Fault detection method for doubly-fed wind turbine generator based on event-triggered sliding mode observer based on particle swarm optimization

Abstract

This paper invents a fault detection method for a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm, comprising the following steps: constructing a rotor current sliding mode observer of the wind turbine; introducing a particle swarm algorithm to optimize the parameters of the constructed sliding mode observer; introducing an event-triggered control mechanism to control the constructed sliding mode observer; obtaining the maximum tracking residual value of the wind turbine under the undisturbed working state by subtracting the rotor current output value of the wind turbine under the fault-free state from the sliding mode observation value; introducing a rotor current sensor fault, a grid voltage drop fault, and a stator inter-turn short circuit fault, and using the residual of the current output value and the sliding mode observation value to realize the fault detection of the wind turbine; This paper designs a fault detection method for a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm, which has the advantages of simple design, strong anti-interference to unknown signals, reduced computing and communication resources, and weakened sliding mode jitter.

Description

Technical Field

[0001] The invention relates to a fault detection method for a double-fed wind generator based on an event-triggered sliding mode observer of a particle swarm algorithm, and belongs to the field of electrical control. Background Art

[0002] The energy crisis in today's society is becoming increasingly serious. The development and utilization of renewable energy such as wind energy and solar energy has become the focus of global attention. Among them, wind energy has an overwhelming advantage in global renewable energy. However, although wind power generation systems do not require additional energy costs, they are prone to failure due to their harsh working environment. Therefore, studying a technology that can detect and suppress faults in sliding mode state variables is of great significance for increasing the service life of motors and improving the economic benefits of wind power generation systems.

[0003] Since doubly-fed wind turbines work in harsh environments such as the field, thunderstorms, typhoons, and hail for a long time, they are affected by temperature changes, mechanical stress, and electrical stress, and many of their main parts are prone to electrical and mechanical failures. Fault detection and maintenance of key components of doubly-fed wind turbines have always been of great importance. However, with the large-scale development of wind turbines from land to sea and the increasing capacity of single units, the maintenance cost of wind turbines has greatly increased, and at the same time, fault shutdown has caused great damage to the stability of the power grid. At this time, traditional fault detection and maintenance methods are difficult to meet the requirements. The ideal wind power fault detection device needs to be more sensitive to the early faults of wind turbines, and can prevent the spread of faults through appropriate control methods, avoid temporary shutdowns, and facilitate the operation and maintenance personnel to arrange maintenance plans reasonably. As a key component of a wind turbine, the generator has a high fault shutdown rate, so its early fault detection is extremely important. However, due to the characteristics of small amplitude, non-stationary, and easily affected by working conditions, the early faults of the motor have always lacked reliable online detection methods for the early faults of doubly-fed wind turbines, resulting in many wind turbines "running with illness" and eventually causing more serious shutdown failures. In summary, seeking more effective online detection methods for early fan failures is of great practical significance for avoiding fan shutdowns caused by early fan failures. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a fault detection method for a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm. The present invention accurately monitors the operating status of the doubly-fed wind turbine generator, extracts the fault information of the system according to the tracking error between the rotor current observation value and the rotor current output value of the proposed event-triggered sliding mode observer based on the particle swarm algorithm, and successfully detects the rotor current sensor fault, grid voltage drop fault and stator turn-to-turn fault of the doubly-fed wind turbine generator.

[0005] The present invention proposes a fault detection method for a doubly-fed wind turbine generator based on an event-triggered sliding mode observer using a particle swarm algorithm, which mainly includes the following steps:

[0006] Step 1: Construct a rotor current sliding mode observer for a doubly-fed wind turbine generator;

[0007] Step 2: Introduce the particle swarm algorithm to optimize the parameters for constructing the sliding mode observer;

[0008] Step 3: Introduce an event-triggered control mechanism to control the constructed sliding mode observer;

[0009] Step 4: by subtracting the rotor current output value of the doubly-fed wind turbine generator in a fault-free state from the sliding mode observation value, the maximum tracking residual value of the doubly-fed wind turbine generator in an undisturbed working state is obtained;

[0010] Step 5: Introduce the rotor current sensor fault, and compare the residual between the rotor current output value and the sliding mode observation value with the maximum tracking residual value under the fault-free state, so as to realize the fault detection of the doubly fed wind turbine generator;

[0011] Step 6: Introduce a grid voltage drop fault, and compare the residual between the rotor current output value and the sliding mode observation value with the maximum tracking residual value under a fault-free state, thereby realizing fault detection of the doubly-fed wind turbine generator;

[0012] Step 7: Introduce a stator turn-to-turn short-circuit fault, and compare the residual between the rotor current output value and the sliding mode observation value with the maximum tracking residual value in the fault-free state, thereby realizing fault detection of the doubly fed wind turbine generator.

[0013] In the above-mentioned method for fault detection of a doubly-fed wind turbine generator using an event-triggered sliding mode observer based on a particle swarm algorithm, the expression of the sliding mode observer in step 1 is:

[0014]

[0015] According to the rotor current state equation of the doubly-fed wind turbine, a sliding mode observer is constructed and the sliding mode approach rate is introduced. In the formula Δ is the residual between the sliding mode estimation value and the system output value, ξ is the maximum steady-state residual critical value that the system can accept, k is the gain parameter of the constant speed approach rate part, η is the parameter of the exponential switching term, β is the parameter of the switching term, ε and δ0 are the switching function gain parameters, α is a positive parameter, s represents the sliding mode switching surface and c is the parameter of the sliding mode switching surface. The method of obtaining the rotor current sliding mode observer model based on the sliding mode approach rate through the sliding mode variable structure control theory is:

[0016] The rotor voltage equation of the system is:

[0017]

[0018] The stator voltage equation of the system is:

[0019]

[0020] The flux equation of the system is:

[0021]

[0022] In the formula, u ds ,u qs ,u dr ,u qr are the voltage vectors of the stator and rotor of the d and q axes respectively, i ds ,i qs ,i dr ,i qr are the stator and rotor current vectors of the d and q axes respectively, ψd s， ψ qs , ψ dr , ψ qr are the stator and rotor magnetic flux vectors of the d and q axes respectively, R S , R r are stator resistance and rotor resistance respectively, L s ,L r are the stator self-inductance and rotor self-inductance, L m is the mutual inductance between the stator and the rotor, L σs ,L σr are the leakage inductance of the stator and rotor respectively, ω r ,ω s ,ω f are the rotor speed, stator speed and slip speed respectively, the d-axis and q-axis current components are used as state variables, and the d-axis and q-axis voltage components are used as input quantities. The state space equation of DFIG is:

[0023]

[0024] in represents the leakage inductance coefficient;

[0025] Considering the state equation of the rotor current, the state equation is:

[0026]

[0027] where i s =[i ds iq s ] T and u=[u ds u qs u dr uqr ] T is the input variable of the system, y is the output vector of the system, and C is the output matrix of the system, where:

[0028]

[0029]

[0030]

[0031] Using the stator voltage oriented control method, the method to find the relationship between the stator current term and the rotor current term is:

[0032] The present invention ignores the stator resistance, and there is no difference between the stator flux orientation and the stator voltage orientation. Ignoring the stator resistance, the equation satisfied after adopting the voltage and flux orientation is:

[0033]

[0034] ψ s is the amplitude of the stator flux;

[0035] The relationship between the stator current term and the rotor current term is established as:

[0036]

[0037] Where ψ s is the stator flux

[0038] The sliding mode observer model is constructed based on the rotor current state equation:

[0039]

[0040] in

[0041]

[0042] is the observed value of the current, Design the control law for the sliding mode observer; Combining sliding mode variable structure control theory, design sliding mode control law for:

[0043]

[0044] Substituting the sliding mode control law into the rotor current sliding mode observer model, we can obtain the sliding mode observer based on the approach rate:

[0045]

[0046] Among them, the fitness function expression of the particle swarm algorithm in step 2 is:

[0047]

[0048] C1 and C2 are weight parameters, and C1+C2=1, u q and u d It is the sliding mode control law of the rotor current. The parameters of the sliding mode observer and the sliding mode switching surface are optimized by the particle swarm algorithm to determine the sliding mode observer parameters k, α, β, ε, δ0, η, c so that the fitness function reaches the target value.

[0049] The fault detection method of the doubly-fed wind turbine generator based on the event-triggered sliding mode observer of the particle swarm algorithm is characterized in that the expression of the judgment condition of the event-triggered control in the step three is:

[0050] t i+1 = inf{t:t>t i ,||A+c||||e(t i )-e(t)||≥(k+εe(t i ))},

[0051] Where A and c are parameters in the state equation of the doubly-fed wind turbine generator, and e(t) is the tracking error at time t. The event-triggered control mechanism is introduced, the event-triggered judgment conditions are designed, and the principle of event-triggered control is combined with the sliding mode observer. While ensuring the stable control effect of the system, the communication between the computing unit and the execution unit is reduced, and the chattering of the sliding mode observer is weakened.

[0052] The method for detecting faults of a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm is characterized in that in step 4, the rotor current output value i r Event-triggered sliding mode observations based on particle swarm optimization By subtracting, we can get the maximum tracking residual value Δ of the doubly fed wind turbine in the undisturbed working state. max .

[0053] The fault detection method of the doubly-fed wind turbine generator based on the event-triggered sliding mode observer of the particle swarm algorithm is characterized in that in the step 5, when a rotor current sensor fault occurs in any time period of the doubly-fed wind turbine generator, the rotor current output value under the fault state is subtracted from the event-triggered sliding mode observation value based on the particle swarm algorithm, and the obtained residual Δ is compared with the maximum tracking residual value Δ under the undisturbed working state. max By comparison, fault detection of the doubly-fed wind turbine generator can be realized;

[0054] Observe the rotor current output value of the doubly-fed wind turbine generator and add the fault value to the rotor current sensor at any time When the obtained tracking residual Δ>Δ max , it is judged that the system has a rotor current sensor failure.

[0055] The fault detection method of the doubly-fed wind turbine generator based on the event-triggered sliding mode observer of the particle swarm algorithm is characterized in that in the step 6, when a voltage drop fault occurs in any time period of the doubly-fed wind turbine generator, the rotor current output value under the fault state is subtracted from the event-triggered sliding mode observation value based on the particle swarm algorithm, and the obtained residual Δ is compared with the maximum tracking residual value Δ under the undisturbed working state. max Compare and realize the fault detection of doubly fed wind turbine generator; establish the grid voltage drop fault of doubly fed wind turbine generator as Among them U g is the grid voltage amplitude, ω g is the grid voltage electrical angular velocity, l represents the percentage of fault degree. When the tracking residual Δ>Δ max When the voltage drops, the system is judged to have a voltage drop fault.

[0056] The fault detection method of the doubly-fed wind turbine generator based on the event-triggered sliding mode observer of the particle swarm algorithm is characterized in that in the step 7, a stator turn-to-turn short circuit fault occurs in any time period of the doubly-fed wind turbine generator, and the rotor current output value under the fault state is subtracted from the event-triggered sliding mode observation value based on the particle swarm algorithm, and the obtained residual Δ is compared with the maximum tracking residual value Δ under the undisturbed working state. max By comparison, fault detection of the doubly-fed wind turbine generator can be realized;

[0057] Assume that the stator resistance change when a stator turn-to-turn fault occurs in a doubly-fed wind turbine is When the obtained tracking residual Δ>Δ max When the stator turns fault occurs in the doubly-fed wind turbine generator, it is determined that the stator turns fault occurs in the doubly-fed wind turbine generator.

[0058] The present invention has the following advantages.

[0059] (1) The present invention designs a method for fault detection of a doubly-fed induction generator based on an event-triggered sliding mode observer of a particle swarm algorithm, which has the advantages of simple structure, finite time convergence, reduced sliding mode chattering, and strong robustness.

[0060] (2) The present invention adopts an event-triggered control mechanism to reduce the communication between the computing unit and the execution unit while ensuring the stable control effect of the system.

[0061] (3) This paper adopts the particle swarm algorithm to optimize the parameters of the sliding mode observer, which has higher tracking accuracy than the traditional method.

[0062] (4) In this paper, the rotor current output value is subtracted from the observed value of the event-triggered sliding mode observer based on the particle swarm algorithm to obtain the tracking residual Δ. The maximum tracking residual Δ under no disturbance is obtained by subtracting Δ from the observed value of the event-triggered sliding mode observer based on the particle swarm algorithm. max By comparison, it is possible to quickly determine whether the doubly-fed wind turbine generator has a fault. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 The present invention provides a fault detection structure framework for a doubly-fed wind turbine generator using an event-triggered sliding mode observer based on a particle swarm algorithm.

[0064] Figure 2 The schematic diagram is a principle diagram of fault detection of a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm according to the present invention.

[0065] Figure 3 The schematic diagram is a principle diagram of fault detection when a rotor current fault occurs in a doubly-fed wind generator.

[0066] Figure 4 The schematic diagram is a principle diagram of fault detection when a stator turn-to-turn fault occurs in a doubly-fed wind turbine generator.

[0067] Figure 5 It is a schematic diagram of the principle of fault detection when a voltage drop fault occurs in a doubly-fed wind turbine generator. Specific implementation methods

[0068] The present invention will be further described below in conjunction with the accompanying drawings.

[0069] like Figure 1 As shown, a fault detection method for a doubly-fed generator based on an event-triggered sliding mode observer of a particle swarm algorithm is characterized by mainly comprising the following steps:

[0070] Step 1: Construct a rotor current sliding mode observer for a doubly-fed wind turbine generator;

[0071] Step 2: Introduce the particle swarm algorithm to optimize the parameters for constructing the sliding mode observer;

[0072] Step 3: Introduce an event-triggered control mechanism to control the constructed sliding mode observer;

[0073] Step 4: by subtracting the rotor current output value of the doubly-fed wind turbine generator in a fault-free state from the sliding mode observation value, the maximum tracking residual value of the doubly-fed wind turbine generator in an undisturbed working state is obtained;

[0074] Step 5: Introduce the rotor current sensor fault, and compare the residual between the rotor current output value and the sliding mode observation value with the maximum tracking residual value under the fault-free state, so as to realize the fault detection of the doubly fed wind turbine generator;

[0075] The above-mentioned method for fault detection of a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm is characterized in that the expression of the sliding mode observer in step 1 is:

[0076]

[0077] According to the rotor current state space equation of the doubly fed wind turbine, a sliding mode observer of the rotor current is constructed, and the sliding mode approach rate is introduced. In the formula △ is the residual between the sliding mode estimation value and the system output value, ξ is the maximum steady-state residual critical value that the system can accept, k is the gain parameter of the constant velocity approach rate part, η is the parameter of the exponential switching term, β is the parameter of the switching term, ε and δ0 are the switching function gain parameters, α>0, s represents the sliding mode switching surface and is the sliding mode switching surface parameter and c>0, ‖s‖ represents the parameter norm of s. The method of obtaining the rotor current sliding mode observer model based on the sliding mode approach rate through the sliding mode variable structure control theory is:

[0078] The rotor voltage equation of the system is:

[0079]

[0080] The stator voltage equation of the system is:

[0081]

[0082] The flux equation of the system is:

[0083]

[0084] In the formula, u ds ,u qs ,u dr ,u qr are the voltage vectors of the stator and rotor of the d and q axes respectively, i ds ,i qs ,i dr ,i qr are the stator and rotor current vectors of the d and q axes respectively, ψ ds ,ψ qs ,ψ dr ,ψ qr are the stator and rotor magnetic flux vectors of the d and q axes respectively, R S , R r are stator resistance and rotor resistance respectively, L s,L r are the stator self-inductance and rotor self-inductance, L m is the mutual inductance between the stator and the rotor, L σs ,L σr are the leakage inductance of the stator and rotor respectively, ω r ,ω s ,ω f are the rotor speed, stator speed and slip speed respectively, the d-axis and q-axis current components are used as state variables, and the d-axis and q-axis voltage components are used as input quantities. The state space equation of DFIG is:

[0085]

[0086] in represents the leakage inductance coefficient;

[0087] Considering the state equation of the rotor current, the state equation is:

[0088]

[0089] where i s =[i ds i qs ] T and u=[u ds u qs u dr u qr ] T is the input variable of the system, y is the output vector of the system, and C is the output matrix of the system, where: Using the stator voltage oriented control method, the method to find the relationship between the stator current term and the rotor current term is:

[0090] The present invention ignores the stator resistance, and there is no difference between the stator flux orientation and the stator voltage orientation. Ignoring the stator resistance, the equation satisfied after adopting the voltage and flux orientation is:

[0091]

[0092] ψ s is the amplitude of the stator flux;

[0093] The relationship between the stator current term and the rotor current term is established as:

[0094]

[0095] Where ψ s is the stator flux,

[0096] The sliding mode observer model is constructed based on the rotor current state equation:

[0097]

[0098] in is the observed value of the current, is the control rate of the sliding mode observer; the approach rate of the rotor current sliding mode observer is designed as:

[0099]

[0100] Combining sliding mode variable structure control theory, design sliding mode control law for:

[0101]

[0102] Substituting the sliding mode control law into the rotor current sliding mode observer model, we can obtain the sliding mode observer based on the approach rate:

[0103]

[0104] The above-mentioned method for fault detection of a doubly-fed wind turbine generator using an event-triggered sliding mode observer based on a particle swarm algorithm is characterized in that the fitness function expression of the particle swarm algorithm in step 2 is:

[0105]

[0106] C1 and C2 are weight parameters, and C1+C2=1, u q and u d It is the sliding mode control law of the rotor current. The parameters of the sliding mode observer and the sliding mode switching surface are optimized by the particle swarm algorithm to determine the sliding mode observer parameters k, α, β, ε, δ0, η, c so that the fitness function reaches the target value;

[0107] The above-mentioned method for fault detection of a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm is characterized in that the expression of the judgment condition of the event-triggered control in step three is:

[0108] t i+1 = inf{t:t>t i ,||A+c||||e(t i )-e(t)||≥(k+εe(t i ))},

[0109] Wherein A and c are parameters in the state equation of the doubly fed wind turbine generator, e(t) is the tracking error at time t, an event-triggered control mechanism is introduced, and the event-triggered judgment condition is designed. The principle of event-triggered control is combined with the sliding mode observer, and the communication between the computing unit and the execution unit is reduced while ensuring the stable control effect of the system, and the chattering of the sliding mode observer is weakened. The above-mentioned fault detection method of the doubly fed wind turbine generator based on the event-triggered sliding mode observer of the particle swarm algorithm is characterized in that in the step four, the rotor current output value ir is compared with the event-triggered sliding mode observation value based on the particle swarm algorithm. By subtracting, we can get the maximum tracking residual value Δ of the doubly fed wind turbine in the undisturbed working state. max .

[0110] like Figure 2 As shown in the figure, since the performance of the sliding mode observer depends on whether the parameters k, α, β, ε, δ0, η, and c are reasonable, and the size of the parameters is usually selected through experience, it is of great significance to use the particle swarm algorithm to optimize the parameters of the sliding mode observer. The method for determining the parameters of the sliding mode observer is:

[0111] The particle swarm algorithm generates a particle swarm, randomly generates the positions and velocities of all particles, and assigns the parameters k, α, β, ε, δ0, η, and c in the sliding mode observer to the examples in the particle swarm in turn, and determines the particle P t and G t ;

[0112] For each particle, its fitness value is compared with the optimal position P that the particle has ever experienced. t The fitness value of the P is compared, and if it is better, it is used as the current P t ;

[0113] For each particle, its fitness value is compared with the optimal position G experienced by the entire particle group. t The fitness value of G is compared, if it is better, it is used as the current G t ;

[0114] According to the speed formula v t+1 =ωv t +c1r1(P t -x t )+c2r2(G t -x t ) and the position formula x t+1 =x t +v t+1 Update particle speed and formula;

[0115] Designing fitness function If the target value of the fitness function does not meet the termination condition, return to step (2); otherwise, exit the algorithm and obtain the optimal solution; where x represents the position of the particle; v represents the velocity of the particle; ω is the inertia factor; c1 and c2 are acceleration constants; r1 and r2 are random numbers in the interval [0, 1]; P t is the optimal position that the particle has searched so far; G t is the optimal position searched by the entire particle swarm so far; Δ q and Δ d is the tracking error of the sliding mode observer; u d and u q They are the sliding mode control laws for the q and d axes respectively; combining the principle of event-triggered control with the sliding mode observer, the method of the rotor current sliding mode observer based on the event-triggered mechanism of the particle swarm algorithm is obtained as follows:

[0116] According to the tracking residual Δ of the sliding mode observer, a new error is defined For any t∈[t i , t i+1 )exist With {t1, t2, ..., t i} represents the potential time node for the control signal update. For any η∈(0, 1) and δ∈(0, ∞), the next triggering time is determined according to the event triggering condition. When the trigger condition is met, the tracking error at this moment is output, and when the trigger condition is not met, the tracking error at the previous moment is maintained. At the same time, it is necessary to prove that the proposed system does not have a Zeno solution, where the Zeno solution refers to the phenomenon that there are infinite trigger moments within a limited event range, and the key to eliminating the Zeno solution is to ensure that the deviation between any two consecutive moments is always bounded by a lower bound, that is, t i+1 -t i ≥m>0.

[0117] like Figure 3 As shown, by changing the rotor current output value i r With sliding mode observations By subtracting, we can get the maximum tracking residual value Δ of the doubly fed wind turbine in the undisturbed working state. max , adding the rotor current sensor fault at any time of the doubly fed wind turbine The estimated value of the rotor current is obtained by the designed event-triggered sliding mode observer based on the particle swarm algorithm. The tracking residual Δ can be obtained by subtracting the sliding mode estimated value from the system output value. When the obtained tracking residual Δ>Δ max When the rotor current sensor is detected, the system is judged to have occurred.

[0118] like Figure 4 As shown, by changing the rotor current output value ir With sliding mode observations By subtracting, we can get the maximum tracking residual value Δ of the doubly fed wind turbine in the undisturbed working state. max . Establish the grid voltage drop fault of the doubly fed wind turbine generator as Among them U g is the grid voltage amplitude, ω g is the electrical angular velocity of the grid voltage, and l represents the percentage of the fault degree. When a grid voltage drop fault occurs at any time, the estimated value of the rotor current is obtained through the designed event-triggered sliding mode observer based on the particle swarm algorithm. The sliding mode estimated value is subtracted from the current output to obtain the tracking residual Δ. When the obtained tracking error Δ>Δ max When the voltage drops, it is judged that the double-fed wind turbine generator has a voltage drop fault.

[0119] like Figure 5 As shown, by changing the rotor current output value i r With sliding mode observations By subtracting, we can get the maximum tracking residual value Δ of the doubly fed wind turbine in the undisturbed working state. max When a stator turn-to-turn fault occurs in a doubly-fed wind turbine, the stator resistance change is When a stator turn-to-turn fault occurs at any time, the estimated value of the rotor current is obtained through the designed event-triggered sliding mode observer based on the particle swarm algorithm. The tracking residual Δ can be obtained by subtracting the sliding mode estimated value from the current output. When the obtained tracking residual Δ>Δ max When the stator turns fault occurs in the doubly-fed wind turbine generator, it is determined that the stator turns fault occurs in the doubly-fed wind turbine generator.

Claims

1. A doubly-fed wind turbine fault detection method based on an event-triggered sliding mode observer using a particle swarm algorithm comprises the following steps: Step 1: Construct a rotor current sliding mode observer for a doubly-fed wind turbine generator; Step 2: Introduce the particle swarm algorithm to optimize the parameters for constructing the sliding mode observer; Step 3: Introduce an event-triggered control mechanism to control the constructed sliding mode observer; Step 4: by subtracting the rotor current output value of the doubly-fed wind turbine generator in a fault-free state from the sliding mode observation value, the maximum tracking residual value of the doubly-fed wind turbine generator in an undisturbed working state is obtained; Step 5: Introduce the rotor current sensor fault, and compare the residual between the rotor current output value and the sliding mode observation value with the maximum tracking residual value under the fault-free state, so as to realize the fault detection of the doubly fed wind turbine generator; Step 6: Introduce a grid voltage drop fault, and compare the residual between the rotor current output value and the sliding mode observation value with the maximum tracking residual value under a fault-free state, thereby realizing fault detection of the doubly-fed wind turbine generator; Step 7: Introduce a stator turn-to-turn short-circuit fault, and compare the residual between the rotor current output value and the sliding mode observation value with the maximum tracking residual value under a fault-free state, thereby realizing fault detection of the doubly-fed wind turbine generator; Wherein, the expression of the sliding mode observer in step 1 is: According to the rotor current state space equation of the doubly fed wind turbine, a sliding mode observer of the rotor current is constructed, and the sliding mode approach rate is introduced. In the formula △ is the residual between the sliding mode estimation value and the system output value, ξ is the maximum steady-state residual critical value that the system can accept, k is the gain parameter of the constant velocity approach rate part, η is the parameter of the exponential switching term, β is the parameter of the switching term, ε and δ0 are the switching function gain parameters, α>0, s represents the sliding mode switching surface and c is the sliding mode switching surface parameter and c>0, ‖s‖ represents the parameter norm of s. The method of obtaining the rotor current sliding mode observer model based on the sliding mode approach rate through the sliding mode variable structure control theory is: The rotor voltage equation of the system is: The stator voltage equation of the system is: The flux equation of the system is: In the formula, u ds ,u qs ,u dr ,u qr are the voltage vectors of the stator and rotor of the d and q axes respectively, i ds ,i qs ,i dr ,i qr are the stator and rotor current vectors of the d and q axes respectively, ψ ds , ψ qs , ψ dr , ψ qr , are the stator and rotor magnetic flux vectors of the d and q axes, respectively, R S , R r are stator resistance and rotor resistance respectively, L s ,L r are the stator self-inductance and rotor self-inductance, L m is the mutual inductance between the stator and the rotor, L σs ,L σr are the leakage inductance of the stator and rotor respectively, ω r ,ω s ,ω f are the rotor speed, stator speed and slip speed respectively, the d-axis and q-axis current components are used as state variables, and the d-axis and q-axis voltage components are used as input quantities. The state space equation of DFIG is: in represents the leakage inductance coefficient; Considering the state equation of the rotor current, the state equation is: where i s =[i ds i qs ] T and u=[u ds u gs u dr u qr ] T is the input variable of the system, y is the output vector of the system, and C is the output matrix of the system, where: Using the stator voltage oriented control method, the method to find the relationship between the stator current term and the rotor current term is: The present invention ignores the stator resistance, and there is no difference between the stator flux orientation and the stator voltage orientation. Ignoring the stator resistance, the equation satisfied after adopting the voltage and flux orientation is: ψ s is the amplitude of the stator flux; The relationship between the stator current term and the rotor current term is established as: Where ψ s is the stator flux, The sliding mode observer model is constructed based on the rotor current state equation: in is the observed value of the current, is the control rate of the sliding mode observer; Design the approach rate of the rotor current sliding mode observer: Combining sliding mode variable structure control theory, design sliding mode control law for: Substituting the sliding mode control law into the rotor current sliding mode observer model, we can obtain the sliding mode observer based on the approach rate: Among them, the fitness function expression of the particle swarm algorithm in step 2 is: C1 and C2 are weight parameters, and C1+C2=1, u q and u d It is the sliding mode control law of the rotor current. The parameters of the sliding mode observer and the sliding mode switching surface are optimized through the particle swarm algorithm to determine the sliding mode observer parameters k, α, β, ε, δ0, η, c so that the fitness function reaches the target value.

2. The fault detection method for a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm according to claim 1 is characterized in that: The expression of the judgment condition of the event trigger control in step 3 is: t i+1 =inf{t:t>t i ,||A+c||||e(t i )-e(t)||≥(k+se(t) i ))}, Where A and c are parameters in the state equation of the doubly fed wind turbine generator, e(t) is the tracking error at time t, and an event-triggered control mechanism is introduced. The event-triggered judgment conditions are designed, and the principle of event-triggered control is combined with the sliding mode observer. This reduces the communication between the computing unit and the execution unit while ensuring the stable control effect of the system, and at the same time weakens the jitter of the sliding mode observer.

3. The fault detection method for a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm according to claim 1, characterized in that: In step 4, the rotor current output value i is r Event-triggered sliding mode observations based on particle swarm optimization By subtracting, we can get the maximum tracking residual value Δ of the doubly fed wind turbine in the undisturbed working state. max .

4. The fault detection method for a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm according to claim 1, characterized in that: In the step 5, when a rotor current sensor fault occurs in any time period of the doubly fed wind turbine generator, the rotor current output value under the fault state is subtracted from the event-triggered sliding mode observation value based on the particle swarm algorithm, and the obtained residual Δ is compared with the maximum tracking residual value Δ under the undisturbed working state. max By comparison, fault detection of the doubly-fed wind turbine generator can be realized; Observe the rotor current output value of the doubly-fed wind turbine generator and add the fault value to the rotor current sensor at any time When the obtained tracking residual Δ>Δ max , it is judged that the system has a rotor current sensor failure.

5. The fault detection method for a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm according to claim 1, characterized in that: In step 6, when a voltage drop fault occurs in any time period of the doubly fed wind turbine generator, the rotor current output value under the fault state is subtracted from the event-triggered sliding mode observation value based on the particle swarm algorithm, and the obtained residual Δ is compared with the maximum tracking residual value Δ under the undisturbed working state. max By comparison, fault detection of the doubly-fed wind turbine generator can be realized; Establish the grid voltage drop fault of the doubly fed wind turbine as Among them U g is the grid voltage amplitude, ω g is the grid voltage electrical angular velocity, l represents the percentage of fault severity, when the obtained tracking residual Δ>Δ max When the voltage drops, the system is judged to have a voltage drop fault.

6. The fault detection method for a doubly-fed wind turbine generator based on an event-triggered sliding mode observer of a particle swarm algorithm according to claim 1, characterized in that: In the step 7, a stator turn-to-turn short circuit fault occurs in any time period of the doubly fed wind turbine generator. By subtracting the rotor current output value under the fault state from the event-triggered sliding mode observation value based on the particle swarm algorithm, the obtained residual Δ is compared with the maximum tracking residual value Δ under the undisturbed working state. max By comparison, fault detection of the doubly-fed wind turbine generator can be realized; Assume that the stator resistance change when a stator turn-to-turn fault occurs in a doubly-fed wind turbine is When the obtained tracking residual Δ>Δ max When the stator turns fault occurs in the doubly-fed wind turbine generator, it is determined that the stator turns fault occurs in the doubly-fed wind turbine generator.

Images