The present invention will be described in detail below with reference to the drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation mode and specific operation process are given, but the protection scope of the present invention is not limited to the following embodiments.
 This embodiment provides a method for selecting and optimizing wind farm blades, which includes the following steps: 1) Obtain the wind farm unit capacity, and obtain the selectable wind turbine type and its corresponding wind turbine blade length according to the capacity; 2) Take the wind farm The goal is to have the lowest cost per kilowatt-hour throughout the life cycle, and the discrete wind turbine blade length is used as a constraint to establish an objective function; 3) the objective function is solved by genetic algorithm to obtain the optimal wind turbine type combination.
 This method establishes a calculation model for the cost per kilowatt hour of the entire life cycle of the wind farm. Among them, the total cost calculation mainly considers the cost of a single wind turbine, the investment cost of the wind farm, and the operation and maintenance costs related to blade maintenance and replacement. The power generation of the wind farm first calculates the wind speed at the wind turbine according to the wake effect model, and then calculates the power generation at the corresponding wind speed according to the active output model of the wind turbine. The power generation includes the power generation calculation of the wind farm's life cycle of 20-25 years. This method aims at the lowest cost per kilowatt-hour throughout the life cycle of the wind farm, and optimizes the blade length of each wind turbine in the wind farm under the premise of fully considering the wind speed distribution, wind direction distribution and the wake effect between the wind farms. Select and use genetic algorithm to solve the optimization problem.
 The objective function established by this method is:
 Where: C cost Is the total cost of the wind turbine; P total It is the power generation of the entire wind farm in the whole life cycle. R is the blade radius of the unit; R f It is the set of all available blade radii.
 (1) Cost model: The total cost of wind turbines is determined by the initial investment of the project and the annual operation and maintenance costs of the turbines. The initial investment of the project consists of two parts: the cost of wind turbines and the investment of wind farms; the annual operation and maintenance costs only consider the costs related to blade maintenance and replacement. The total cost of wind turbines can be expressed as:
 Where: C W Cost for a single wind turbine; C f Is the wind farm investment cost; C m For operation and maintenance costs.
 The cost of a single wind turbine includes the cost of each component of the unit. The cost of blades, pitch mechanism, low-speed shaft, main bearing, yaw mechanism, nacelle body, and tower are related to the length of the unit blade R and the height of the hub H. w1 ~C w7 Respectively. The cost of the remaining components is only related to the power level of the unit. The present invention adopts the empirical formula of a three-stage gearbox double-fed variable speed wind power generator in the cost calculation:
 C w2 =a 2 (2R) 2.6578 (4)
 C w3 =a 3 (2R) 2.887 (5)
 C w4 =a 4 ((b 4 ×2R-c 4 )×d 4 (2R) 2.5 ) (6)
 C w5 =a 5 (2R) 2.964 (7)
 C w6 =a 6 (2R) 1.953 (8)
 C w7 =a 7 (b 7 R 2 H-c 7 ) (9)
 Among the investment costs of wind farms, the relationship between capital construction and installation costs, blade length and hub height is:
 C f1 =e 1 (πR 2 H) 0.4037 (10)
 C f2 =e 2 (2RH) 1.1736 (11)
 The annual operation and maintenance costs only consider the costs related to blade maintenance and replacement:
 C m =kpC w1 (12)
 In formula (12), k is the full life cycle years; p is the blade failure rate, unit times/year.
 (2) Wind turbine output model: The classic wind turbine active output model is a piecewise function:
 Where: v in , V rate , V out , P rate They are the cut-in wind speed, rated wind speed, cut-out wind speed and rated power of the unit; the values of a, b, and c depend on v in , V rate.
 In order to solve the problem that the active power curve of wind turbines is not very obvious due to different blade lengths under the same capacity wind turbines, the present invention fits the unit power curves according to the power curves of the same series of wind turbines given by wind turbine manufacturers, and then Use the interpolation method to obtain the active power output under the corresponding wind speed.
 (3) Wake model: P total The calculation of Wake effect needs to be considered. The present invention uses the Lissaman model that can unify the calculation of all occlusion and partial occlusion.
 The impact of upstream units on downstream units is as figure 1 Shown.
 Let the initial wind speed be v i , The blade radius of the upstream fan and downstream fan are r i And r j , The height of the hub is and h i , H j , Then the distance from the upstream fan d ij The wake radius at is:
 r d =r i +k w d ij (14)
 Where: when the wind turbine receives natural wind speed, k w =0.04, otherwise k w = 0.08.
 The shielding area of the upstream fan wake effect on the downstream fan is:
 After the wake effect, the downstream fan wind speed v j Calculated as follows:
 Where: A s Is the area of the covered part; A j Sweep the area of the downwind wind wheel; C T Is the wind turbine thrust coefficient; d ij Is the distance of the unit along the wind direction; x ij Is the distance of the unit along the direction perpendicular to the wind direction; Δh is the height difference of the unit hub.
 Since the unit in the downwind direction may be affected by the wake of multiple upstream units, the combined effect of multiple upstream units on the downstream units is equivalent to:
 Where: v 0 Is the wind speed not affected by the wake effect; v j Is the wind speed under the influence of a single fan; N is the total number of units that are jointly affected by the upstream.
 (4) Wind speed and wind direction model: In actual wind farms, wind speed changes continuously with time, which is random and difficult to predict accurately. A lot of statistical data shows that Weibull distribution can describe the characteristics of wind speed throughout the year, and its probability density function is:
 Where: f(v) is the probability density of wind speed v; k is the sex conversion parameter; c is the scale parameter.
 The changing law of wind direction is more stable than wind speed. According to the observation data of the wind tower, the proportion of each main wind direction can be calculated, and the wind direction rose diagram of the wind farm can be obtained. Each radiation line in the rose diagram represents a wind direction, and the length of the radiation line represents the proportion of the wind direction. The wind rose diagram can well describe the year-round changes in wind direction in a wind farm.
 (5) Wind farm life cycle power generation calculation model: n wind direction is obtained from the wind rose diagram, m wind speed is obtained from Weibull distribution, then the active power output of the wind farm in the life cycle is:
 In the formula: k is the full life cycle years; θ is the angle between the wind direction and the Y axis; v is the wind speed; N is the number of units; P w It is the active power output of a unit at a specific wind speed and wind; t(v) is the number of hours in a year when the wind speed is v; p(θ) is the probability that the wind direction is θ.
 In the wake calculation, it can be seen that the wake calculation has nothing to do with the absolute position of the fan, only the relative position of the fan along the wind direction, that is, the horizontal distance and the longitudinal distance between the fans. Once the wind direction changes, the horizontal and vertical distances between the fans in the new wind direction will change. It is complicated and tedious to directly derive the relationship between the wind direction angle and the relative position of the fan. The present invention proposes a simple method for multi-wind direction wake calculation, which can obtain the relative position relationship of the fan by transforming rectangular coordinates and polar coordinates, thereby calculating the wake.
 When the wind direction is shown in Figure (2a), the initial wind speed is v i Then remember that after unit A affects the wake effect of unit B, the wind speed at unit B v j It can be obtained by formula (16). From equation (16), v j The functional relationship with other variables can be expressed as:
 v j =g 1 (v i ,h i ,h j ,r i ,r d ,A s ) (20)
 among them:
 From equations (20) and (21), we can get:
 v j =f(v i ,h i ,h j ,r i ,r j ,x ij ,d ij ) (twenty two)
 Set the rectangular coordinates of unit A as (x i , Y i ), the polar coordinates are (ρ i ,θ i ); The rectangular coordinate of unit B is (x j , Y j ), the polar coordinates are (ρ j ,θ j ). Then use polar coordinates to represent the variable x ij , D ij for:
 When the wind direction becomes as shown in Figure (2b), the wind speed vj at computer group B is the parameter v i , R i , R j , H i , H j No change, only x ij , D ij Changes. Rotate X axis and Y axis to X’, Y’, the polar coordinates of unit A and B will become (ρ i ,θ i +α), (ρ j ,θ j +α), α is. Variable x ij , D ij Can be expressed as:
 Change the polar coordinates to rectangular coordinates, then x ij , D ij Expressed as:
 After the wind direction changes, the wind speed at unit B is:
 v’ j =f(v i ,r i ,r j ,h i ,h j ,x′ ij ,d′ ij ) (26)
 By making the coordinate axis change with the change of wind direction, the relative position of the unit can be fixedly expressed by equation (25), which achieves the purpose of simplifying the calculation of wakes in multiple wind directions.
 (7) Optimization algorithm:
 The optimization of fan blade selection is a nonlinear optimization problem with multiple discrete variables. The present invention uses genetic algorithm to solve this problem, and the algorithm flow is as follows image 3 Shown.
 Take a wind farm containing 57 wind turbines with a rated power of 3MW as an example to implement the above method. The total installed capacity of the wind farm is 171MW, and the row spacing and column spacing of the wind turbines are 500m and 400m. Figure 4 Shown.
 The probability density function of the wind speed distribution can be approximated by the Weibull distribution function. From the Weibull distribution parameter of the wind farm, the probability density of the wind speed in one year can be obtained as Figure 5 Shown. According to the long-term wind measurement data of a wind farm, the present invention obtains the wind direction rose diagram of the wind farm as Image 6 Shown.
 According to wind farm unit capacity requirements, the types of wind turbines available for selection in 3MW units and their related parameters are shown in Table 1. The present invention optimizes the selection of the 4 types of wind power given in Table 1.
 Table 1 Fan parameters
 According to the calculation method proposed by the present invention, the calculation result is as Figure 8-11 Shown. among them, Figure 8 For 57 units of electricity per hour. It can be seen that the objective function eventually tends to converge, and the method proposed by the present invention can effectively improve the cost per kilowatt-hour of the wind farm. Picture 9 It is the comparison data of the cost per kilowatt-hour of the wind farm under the scheme optimized by the method of the present invention and the scheme with a single blade length.
 From Picture 9 It can be seen from the data that the cost per kilowatt-hour of using four single blades are 0.4627, 0.4274, 0.4694, and 0.4753 (yuan/kW·h) respectively. The length of the blade and the cost per kilowatt-hour are wirelessly related, and the optimized method of the present invention realizes The cost per kilowatt-hour of the wind farm is reduced to 0.3868 (yuan/kW·h), which can be reduced by 9.4% compared with the optimal single blade solution. Picture 10 It is the optimal fan blade distribution map. The optimization results show that the choice of blades is related to the wind direction. In the case of multiple dominant wind directions, the general rule of wind turbine blade length is that the dominant wind has short blades in the front row and long blades in the back row. Due to the mutual influence of multiple dominant wind directions and wakes, the unit presents a more complex combination under the overall law. The calculation results of wind farm unit blades under complex terrain conditions are as follows Picture 11 Shown.
 In summary, the present invention comprehensively considers the wind speed distribution, wind direction distribution, wake effects between units and the entire life cycle cost of the wind farm, and proposes a method for optimizing the selection of wind farm blades. This method can reduce the cost per kilowatt-hour of a wind farm under the same wind farm capacity, and provides a basis for the selection of wind turbine blades for onshore and offshore wind farms.