Dynamic event triggered iterative filtering and fault estimation method for permanent magnet direct-driven wind turbine system considering spoofing attack

Abstract

The invention discloses a permanent magnet direct drive wind turbine system dynamic event triggering iterative filtering and fault estimation method considering spoofing attacks. According to the method, firstly, a fusion event triggering mechanism based on a ring event and a dynamic threshold updating law formed by system output and error dependent triggering parameters are designed to balance triggering performance and system performance, and then an iterative filtering and fault estimation algorithm is designed under spoofing attack to estimate system faults. And thus, more accurate fault estimation can be obtained. When the method is applied to a permanent magnet direct drive wind turbine system, the fault estimation effect is good, and the communication frequency is remarkably reduced.

Description

Technical Field

[0001] The present invention relates to a fault estimation method, and in particular to a dynamic event-triggered iterative filtering and fault estimation method for a permanent magnet direct-drive wind turbine system taking deception attacks into account. Background Art

[0002] Nowadays, governments and enterprises are making further efforts to promote the large-scale application of renewable energy and guide the use of clean energy to produce electricity. Among these energy sources, wind energy is recognized as a clean and sustainable energy solution due to its higher cost-effectiveness compared with traditional fossil fuels and nuclear energy. This advantage has stimulated researchers' interest in the stability problem of wind turbine systems. The most commonly used generator in wind power generation is the permanent magnet synchronous generator, which has the advantages of direct drive capability, low speed and low maintenance cost. It is well known that industrial systems are inevitably affected by external interference, environmental noise, component failure and other factors. These factors can cause abnormal operation or even shutdown of wind turbine systems. Therefore, it is particularly important to develop effective fault estimation methods to detect and compensate for the impact of faults. In addition, wind speed has the characteristics of random changes, and traditional methods can only handle systems under specific wind conditions. Markov process can effectively model this random characteristic of wind speed. Therefore, it is necessary to apply Markov jump system theory and fuzzy control theory to develop filtering and fault estimation methods for permanent magnet direct-drive wind turbine systems.

[0003] In addition, with the popularization of communication technology and its integration with industrial systems, networked control has become an important way to promote the development of permanent magnet direct-drive wind turbine systems. In the network control framework, data exchange is carried out through a network channel that adopts a time-triggered (periodic) mode. However, with the expansion of the system scale and the arbitrary release of redundant data packets, the shadow of network congestion looms over the entire system. Therefore, there is an urgent need to adopt an effective communication scheme in permanent magnet direct-drive wind turbine systems to reduce the frequency of data transmission and save limited network resources. However, existing research work mainly focuses on static event triggering strategies, and the network burden reduction is limited. Moreover, the impact of outlier strategy data on event generators is not considered. Therefore, it is necessary to improve the existing event triggering strategy to save bandwidth usage.

[0004] In addition to considering limited network resources, exposed data transmission is also vulnerable to hacker attacks. In recent years, large-scale power outages caused by cyber attacks have occurred in foreign countries, and network security has become a new challenge facing the power system. Therefore, the network security strategy of wind turbine systems needs further research. Summary of the invention

[0005] The purpose of the present invention is to propose a dynamic event triggered iterative filtering and fault estimation method for a permanent magnet direct-drive wind turbine system taking into account deception attacks, which can effectively improve the fault estimation performance of the permanent magnet direct-drive wind turbine system under limited bandwidth and deception attacks.

[0006] The specific technical solution of the present invention is as follows: A method for dynamic event-triggered iterative filtering and fault estimation of a permanent magnet direct-drive wind turbine system taking into account deception attacks, comprising the following steps:

[0007] A fusion event trigger mechanism based on ring events and a dynamic threshold update law consisting of system output and error-dependent trigger parameters are designed to balance trigger performance and system performance:

[0008] According to the principles of aerodynamics, the power generated by wind energy is expressed as follows:

[0009]

[0010] in, Indicates air density, its unit is Kg / m 3 ;π represents pi; It represents the length of turbine blades, and its unit is m; Indicates the power factor; v w Indicates wind speed, the unit is m / s; represents the tip speed ratio; v t Indicates the turbine speed, the unit is rpm; Indicates the pitch angle, its unit is rad;

[0011] The corresponding mechanical torque is calculated as follows:

[0012]

[0013] Typically, the dynamic characteristics of a pitch pushrod are:

[0014]

[0015] in, is the time constant, is the control signal;

[0016] Next, the wind speed is split into the low-frequency stable part v ws and high frequency interference part v wp , and for mechanical torque Taylor expansion at a stable point:

[0017]

[0018] in, v wd =vw -v ws Indicates deviation;

[0019] Mechanical torque The electric torque applied by the permanent magnet synchronous motor cancel each other out, and this interaction has an impact on the motor speed v t The impact is as follows:

[0020]

[0021] in, represents the moment of inertia, represents the friction coefficient;

[0022] The upper type is at the operating point By performing Taylor expansion, we can obtain the following deviation equation:

[0023]

[0024] in, The horizontal bar and wave above the variable represent the operating point and deviation, respectively;

[0025] Next, the nonlinear model of the permanent magnet direct-drive wind turbine system is established as follows:

[0026]

[0027] In the formula, Represents stator resistance, its unit is Ω; It represents the stator inductance on the d-axis, and its unit is H; It represents the stator inductance on the q-axis, and its unit is H; It represents the magnetic flux of the permanent magnet, and its unit is Wb; Indicates the number of poles; V d and V q Represent the q-axis and d-axis voltages respectively; i d and i q represent the q-axis and d-axis currents respectively;

[0028] The electric torque applied by the permanent magnet synchronous motor can be further rewritten as:

[0029]

[0030] For the nonlinear terms in the system, fuzzy control theory is used to deal with them; assuming And choose the fuzzy membership function as:

[0031]

[0032] Among them, η1, η2 are available premise variables, ηmax is a known upper bound;

[0033] Next, set the wind speed operating points and model the random variation of wind speed as a Markov process Then, the model of permanent magnet direct drive wind turbine can be further expressed as:

[0034] Fuzzy rule i: If is η j ,So

[0035]

[0036] in, d=v wd , f, y and z are the system state, control input, disturbance, fault, system output and regulation output respectively; in addition, the system matrix is ​​as follows:

[0037]

[0038] in, Indicates the i-th operation partition The same is true; to simplify the notation, let

[0039] Next, the sampling interval is set to T, and the Euler method is used to transform the above continuous-time TS fuzzy Markov jump system into the following discrete system model:

[0040] Fuzzy rule i: If is η j ,So

[0041]

[0042] in,

[0043] For the above wind turbine system, in order to reduce the network transmission burden, a dynamic event trigger mechanism based on ring events and a dynamic threshold update law consisting of system output and error-dependent trigger parameters are designed to balance the trigger performance and system performance; the estimated signal is first verified by the event criterion and then transmitted to the controller through the ideal network; the dynamic event trigger mechanism is constructed as follows:

[0044]

[0045] in, Represents the output trigger error, Φ irepresents the weight matrix to be designed, k s Indicates the last triggering time, k s+1 Indicates the next trigger time, dynamic threshold Updated by:

[0046]

[0047] Among them, q belongs to the set {1, 2}, μ q is the weight parameter, γ aq and γ bq is the sensitivity parameter and h is the integration length.

[0048] Furthermore, an iterative filtering and fault estimation algorithm is designed to estimate system faults under deception attacks. The specific steps are as follows:

[0049] When the trigger output is transmitted in an insecure network channel, the enemy can attack it by tampering with the transmitted data; this attack mode is called a spoofing attack; once the attack is successful, the transmitted data will be rewritten as:

[0050]

[0051] Among them, α k ∈{0, 1} is an expectation The random variable, attack signal φ k The energy of is limited to ||φ k ||≤κ||x k ||, κ is a positive scalar;

[0052] Next, the iterative filter and fault estimator are designed as follows:

[0053]

[0054] Where [m] represents the number of iterations, represents the state of the filter at the mth iteration, represents the estimated value of the fault at the mth iteration; and is the gain to be designed; represents the estimated error signal of the mth iteration generated by the following observer:

[0055]

[0056] in, represents the estimate of the system state at the mth iteration, represents the estimate of the adjusted output at the mth iteration, is the gain to be designed;

[0057] set up The following error system can be obtained:

[0058]

[0059] in, H=diag{I x ;I x ;I xf ; 0 f×f}, E{a} represents the mathematical expectation of a; in addition, the system matrices are:

[0060]

[0061] The iterative filter and fault estimator can ensure the random stability of the error system under deception attacks. The proof process is as follows:

[0062] D001: Select the composite energy function of the following form:

[0063]

[0064] D002: Among them, is a positive definite symmetric matrix;

[0065] D003: Next, calculate the difference And consider H ∞ Performance available:

[0066]

[0067] D004: In the formula,

[0068] D005: On the one hand, considering the energy limit of the attack signal, we can get:

[0069]

[0070] D006: In the formula, ∈1 is an arbitrary positive constant;

[0071] D007: Next, introduce the event triggering conditions into the analysis.

[0072]

[0073] D008: In the formula, ∈2 and ∈3 are positive scalars;

[0074] D009: Substitute D005 and D007 into D003, and we get:

[0075]

[0076] D010: In the formula, ∈5=∈2-∈3;

[0077] D011: Definition Available

[0078]

[0079] D012: Next, introduce the free matrix And applying the lemma, we can get

[0080]

[0081] D013: In the formula,

[0082] D014: Notice There is coupling, so let Available

[0083]

[0084] D015: Substitute D014 into D012, and we get

[0085]

[0086] D016: In the formula,

[0087] D017: Then, repeating the Schur's complement lemma for D015, we can obtain the following stability condition:

[0088]

[0089] D018: If condition D017 can be guaranteed to be met, then holds; therefore, the error system is stochastically stable. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 is a system dynamic response diagram of an embodiment;

[0091] Figure 2 It is an iterative estimation diagram of z using the method proposed by the present invention in the embodiment;

[0092] Figure 3 It is an iterative estimation diagram of f using the method proposed in the present invention in the embodiment;

[0093] Figure 4 This is a response diagram of the event triggering threshold value using the method proposed by the present invention in the embodiment;

[0094] Figure 5The trigger diagram of the event triggering of the embodiment using the method proposed by the present invention; DETAILED DESCRIPTION

[0095] The present invention will be further described below with reference to specific embodiments. It should be understood that these embodiments are only for illustration purposes. The present invention is not intended to limit the scope of the present invention. After reading the present invention, various equivalent modifications of the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0096] like Figure 1 As shown, a method for dynamic event-triggered iterative filtering and fault estimation of a permanent magnet direct-drive wind turbine system taking into account deception attacks comprises the following steps:

[0097] Step 1: Set the initial values ​​of various parameters;

[0098] Step 2: Sample sensor output y k ;

[0099] Step 3: Update the threshold parameter σ q，k ;

[0100] Step 4: Use threshold parameter σ q，k With the output y k Verify event trigger conditions and update trigger output

[0101] Step 5: Use the output signal y tampered by the attacker a，k Update the filter, fault estimator and observer, and output the mth iteration and And save it into the register;

[0102] Step 6: Repeat steps 2, 3, 4, and 5 until one iteration is completed.

[0103] Step 7: Repeat the above steps for n iterations, analyze all the data stored in the registers, and output the fault estimate.

[0104] An embodiment of the present invention is described below:

[0105] Consider the system parameters shown in the following table:

[0106] Figure 1 is a system dynamic response diagram of an embodiment of the present invention; the regulation output and system fault estimation diagram of the permanent magnet direct-drive wind turbine system under the designed iterative filtering and fault estimation method are as follows Figure 2 and Figure 3 As shown in the figure, the designed dynamic threshold response diagram is as follows Figure 4 As shown in the figure, the trigger interval diagram of the designed dynamic event trigger mechanism is as follows Figure 5 As shown. Figure 2 and Figure 3 It can be seen from the figure that the proposed dynamic event-triggered iterative filtering and fault estimation method successfully estimates the regulation output and system faults, and the estimation effect improves with the increase of the number of iterations. Figure 4 and Figure 5 It can be seen that the proposed event trigger mechanism further reduces the number of signal transmissions and reduces the network bandwidth usage while ensuring the same system performance.

[0107] References

[0108] [1] Yan, S., Gu, Z., Park, JH, & Xie, X. (2021). Adaptive memory-event-triggered static output control of TS fuzzy wind turbine systems. IEEE Transactions on Fuzzy Systems, 30(9), 3894-3904.

[0109] [2]Peng, C., & Yang, TC (2013). Event-triggered communication and H∞control co-design for networked control systems. Automatica, 49(5), 1326-1332.

Claims

1. A method for dynamic event-triggered iterative filtering and fault estimation of a permanent magnet direct-drive wind turbine system taking into account spoofing attacks, characterized in that: The following steps are involved: A fusion event trigger mechanism based on ring events and a dynamic threshold update law consisting of system output and error-dependent trigger parameters are designed to balance trigger performance and system performance. An iterative filtering and fault estimation algorithm is designed to estimate system faults under deception attacks.

2. According to the method for dynamic event triggering iterative filtering and fault estimation of a permanent magnet direct-drive wind turbine system taking into account deception attacks described in claim 1, a fusion event triggering mechanism based on ring events and a dynamic threshold update law consisting of system output and error-dependent trigger parameters are designed to balance trigger performance and system performance. The specific steps are as follows: According to the principles of aerodynamics, the power generated by wind energy is expressed as follows: in, Indicates air density, its unit is Kg / m 3 ;π represents pi; It represents the length of turbine blades, and its unit is m; Indicates the power factor; v w Indicates wind speed, the unit is m / s; represents the tip speed ratio; v t Indicates the turbine speed in rpm; Indicates the pitch angle, its unit is rad; The corresponding mechanical torque is calculated as follows: Typically, the dynamic characteristics of a pitch pushrod are: in, is the time constant, is the control signal; Next, the wind speed is split into the low-frequency stable part v ws and high frequency interference part v wp , and for mechanical torque Taylor expansion at a stable point: in, v wd =v w -v ws Indicates deviation; Mechanical torque The electric torque applied by the permanent magnet synchronous motor cancel each other out, and this interaction has an impact on the motor speed v t The impact is as follows: in, represents the moment of inertia, represents the friction coefficient; The upper type is at the operating point By performing Taylor expansion, we can obtain the following deviation equation: in, The horizontal bar and wave above the variable represent the operating point and deviation, respectively; Next, the nonlinear model of the permanent magnet direct-drive wind turbine system is established as follows: In the formula, Represents stator resistance, its unit is Ω; It represents the stator inductance on the d-axis, and its unit is H; It represents the stator inductance on the q-axis, and its unit is H; It represents the magnetic flux of the permanent magnet, and its unit is Wb; Indicates the number of poles; V d and V q Represent the q-axis and d-axis voltages respectively; i d and i q represent the q-axis and d-axis currents respectively; The electric torque applied by the permanent magnet synchronous motor can be further rewritten as: For the nonlinear terms in the system, fuzzy control theory is used to deal with them; assuming And choose the fuzzy membership function as: Among them, η1, η2 are available premise variables, η max is a known upper bound; Next, set the wind speed operating points and model the random variation of wind speed as a Markov process Then, the model of permanent magnet direct drive wind turbine can be further expressed as: Fuzzy rule i: If is η j ,So in, d=v wd , f, y and z are the system state, control input, disturbance, fault, system output and regulation output respectively; in addition, the system matrix is ​​as follows: in, Indicates the i-th operation partition The same is true; to simplify the notation, let Next, the sampling interval is set to T, and the Euler method is used to transform the above continuous-time TS fuzzy Markov jump system into the following discrete system model: Fuzzy rule i: If is η j ,So in, For the above wind turbine system, in order to reduce the network transmission burden, a dynamic event trigger mechanism based on ring events and a dynamic threshold update law consisting of system output and error-dependent trigger parameters are designed to balance the trigger performance and system performance; the estimated signal is first verified by the event criterion and then transmitted to the controller through the ideal network; the dynamic event trigger mechanism is constructed as follows: Among them, e y，k =y k -y k , represents the output trigger error, Φ i represents the weight matrix to be designed, k s Indicates the last triggering time, k s+1 Indicates the next trigger time, dynamic threshold Updated by: Among them, q belongs to the set {1, 2}, μ q is the weight parameter, γ aq and γ bq is the sensitivity parameter and h is the integration length.

3. According to the method for iterative filtering and fault estimation of a permanent magnet direct-drive wind turbine system under deception attack according to claim 2, an iterative filtering and fault estimation algorithm is designed to estimate system faults under deception attack, and the specific steps are as follows: When the trigger output is transmitted in an insecure network channel, the enemy can attack it by tampering with the transmitted data; this attack mode is called a spoofing attack; once the attack is successful, the transmitted data will be rewritten as: in, α k ∈{0, 1} is an expectation The random variable, attack signal φ k The energy of is limited to ||φ k ||≤κ||x k ||, κ is a positive scalar; Next, the iterative filter and fault estimator are designed as follows: Where [m] represents the number of iterations, represents the state of the filter at the mth iteration, represents the estimated value of the fault at the mth iteration; and is the gain to be designed; represents the estimated error signal of the mth iteration generated by the following observer: in, represents the estimate of the system state at the mth iteration, represents the estimate of the adjusted output at the mth iteration, is the gain to be designed; set up The following error system can be obtained: in, H=diag{I x ;I x ;I xf ; 0 f×f }, E{a} represents the mathematical expectation of a; in addition, the system matrices are: The iterative filter and fault estimator can ensure the random stability of the error system under deception attacks. The proof process is as follows: B001: Select a composite energy function of the following form: B002: Among them, is a positive definite symmetric matrix; B003: Next, calculate the difference And consider H ∞ Performance available: B004: In the formula, B005: On the one hand, considering the energy limit of the attack signal, we can get: B006: In the formula, ∈1 is an arbitrary positive constant; B007: Next, introduce the event triggering conditions into the analysis. B008: In the formula, ∈2 and ∈3 are positive scalars; B009: Substitute B005 and B007 into B003, and we get: B010: In the formula, ∈5=∈2-∈3; B011: Definition Available B012: Next, introduce the free matrix And applying the lemma, we can get B013: In the formula, B014: Notice There is coupling, so let Available B015: Substitute B014 into B012, and we get B016: In the formula, B017: Then, repeating the Schur complement lemma for B015, we can obtain the following stability condition: B018: If condition B017 can be guaranteed to be met, then holds; therefore, the error system is stochastically stable.

Images