regression function exports;

φ (X _i) be Nonlinear Mapping from the input space to high-dimensional feature space;

X _ifor input vector;

W is weight vector;

B is biased;

Weight vector W and biased b by minimizing formulae discovery formula is:

$\frac{1}{2}\left|\right|W|{|}^2+C\frac{1}{N}\underset{i=1}{\overset{N}{\&Sigma;}}{\mathrm{\&xi;}}_i$

$s.t.\left\{\begin{array}{c}{y}_i-W\mathrm{\&phi;}\left(X_i\right)-b\&le;\mathrm{\&epsiv;}+{\mathrm{\&xi;}}_i\\ {\mathrm{\&xi;}}_i\&GreaterEqual;0\end{array}\right.$

Wherein: W is weight vector, the 1st determine the generalization ability of regression function; C is penalty factor (C>0), for controlling the punishment degree to the sample exceeded; N is number of samples; ξ _ifor the slack variable introduced; ε is error;

Step 2023: set up Lagrange's equation according to minimizing formula, solving linear regression function is:

$f\left(X_i\right)=\underset{j=1}{\overset{N}{\&Sigma;}}{\mathrm{\&alpha;}}_{j}K(X_i,X_j)+b$

Wherein: K (X _i, X _j) be kernel function, kernel function is Gaussian function: $K(X_i,X_j)=\exp (-\frac{\left|\right|X_i-X_j|{|}^2}{{\mathrm{\&delta;}}^2}),$ δ ²for the width parameter of gaussian kernel function; α _jfor Lagrange coefficient; X _jfor sample vector, and non-vanishing α _jcorresponding vectorial X _jbe called support vector;

Step 2024: adopt statistic average relative error Δ _mREthe performance of valuation prediction models; Its representation is:

${\mathrm{\&Delta;}}_{MRE}=\frac{1}{N}\underset{i=1}{\overset{N}{\&Sigma;}}\left|\frac{Y-\hat{Y}}{Y}\right|\&times;100\%$

In formula:

Δ _mREfor statistic average relative error;

Y is the true value of sample;

for the estimated value of Y;

Step 2025: evenly extract 3/5ths in total sample as training sample, all the other get different C and δ respectively 2/5ths as test samples ², utilize training sample to learn, and calculate the Δ on test samples _mRE, select minimum Δ _mREcorresponding model is as final forecasting model;

Step 203: by wind velocity vector V=[V ₁, V ₂..., V _n] ^tobtaining blower fan maximum power point rotor speed forecast value by final forecasting model is ω _ref.