Wind generating set transmission system fault detection method based on finite frequency observer

Abstract

The invention provides a wind generating set transmission system fault detection method based on a finite frequency observer, and the method comprises the steps: firstly, designing an improved observer according to a finite frequency performance index in consideration of the frequency domain characteristics of an actuator stuck fault signal and a suffered wind disturbance signal of a wind generating set transmission system; and the detection performance is superior to that of the observers provided before. Thirdly, the finite frequency performance index required by design is converted into a linear matrix inequality, and observer parameters are set through a convex optimization problem under the constraint condition of the linear matrix inequality; the biggest advantage of improving the observer is that different frequency domain characteristics of a fault signal and an interference signal are considered at the same time, so that the observer is more sensitive to such faults and has relatively strong robustness to external wind disturbance, and the detection performance is improved. Finally, a simulation result shows that the design method has relatively good fault detection performance.

Description

Technical Field

[0001] The invention belongs to the technical field of wind turbine generator set fault detection, and in particular to a wind turbine generator set transmission system fault detection method based on a finite frequency observer. Background Art

[0002] Due to the massive consumption of fossil fuels, the depletion of natural resources and the deterioration of the environment, the development and utilization of clean and renewable energy is the only way to achieve sustainable economic development. Wind energy has attracted global attention because it is recyclable and does not pollute the environment. However, most wind turbines are installed in unpredictable weather and harsh conditions such as the western desert or coastal areas, which will lead to frequent failures of wind turbines, which will affect the safe operation of wind turbines and even cause downtime, resulting in economic losses and energy waste. As a key component of wind turbines, the transmission system of wind turbines is easily damaged by external environmental factors such as lightning, gusts, and hail. Therefore, the design of effective fault detection technology can detect and warn the faulty units as early as possible and take effective intervention measures to improve the reliability of the entire wind farm.

[0003] Model-based fault detection technology is used for fault detection of various devices, components and sensors. The general steps are: first, use the system model to reconstruct the state of the controlled object and connect it in parallel with the actual system; then, compare the estimated value generated by the model with the actual measured value, and construct a residual discriminant function; finally, compare the output result of the residual function with the set threshold, and detect the fault. However, unknown noise and external interference will affect the accuracy of fault detection, resulting in frequent false alarms. Therefore, robust fault observers have received great attention. For example, H ∞ / H - The fault observer detects the fault. - Performance indicators are used to describe the sensitivity of the detection system to fault signals. The indicator H - The higher it is, the smaller the fault signal will be detected. Frequency domain H ∞ The performance index is used to evaluate the system's resistance to external interference during fault detection. ∞ The higher the index, the higher the reliability of fault detection.

[0004] This patent designs a finite frequency observer to detect faults based on the finite frequency characteristics of external wind disturbance and transmission system actuator jamming faults. The output of the observer and the actual output are used to generate the residual of the transfer function and compared with the set threshold to determine whether the fault has occurred. More precisely, on the one hand, H - The finite frequency performance index is used to represent the sensitivity of the residual to the fault in the low frequency domain. On the other hand, H ∞The finite frequency performance index is used to represent the performance of the residual in the mid-frequency domain against external disturbances. In addition, the finite performance index is converted into a linear matrix inequality, and the parameters of the designed observer are solved through a convex optimization problem. The simulation experiment shows that the fault detection method of the finite frequency observer can effectively improve the sensitivity of the detection performance and enhance its robustness when detecting the transmission system of the wind turbine generator. Summary of the invention

[0005] To solve the above technical problems, the present invention proposes a wind turbine transmission system fault detection method based on a finite frequency observer, which can enable the microgrid system to quickly cut off the faulty wind turbine, reduce the damage of the fault to the unit itself, and improve the reliability of the system.

[0006] To achieve the above object, the technical solution adopted by the present invention is:

[0007] The wind turbine transmission system fault detection method based on finite frequency observer has the following specific steps:

[0008] Step 1, according to the structural characteristics of the wind turbine transmission system, construct the state space equation of the wind turbine transmission system model with faults and wind disturbances;

[0009] Step 2, performing finite frequency domain analysis on the wind disturbance signal and the fault signal;

[0010] Step 3, design a fault observer based on the transmission system model, obtain the augmented state space model of the system, and connect it in parallel with the actual system;

[0011] Step 4, construct the fault detection problem, and use the linear matrix inequality technique to transform the design problem of the finite frequency fault observer into an LMI convex optimization problem, so as to adjust the observer parameters;

[0012] Step 5: construct a residual discriminant function and set a threshold, use the threshold logic rule to detect the fault, and isolate and shut down the faulty unit.

[0013] As a further improvement of the present invention, the specific process of step 1 is as follows:

[0014] Step 1.1, construct the low-speed shaft inertia dynamic equation:

[0015]

[0016] Among them, J r is the moment of inertia of the rotor side, kg·m 2 ;ω r (t) is the wind wheel speed, rad / s; counter torque T d (t) and aerodynamic torque Ta The expression of (t) is:

[0017] T d (t) = k d (θ r (t)-θ g (t))+C d (ω r (t)-ω g (t))

[0018]

[0019] Among them, θ r (t) and θ g (t) is the azimuth angle between the wind rotor and the generator, rad; ρ is the air flow density, kg·m -3 ; v(t) is wind speed, m / s; R is blade radius, m, β(t) is pitch angle, °; C p (λ(t),β(t)) is the wind energy utilization coefficient; λ(t) is the tip speed ratio; C d is the transmission system damping, Nm·(rad·s) -1 ;k d is the transmission chain stiffness, Nm·rad -1 ;

[0020] Step 1.2, wind wheel aerodynamic torque T a (t) is the wind speed v(t) and the wind rotor speed ω r (t) and the nonlinear function related to the pitch angle β(t), which is linearized at the working point (v0, ω0, β0) as follows:

[0021] T a (t) = T a0 +ΔT a =T a0 +χΔv(t)+μΔω r (t)+δΔβ(t)

[0022] Among them, v0, ω0, β0 are the wind speed, rotor speed, and pitch angle in the equilibrium state, Δv(t), Δω r (t), Δβ(t) are the changes of wind speed, rotor speed and pitch angle relative to the working point v0, ω0, β0, and

[0023] Step 1.3, high-speed shaft generator side, the dynamic equation is:

[0024]

[0025] Among them, J g is the generator moment of inertia, kg·m2 ;ω g (t) is the generator speed, rad / s; T g (t) is the generator torque, N·m. The above equation is linearized at the equilibrium point as follows:

[0026]

[0027] Where, ΔT e (t) is the change in generator torque, N·m;

[0028] Step 1.4, define the following variables:

[0029]

[0030] According to the above steps, the state space model equation of the wind turbine transmission system is obtained as follows:

[0031]

[0032] in;

[0033]

[0034] As a further improvement of the present invention, the specific process of step 2 is as follows:

[0035] When the actuator stuck fault f(t) occurs in the transmission system of the wind turbine generator system, the state space model of the system under the fault condition can be written as:

[0036]

[0037] Among them, B2, B3, B4 are matrices of appropriate dimensions;

[0038] The spectral characteristics of the fault signal p(t) and the wind disturbance signal f(t) are analyzed. The fault signal is regarded as a typical step signal and is defined in the low-frequency domain, while the wind disturbance signal is regarded as a disturbance signal and is defined in the medium-frequency domain. The finite frequency ranges that satisfy the fault signal and the wind disturbance signal are defined respectively as follows:

[0039]

[0040] Among them, ω2>0 and ω1>0 are the upper and lower bounds of the wind disturbance frequency domain, respectively. f >0 is the upper bound of the fault frequency domain.

[0041] As a further improvement of the present invention, the specific process of step 3 is as follows:

[0042] Design the following fault observer:

[0043]

[0044] in, is the state vector of the observer, is the output vector of the observer, r(t) is the residual signal, L and N are the observer parameters that need to be designed, and then the augmented state space equation is

[0045]

[0046] in;

[0047]

[0048] In addition, the transfer function from the fault signal f(t) to the residual signal r(t) is G rf (s), the transfer function from the disturbance signal p(t) to the residual signal r(t) is G rp (s), expressed as:

[0049]

[0050] As a further improvement of the present invention, the specific process of step 4 is as follows:

[0051] Step 4.1, construct the fault detection problem of the wind turbine transmission system: design the fault observer in step 3 so that the augmented system is asymptotically stable and for given scalars α>0 and γ>0, H in the intermediate frequency domain ∞ Performance indicators and low-frequency H - The performances satisfy the following inequalities respectively;

[0052]

[0053] Then the fault can be detected using the designed observer, where G rf (s) finite frequency H - Performance indicators, G rp (s) finite frequency H ∞ Performance indicators are defined as follows:

[0054]

[0055]

[0056] Among them, σ(·) represents the minimum singular value of (·), represents the maximum singular value of (·);

[0057] Step 4.2, the fault detection problem of the wind turbine transmission system is converted into a convex optimization problem with LMI constraints and solved. For a given positive scalar γ, the fault observer parameters L and N are determined so that the positive scalar α is maximized, and the mathematical expression is as follows;

[0058]

[0059] Then, the observer parameter L is solved by the following relationship;

[0060] L=F3 -T Z T

[0061] Among them, Θ1, Θ2, Θ3 are represented as follows:

[0062]

[0063]

[0064]

[0065] Where * is a matrix block obtained from the symmetry of the matrix.

[0066] As a further improvement of the present invention, the specific process of step 5 is as follows:

[0067] Select the residual discriminant function J r (t) and threshold J l To detect the fault, J r (t) and J l They are

[0068]

[0069] Among them, the threshold J l It can be obtained empirically and fault detection is performed through threshold logic rules, which can be expressed as

[0070]

[0071] If a transmission system failure occurs, the faulty unit is detected through the above threshold logic rules.

[0072] Compared with the prior art, the above technical solution adopted by the present invention has the following technical effects:

[0073] 1. Aiming at the problem that it is difficult to detect faults in the transmission system of a wind turbine generator set under external wind disturbance, the present invention designs a fault detection method of a finite frequency observer. The biggest advantage of the improved observer is that it takes into account the different frequency domain characteristics of the fault signal and the wind disturbance signal at the same time, making the observer more sensitive to faults and having strong robustness to external wind disturbances, which greatly improves the detection performance.

[0074] 2. This patent provides a new fault detection method based on a finite frequency observer, and utilizes the frequency domain characteristics of wind disturbance signals and fault signals from the perspective of the finite frequency domain of the signal, so that the output of its residual function can reflect the detection sensitivity, so as to better improve the detection performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 It is a structural diagram of the transmission system of the wind turbine generator set of the present invention;

[0076] Figure 2 It is a fault detection logic rule diagram of the wind turbine transmission system;

[0077] Figure 3 It simulates the actual wind disturbance signal;

[0078] Figure 4 It simulates the actual fault signal;

[0079] Figure 5 are the actual and estimated values ​​of the generator speed variation under no fault conditions;

[0080] Figure 6 is the residual discriminant function J of the wind turbine transmission system under wind disturbance r (t) curve;

[0081] Figure 7 It is a comparison diagram of the residual r(t) output curves of the finite frequency and full frequency fault observers;

[0082] Figure 8 It is the residual r(t) output curve under different γ;

[0083] Fig. 9 is the residual discriminant function J under different γ r (t) output curve;

[0084] Fig.10 is the different fault frequencies Ω f The residual r(t) output curve under ; DETAILED DESCRIPTION

[0085] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0086] The present invention provides a fault detection method for a wind turbine transmission system based on a finite frequency observer, comprising the following steps:

[0087] Step 1: Modeling the wind turbine drive train

[0088] Step 101, the low speed shaft inertia dynamic equation is:

[0089]

[0090] Among them, J r is the moment of inertia of the wind rotor, kg·m 2 ;ω r (t) is the wind wheel speed, rad / s; counter torque T d (t) and aerodynamic torque T a The expression of (t) is

[0091] T d (t) = k d (θ r (t)-θ g (t))+C d (ω r (t)-ω g (t)) (2)

[0092]

[0093] Among them, θ r (t) and θ g (t) is the azimuth angle between the wind rotor and the generator, rad; ρ is the air flow density, kg·m -3 ; v(t) is wind speed, m / s; R is blade radius, m, β(t) is pitch angle, °; C p (λ(t),β(t)) is the wind energy utilization coefficient; λ(t) is the tip speed ratio; C d is the transmission system damping, Nm·(rad·s) -1 ;k d is the transmission chain stiffness, Nm·rad -1 .

[0094] Step 102: wind wheel aerodynamic torque T a (t) is the wind speed v(t) and the wind rotor speed ω r (t) and the nonlinear function related to the pitch angle β(t), which is linearized at the operating point (v0, ω0, β0) as

[0095]

[0096] in, v0, ω0, β0 are the wind speed, rotor speed, and pitch angle in equilibrium, Δv(t)=v(t)-v0, Δω r (t) = ω r (t)-ω r0 , Δβ(t)=β(t)-β0 is the change of wind speed, wind rotor speed and pitch angle relative to the working point.

[0097] Step 103, rewrite equation (2) into a perturbation form

[0098] ΔT d (t) = k d (Δθ r (t)-Δθ g (t))+C d (Δω r (t)-Δω g (t)) (5)

[0099] Therefore, equation (1) can be written as follows

[0100]

[0101] Since the operating point is in steady state, the rotor angular acceleration is 0, and T a0 =T d0 , formula (6) can be converted to

[0102]

[0103] Step 104, high-speed shaft generator side, the dynamic equation is

[0104]

[0105] Among them, J g is the generator moment of inertia, kg·m 2 ,ω g (t) is the generator speed, rad / s; T g (t) is the generator torque, N·m.

[0106] Similarly, the generator equation (8) can be linearized at its equilibrium point as follows:

[0107]

[0108] Where, ΔT e (t) is the change in generator torque, N·m.

[0109] Step 104, define the following state variables

[0110]

[0111] The state space model equation of the wind turbine transmission system is obtained as follows:

[0112]

[0113] in

[0114]

[0115] Step 105: When the actuator stuck fault f(t) occurs in the wind turbine transmission system, the state space model of the system under the fault condition can be written as

[0116]

[0117] Among them, B2, B3, B4 are known matrices with appropriate dimensions.

[0118] Step 2: When a fault occurs in the transmission system of a wind turbine generator set, a finite frequency observer is designed to detect the fault based on the finite frequency domain performance of the wind disturbance signal and the fault signal. This includes the following steps:

[0119] Step 201, based on the spectral characteristics of the fault signal p(t) and the wind disturbance signal f(t), we define the finite frequency ranges that satisfy the fault signal and the wind disturbance signal, respectively, as shown below:

[0120]

[0121] Among them, ω2>0 and ω1>0 are the upper and lower bounds of the finite frequency domain range of the wind disturbance signal, respectively. f > 0 is the upper limit of the fault signal. Therefore, in actual engineering, the appropriate finite frequency domain range can be selected according to the frequency spectrum of the fault signal and the power spectrum of the pulsating wind.

[0122] Step 202, the fault observer is designed as follows

[0123]

[0124] in, is the state vector of the observer, is the output vector of the observer, r(t) is the residual signal, and L and N are the observer parameters to be designed.

[0125] Based on the designed observer and the state space equation of the transmission system, the augmented state space model is obtained as follows:

[0126]

[0127] in

[0128]

[0129] Step 3: Troubleshooting

[0130] Design the fault observer (11) so that the augmented system is asymptotically stable and for given scalars α>0 and γ>0, H in the intermediate frequency domain ∞ Performance indicators and low-frequency H - The performance satisfies the following inequalities:

[0131]

[0132] Then the fault can be detected using the designed observer. G rf (s) finite frequency H - Performance indicators, G rp (s) finite frequency H ∞ Performance indicators, respectively defined

[0133]

[0134] Among them, σ(·) represents the minimum singular value of (·), represents the largest singular value of (·).

[0135] Step 4: Transformation of the fault problem design

[0136] The wind turbine transmission system fault detection problem can be converted into a convex optimization problem of linear matrix inequality, which is expressed as follows:

[0137] For a given positive scalar γ, the fault detection observer parameters L, N are determined so that the positive scalar α is maximized, and the mathematical expression is as follows.

[0138]

[0139] Then, the observer parameter L can be solved by the following relationship.

[0140] L=F3 -T Z T

[0141] Among them, Θ1, Θ2, Θ3 are represented as follows

[0142]

[0143]

[0144]

[0145] Among them, * is the matrix block obtained by the symmetry of the matrix.

[0146] Step 401: H in the low frequency domain - Equivalent transformation of performance index inequality and linear matrix inequality Θ1.

[0147] After simple algebraic operations, the linear matrix inequality Θ1 can be written as follows

[0148]

[0149] To simplify, we define the following variables

[0150]

[0151] And, Θ 12 The following conversions are possible

[0152]

[0153] Then, the inequality Θ1 can be transformed as follows

[0154] λΞ f λ T +RΠ f R T +He(η[F 0]σ)<0 (15)

[0155] According to the projection theorem, if the above inequality (15) exists, then and only if the following inequality holds.

[0156]

[0157] in

[0158] K=[I 0 0] T ,

[0159] After calculation, the second inequality of formula (16) is equivalent to

[0160]

[0161] Among them, f and Π f They are respectively expressed as follows

[0162]

[0163] According to the GKYP lemma, we can get that equation (17) is equivalent to the following frequency domain inequality:

[0164]

[0165] After simple calculation, the above inequality is equivalent to This is consistent with the H - Performance Inequality same.

[0166] Step 402: for the H in the intermediate frequency domain ∞ Equivalent transformation of performance index inequality and linear matrix inequality Θ2.

[0167] Similarly, using the same method as above, it can be proved that the linear matrix inequality Θ2 is equivalent to H in the intermediate frequency domain ∞ Performance Inequality

[0168] Step 403, equivalent transformation of the linear matrix inequality θ3 and the stability condition:

[0169] After simple algebraic operations, the linear matrix inequality Θ3 can be transformed into

[0170]

[0171] because and [I -I][II] T =0, so we can get the following formula.

[0172]

[0173] Through the projection theorem, we can conclude that if the above inequality holds, then the inequality holds, so the system is asymptotically stable.

[0174] Step 5: Output of fault detection results

[0175] In order to detect the fault of the wind turbine transmission system, the residual discriminant function J is selected respectively. r (t) and threshold J l It is expressed as follows

[0176]

[0177] Among them, the threshold J l It can be obtained empirically and fault detection is performed through threshold logic rules, which can be expressed as

[0178]

[0179] If a transmission system failure occurs, the faulty unit can be detected using the above threshold logic rules.

[0180] In order to better illustrate the invention, a system model is established by Matlab and illustrated with the accompanying drawings.

[0181] The present invention uses the data of a 1.5MW unit provided by a coastal wind farm, and the parameters are R = 61.5 / m, ρ = 1.225 / kg·m -3 , J g =534.116 / kg·m 2 , J r =3.875×10 7 / kg·m 2 , C d =6.215×10 6 / Nm·(rad·s) -1 , k d =8.676×10 8 / Nm·rad -1 , the wind turbine generator set is at the operating point v0=18 / ms, ω r0 =12.1rad / s, linearized at β0=14.7°, the frequency domain of wind disturbance is 0.6Hz<ω d <1.3Hz, the range of fault frequency domain is |ω|<0.5Hz, select parameter γ=3, and adjust the observer parameters.

[0182] When γ = 3, solving the optimization problem (14), we get the optimal α = 0.29 and the observer parameters L, N

[0183]

[0184] Figure 1 is the structure diagram of the wind turbine transmission system. a (t) is the aerodynamic torque of the wind rotor, N·m; T d (t) is the counter torque, N·m; ω r (t) is the wind wheel speed, rad / s; ω g (t) are the generator speed, rad / s; J r is the moment of inertia of the rotor side, kg·m 2 ; J g is the moment of inertia of the generator side, kg·m 2 ; C d is the torsional damping of the transmission system, Nm·(rad·s) -1 ;k d is the transmission chain stiffness, Nm·rad -1. The transmission system is the core component of the wind turbine that connects the generator and the blade main shaft. Its main function is to transmit the mechanical energy generated by the hub to the generator so that the generator can operate normally. Its working principle is: first, the pitch angle of the wind rotor blades is adjusted by the rotation of the hub, so that the windward surface of the wind rotor is perpendicular to the airflow direction, which is convenient for absorbing wind energy. When the external wind speed reaches the starting wind speed of the wind turbine, the blades are subjected to force and cause the wind rotor to rotate. Secondly, through the action of the main shaft and gearbox in the transmission system, the mechanical torque is transferred from the low-speed shaft to the high-speed shaft, and finally the high-speed shaft is connected to the generator, thereby completing the conversion of mechanical energy into electrical energy.

[0185] Figure 2 The figure is a fault detection logic rule diagram for the wind turbine transmission system. According to the structural characteristics of WTTS, the main steps are: (1) construct the WTTS state space equation with faults and wind disturbances; (2) perform finite frequency domain analysis on wind disturbance signals and fault signals, and design a fault observer based on the WTTS model, and connect it in parallel with the actual system; (3) use LMI technology to convert the design problem of the finite frequency fault observer into an LMI convex optimization problem to solve the observer; (4) construct a residual discriminant function and set a threshold, use the threshold logic rule to detect faults, and isolate and shut down the faulty unit.

[0186] Figure 3 and Figure 4 They are simulated wind disturbance signals and fault signals respectively. We choose to use Davenport's wind speed spectrum that is constant along the height, and select the wind disturbance frequency domain as 0.6Hz<ω d <1.3Hz, simulate the pulsating wind speed, such as Figure 3 In addition, when the rolling bearing of WTTS is suddenly stuck, the frequency domain where the fault occurs is selected as |ω|<0.5Hz, and the actuator stuck fault signal f(t) is simulated, as shown in Figure 4 shown.

[0187] Figure 5 are the actual value and estimated value of the generator speed change under no fault conditions. Figure 5 As shown in the figure, the actual value of the generator speed change affected by wind disturbance is the black solid line, and the estimated value of the observer is the red dotted line. It is easy to see that in the absence of faults, the estimated value can better reflect the actual value.

[0188] Figure 6 is the residual discriminant function curve of the wind turbine transmission system under wind disturbance. Its output process is to use formula (18) and the residual discriminant function J r (t) The expression simulates the output curve and sets the threshold value to J l =0.1. Figure 6It can be seen that when the actuator stuck fault occurs in the system after 8 seconds, the residual discriminant function exceeds the set threshold, so the fault can be easily detected quickly. At this time, the system needs to remove or shut down the faulty unit.

[0189] Figure 7 is a comparison of the residual output curves of the finite-frequency and full-frequency fault observers. As we all know, most of the WTTS fault observers are designed at full frequency. In order to illustrate the advantage of the proposed finite-frequency observer in terms of fault sensitivity, we set Q1 = 0, Q2 = 0 in LMIΘ1, LMIΘ2, LMIΘ3 and solve the optimization problem (14) to obtain the observer in the full-frequency sense. Figure 7 The comparison of residual output curves of finite frequency and full frequency fault observers is given. It is not difficult to see that when the fault occurs at t≥8s, the finite frequency observer is more sensitive to the fault than the full frequency observer.

[0190] Figure 8 and Fig. 9 They are the residual r(t) output graph and residual discriminant function J under different γ respectively. r (t) Output graph. Fault sensitivity is an important indicator to measure fault detection performance. Designing an observer that is sensitive to faults can detect minor faults earlier and give early warnings for faulty units. Therefore, different finite frequency H ∞ The performance index γ affects the optimal value of α and affects the sensitivity of fault detection. We choose different γ∈{1.7,3,5.2}, solve the optimization problem (14), obtain different optimal values ​​α, and different observer parameters L and N, and plot the residual discriminant function J under different γ values. r (t), such as Figure 8 As shown in Figure 2. As the value of γ increases, the optimal value of α also increases, and the residual response is stronger. This means that the designed fault observer is more sensitive to faults, that is, the detection performance is more sensitive. However, Fig. 9 When γ = 5.2, the residual discriminant function J r (t) exceeds the set threshold J within 8 seconds without failure l , that is, a fault alarm is issued at this time. Therefore, it is very important to set a reasonable γ value, so it is more appropriate to choose γ = 3.

[0191] Fig.10 For different fault frequencies Ω f In order to try to reveal different fault frequency intervals Ω f Will it affect the sensitivity of fault detection? When γ=3, different Ω f , solve the optimization problem (14) to obtain different Ω fThe optimal α value under different observer parameters L and N, and plot different Ω f The residual r(t) under Fig.10 It is not difficult to see that as the fault finite frequency range becomes narrower, the optimal α becomes larger and the detection performance becomes more sensitive. Therefore, in practice, a narrower finite frequency domain fault is selected as much as possible.

[0192] The above description is only a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent change made based on the technical essence of the present invention still falls within the scope of protection required by the present invention.

Claims

1. A wind turbine generator transmission system fault detection method based on a finite frequency observer, the specific steps are as follows, characterized in that: Step 1, according to the structural characteristics of the wind turbine transmission system, construct the state space equation of the wind turbine transmission system model with faults and wind disturbances; Step 2, performing finite frequency domain analysis on the wind disturbance signal and the fault signal; Step 3, design a fault observer based on the transmission system model, obtain the augmented state space model of the system, and connect it in parallel with the actual system; Step 4, construct the fault detection problem, and use the linear matrix inequality technique to transform the design problem of the finite frequency fault observer into an LMI convex optimization problem, so as to adjust the observer parameters; Step 5: construct a residual discriminant function and set a threshold, use the threshold logic rule to detect the fault, and isolate and shut down the faulty unit.

2. The wind turbine generator transmission system fault detection method based on finite frequency observer according to claim 1 is characterized in that: The specific process of step 1 is as follows: Step 1.1, construct the low-speed shaft inertia dynamic equation: Among them, J r is the moment of inertia of the rotor side, kg·m 2 ;ω r (t) is the wind wheel speed, rad / s; counter torque T d (t) and aerodynamic torque T a The expression of (t) is: T d (t)=k d (i r (t)-θ g (t))+C d (oh r (t)-ω g (t)) Among them, θ r (t) and θ g (t) is the azimuth angle between the wind rotor and the generator, rad; ρ is the air flow density, kg·m -3 ; v(t) is wind speed, m / s; R is blade radius, m, β(t) is pitch angle, °; C p (λ(t),β(t)) is the wind energy utilization coefficient; λ(t) is the tip speed ratio; C d is the transmission system damping, Nm·(rad·s) -1 ;k d is the transmission chain stiffness, Nm·rad -1 ; Step 1.2, wind wheel aerodynamic torque T a (t) is the wind speed v(t) and the wind rotor speed ω r (t) and the nonlinear function related to the pitch angle β(t), which is linearized at the working point (v0, ω0, β0) as follows: T a (t)=T a0 +ΔT a =T a0 +xΔv(t)+μΔω r (t)+δΔβ(t) Among them, v0, ω0, β0 are the wind speed, rotor speed, and pitch angle in the equilibrium state, Δv(t), Δω r (t), Δβ(t) are the changes of wind speed, rotor speed and pitch angle relative to the working points v0, ω0, β0, and Step 1.3, high-speed shaft generator side, the dynamic equation is: Among them, J g is the generator moment of inertia, kg·m 2 ;ω g (t) is the generator speed, rad / s; T g (t) is the generator torque, N·m. The above equation is linearized at the equilibrium point as follows: Where, ΔT e (t) is the change in generator torque, N·m; Step 1.4, define the following variables: According to the above steps, the state space model equation of the wind turbine transmission system is obtained as follows: in; 3. The wind turbine generator transmission system fault detection method based on finite frequency observer according to claim 1 is characterized in that: The specific process of step 2 is: When the actuator stuck fault f(t) occurs in the transmission system of the wind turbine generator system, the state space model of the system under the fault condition can be written as: Among them, B2, B3, B4 are matrices of appropriate dimensions; The spectral characteristics of the fault signal p(t) and the wind disturbance signal f(t) are analyzed. The fault signal is regarded as a typical step signal and is defined in the low-frequency domain, while the wind disturbance signal is regarded as a disturbance signal and is defined in the medium-frequency domain. The finite frequency ranges that satisfy the fault signal and the wind disturbance signal are defined respectively as follows: Among them, ω2>0 and ω1>0 are the upper and lower bounds of the wind disturbance frequency domain, respectively. f >0 is the upper bound of the fault frequency domain.

4. The method for detecting faults in a transmission system of a wind turbine generator set based on a finite frequency observer according to claim 1, characterized in that: The specific process of step 3 is as follows: Design the following fault observer: in, is the state vector of the observer, is the output vector of the observer, r(t) is the residual signal, L and N are the observer parameters that need to be designed, and then the augmented state space equation is in; In addition, the transfer function from the fault signal f(t) to the residual signal r(t) is G rf (s), the transfer function from the disturbance signal p(t) to the residual signal r(t) is G rp (s), expressed as:

5. The wind turbine generator transmission system fault detection method based on finite frequency observer according to claim 1, characterized in that: The specific process of step 4 is as follows: Step 4.1, construct the fault detection problem of the wind turbine transmission system: design the fault observer in step 3 so that the augmented system is asymptotically stable and for given scalars α>0 and γ>0, H in the intermediate frequency domain ∞ Performance indicators and low-frequency H - The performances satisfy the following inequalities respectively; Then the fault can be detected using the designed observer, where G rf (s) finite frequency H - Performance indicators, G rp (s) finite frequency H ∞ Performance indicators are defined as follows: Among them, σ(·) represents the minimum singular value of (·), represents the maximum singular value of (·); Step 4.2, the fault detection problem of the wind turbine transmission system is converted into a convex optimization problem with LMI constraints and solved. For a given positive scalar γ, the fault observer parameters L and N are determined so that the positive scalar α is maximized, and the mathematical expression is as follows; s.t.LMIΘ1,LMIΘ2,LMIΘ3 Then, the observer parameter L is solved by the following relationship; <h2 style=";text-align:left;direction:ltr">L=F3<h2 style=";text-align:left;direction:ltr"> -T <h2 style=";text-align:left;direction:ltr"> Z<h2 style=";text-align:left;direction:ltr"> T Among them, Θ1, Θ2, Θ3 are expressed as follows: Where * is a matrix block obtained from the symmetry of the matrix.

6. The method for detecting faults in a transmission system of a wind turbine generator set based on a finite frequency observer according to claim 1, characterized in that: The specific process of step 5 is as follows: Select the residual discriminant function J r (t) and threshold J l Detect the fault, J r (t) and J l They are Among them, the threshold J l It can be obtained empirically and fault detection is performed through threshold logic rules, which can be expressed as If a transmission system failure occurs, the faulty unit is detected through the above threshold logic rules.

Images