A Fault Diagnosis Method and System for a Hydraulic Turbine Governing System

Abstract

The present invention discloses a fault diagnosis method and system for a hydroturbine governing system. The method includes: by analyzing the propagation path of sensor faults in the hydroturbine governing system and the influence of parameter changes on the system behavior, establishing a parameter uncertainty model of the hydroturbine governing system including sensor faults; establishing a dedicated observer according to the parameter uncertainty model, and combining the parameter uncertainty model and H ∞ theory, optimizing the parameters of different sub-observers for fault detection, and each sub-observer performs state estimation on the hydroturbine governing system; comparing the state estimation values of each state observer with the state estimation value of the reference observer to generate residuals, and determining the fault diagnosis of the hydroturbine governing system through residual analysis. The beneficial effects of quickly obtaining relevant information such as the time of fault occurrence and the location of the fault source without a large amount of data driving and improving the robustness of the fault diagnosis method are achieved.

Description

Technical Field

[0001] The present invention belongs to the field of fault diagnosis of hydraulic turbine governing systems, and more specifically, relates to a fault diagnosis method and system for a hydraulic turbine governing system. Background Art

[0002] Hydropower energy plays a crucial role in China's energy system and is of great significance for the realization of the "3060" goal. As a key regulating device of hydropower units, the fault diagnosis of the hydraulic turbine governing system can provide support for the implementation of condition-based maintenance and fault-tolerant control of the hydraulic turbine governing system, which is of great significance for ensuring the safety and stability of hydropower units. However, due to the complexity of the governing system, it is very difficult to research and apply the fault diagnosis technology of the hydraulic turbine governing system.

[0003] The fault diagnosis method based on the analytical model can make full use of the prior knowledge of the system and has advantages such as low computational complexity and low cost. Therefore, it has always been a research hotspot in the field of fault diagnosis. However, the fault diagnosis method based on the analytical model has high requirements for the accuracy of the system model. However, due to technical limitations and insufficient understanding of the system, there are often deviations between the analytical model established for the system and the actual system behavior, which cannot accurately describe the behavior of the system, thereby degrading the performance of the fault diagnosis method. Summary of the Invention

[0004] Aiming at the defects of the related technologies, the purpose of the present invention is to provide a fault diagnosis method and system for a hydraulic turbine governing system, aiming to solve the problem that there are often deviations between the analytical model established for the system and the actual system behavior, which cannot accurately describe the behavior of the system, thereby degrading the performance of the fault diagnosis method.

[0005] To achieve the above purpose, the present invention provides the following technical solutions.

[0006] In the first aspect, the present invention provides a fault diagnosis method for a hydraulic turbine governing system, including:

[0007] S1. Establish a hydraulic turbine governing system model including a fault distribution matrix;

[0008] S2. Construct a parameter uncertainty model of the hydraulic turbine governing system according to the system behavior deviation caused by parameter uncertainty in the hydraulic turbine governing system model;

[0009] S3. Establish a fault observer according to the parameter uncertainty model. The fault observer includes a reference channel and a fault detection channel, which are respectively composed of sub-observers with different fault sensitivities. According to the requirements of different fault detection channels for fault sensitivity, combine the parameter uncertainty model and H ∞The theory is used to optimize the parameters of sub - observers for different fault detections. Each sub - observer performs state estimation on the hydro - turbine governing system. Among them, the sub - observer adopts an unknown input observer;

[0010] S4. Compare the state estimation values of each state observer with those of the reference observer to generate a residual sequence, and judge whether the hydro - turbine governing system has a fault according to the threshold method.

[0011] Optionally, in S3, it includes:

[0012] S31. For each fault - detection channel, divide the fault vector of the hydro - turbine governing system into two complementary sets, Set 1 and Set 2. Among them, for the faults in Set 1, it is expected that the fault - detection channel is sensitive to them; for the faults in Set 2, it is expected that the fault - detection channel is not sensitive to them;

[0013] S32. For the reference channel, set a reference unknown input observer UIO 0 , where Set 2 is the set containing all faults, and its state estimation value is decoupled from all faults; for each fault - detection channel except the reference channel, set a corresponding state unknown input observer UIO k , whose state estimation value is decoupled from the faults in Set 2; or, the reference channel is not decoupled from any faults, and for each fault - detection channel except the reference channel, set a corresponding state unknown input observer UIO k , whose state estimation value is decoupled from the faults in Set 1;

[0014] S33. According to the requirements of fault sensitivity for different fault - detection channels, combined with the parameter uncertainty model and H ∞ theory, design the H 0 optimization performance index for the reference unknown input observer UIO ∞ , and design the H k / H - optimization performance index for the state unknown input observer UIO ∞ ;

[0015] S34. Use linear matrix inequalities to solve the optimization performance index in S33 to obtain the optimal values of each performance index, and solve the feedback gain matrix of each unknown input observer to complete the construction of the dedicated observer;

[0016] S35. Use each of the sub - observers to perform state estimation on the hydro - turbine governing system.

[0017] Optionally, in S32, it includes:

[0018] Divide the faults into fk (i.e., Set 1), f k (i.e., Set2). For the reference channel, a reference unknown input observer UIO is set 0 , and take f 0 = φ, f 0 = f; where f is the fault vector of the system;

[0019] For the fault detection channel, for f k respectively, a state unknown input observer UIO with specific sensitivity is set k ;

[0020] Solve the matrix of the corresponding unknown input observer UIO k The form of the unknown input observer is: The errors and residuals between the state estimated value and the actual state value of the unknown input observer are respectively:

[0021]

[0022]

[0023] 1) For UIO

[0024] 0 The state estimation error is:

[0025]

[0026] 2) For UIO k The state estimation error is:

[0027]

[0028] 3) The residual of UIO k is:

[0029]

[0030] where respectively represent the components corresponding to f k in , represents the component corresponding to f k in , and n represents the dimension of the system state vector x.

[0031] Optionally, in S33, the reference unknown input observer UIO 0 designs H ∞ to optimize the performance index, and the optimization performance index is as follows: ​

[0032]

[0033] The state-unknown input observer UIO k Design H ∞ / H - Optimize the performance index, and the optimized performance index is as follows:

[0034]

[0035] Optionally, in S34, it includes:

[0036] Use the bounded real lemma and Suchr complement lemma to solve the optimal solution of the optimization problem under the matrix inequality constraint corresponding to the optimized performance index, and obtain the corresponding feedback gain matrix Use the matrix Solve the matrix Complete the construction of the dedicated observer.

[0037] Optionally, in S1, it includes:

[0038] S11. Establish an analytical model of the hydraulic turbine governing system at the equilibrium point of the hydraulic turbine governing system;

[0039] S12. Analyze the propagation path of the speed sensor fault f x and the servomotor displacement sensor fault f y in the hydraulic turbine governing system, let f T =[f x f y , and establish a fault distribution matrix in combination with the analytical model of the hydraulic turbine governing system:

[0040]

[0041]

[0042] Obtain a hydraulic turbine governing system model including the fault distribution matrix, which has the following form

[0043]

[0044] y = Cx + E s f

[0045] where, k p is the proportional gain, k i is the integral gain, k d is the derivative gain, b p is the permanent speed droop coefficient, e qy is the transfer coefficient of flow rate to guide vane opening, e qhis the transfer coefficient of flow rate to head, is the control signal of the u controller, x is the unit frequency, c is the given frequency of the unit, T y is the response time constant of the main servomotor, T 1v is the differential time constant, E s is the output fault distribution matrix, E a is the process fault distribution matrix, f is the system fault vector.

[0046] Optionally, in S2, it includes:

[0047] Perform a first-order Taylor expansion on the hydraulic turbine governing system model of the fault distribution matrix, model the influence of the parameter changes of each system matrix on the hydraulic turbine governing system, and obtain the parameter uncertainty model of the hydraulic turbine governing system; the state space equation of the parameter uncertainty model is:

[0048]

[0049] y = (C + ΔC)x + Du + E s f

[0050] Among them, ΔA, ΔB, ΔC, ΔE a represents the uncertainty of the model, and there is ΔA = E A Σ A F A , ΔB = E B Σ B F B , ΔC = E C Σ C F C , Among them, E A , E B , E C , F A , F B , F C , are all known matrices and satisfy

[0051] Optionally, in S4, it includes:

[0052] Subtract the state estimate value obtained by the UIO k of each fault detection channel from the state estimate value obtained by the UIO 0 of the reference channel to generate a residual sequence r k ;

[0053] The threshold method is used for fault judgment. If the residual sequence value corresponding to the fault detection channel exceeds the set threshold, it is determined that the corresponding fault occurs, and the time information of the fault occurrence and the location information of the fault source are analyzed and obtained.

[0054] In a second aspect, the present invention also provides a fault diagnosis system for a hydraulic turbine governing system, including:

[0055] A fault distribution matrix determination module for establishing a hydraulic turbine governing system model including a fault distribution matrix;

[0056] A parameter uncertainty model establishment module for constructing a parameter uncertainty model of the hydraulic turbine governing system according to the system behavior deviation caused by parameter uncertainty in the hydraulic turbine governing system model;

[0057] A state estimation module for establishing a dedicated observer according to the parameter uncertainty model. The dedicated observer includes a reference channel and a fault detection channel, which are respectively composed of sub-observers with different fault sensitivities. According to the requirements of different fault detection channels for fault sensitivity, combined with the parameter uncertainty model and H ∞ theory, the parameters of different fault detection sub-observers are optimized, and each sub-observer performs state estimation on the hydraulic turbine governing system; among them, the sub-observer adopts an unknown input observer;

[0058] A fault diagnosis module for comparing the state estimation values of each state observer with the state estimation value of the reference observer to generate a residual sequence, and judging whether the hydraulic turbine governing system has a fault according to the threshold method.

[0059] Generally speaking, compared with the prior art through the above technical solutions conceived by the present invention, the following beneficial effects can be achieved:

[0060] 1. By adopting a fault diagnosis method based on state estimation, relevant information such as the time of fault occurrence and the location of the fault source can be quickly obtained on the premise of obtaining the nominal model of the system, which has the advantages of low computational complexity, no need for a large amount of data driving, and high efficiency;

[0061] 2. The H ∞ theory is introduced for optimization, so that the fault diagnosis method has a certain tolerance for system uncertainty and modeling errors, effectively improving the robustness of the fault diagnosis method. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic flowchart of a fault diagnosis method for a hydraulic turbine governing system provided by an embodiment of the present invention;

[0063] Figure 2Schematic diagram of the mathematical model of the hydraulic turbine governing system provided by the embodiment of the present invention;

[0064] Figure 3 Schematic diagram for fault diagnosis using the dedicated observer provided by the embodiment of the present invention;

[0065] Figure 4 Residual sequence when the speed sensor has a fixed - value fault and the servomotor displacement sensor has no fault in the embodiment of the present invention;

[0066] Figure 5 Residual sequence when the speed sensor has no fault and the servomotor displacement sensor has a fixed - value fault in the embodiment of the present invention;

[0067] Figure 6 Residual sequence when the speed sensor has no fault and the servomotor displacement sensor has a time - varying fault in the embodiment of the present invention;

[0068] Figure 7 Residual sequence when the speed sensor has a fixed - value fault and the servomotor displacement sensor has a time - varying fault in the embodiment of the present invention. Detailed implementation manners

[0069] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0070] Embodiment 1

[0071] As Figure 1 shown, the present invention provides a fault diagnosis method for a hydraulic turbine governing system, including:

[0072] S1. Establish a hydraulic turbine governing system model including a fault distribution matrix.

[0073] S2. Construct a parameter uncertainty model of the hydraulic turbine governing system according to the system behavior deviation caused by parameter uncertainty in the hydraulic turbine governing system model.

[0074] S3. Establish a dedicated observer according to the parameter uncertainty model. The dedicated observer includes a reference channel and a fault detection channel, which are respectively composed of sub - observers with different fault sensitivities. According to the requirements of different fault detection channels for fault sensitivity, combined with the parameter uncertainty model and H ∞Optimize the parameters of the sub - observers for different fault detections according to the theory, and each sub - observer performs state estimation on the hydraulic turbine governing system; among them, the sub - observer adopts an unknown input observer.

[0075] S4. Compare the state estimation values of each state observer with those of the reference observer to generate a residual sequence, and judge whether the hydraulic turbine governing system has a fault according to the threshold method.

[0076] Analyze the propagation path of sensor faults in the hydraulic turbine governing system and the influence of parameter changes on the system behavior, establish a parameter uncertainty model of the hydraulic turbine governing system with sensor faults, use an unknown input observer (UIO) as the sub - observer to construct a dedicated observer, and set sub - observers with different parameters for different fault detection channels according to the requirements of different channels for fault sensitivity, and construct a reference observer to achieve robust fault isolation in the case of multi - fault coupling in the governing system.

[0077] Optionally, in S1, it includes:

[0078] S11. Establish an analytical model of the hydraulic turbine governing system at the equilibrium point of the hydraulic turbine governing system.

[0079] Figure 2 The mathematical model of the hydraulic turbine governing system given in this embodiment is provided, and the model parameters are set. Among them, the transfer variables in the model are all per - unit values, and the definitions of each parameter variable are shown in Table 1:

[0080] Table 1 Definition of model parameters of the hydraulic turbine governing system

[0081] <![CDATA[b p / %]]> Permanent slip coefficient <![CDATA[q t > Water flow rate of the water turbine <![CDATA[k p > Proportional gain <![CDATA[m t > Active torque of the water turbine ki Integral gain <![CDATA[m g > Resisting torque <![CDATA[k d > Derivative gain <![CDATA[e g > Transfer coefficient of the resisting torque to the rotational speed <![CDATA[T 1v / s]]> Derivative time constant <![CDATA[e x > Transfer coefficient of the torque to the rotational speed <![CDATA[T y / s]]> Response time constant of the main servomotor <![CDATA[e h > Transfer coefficient of the torque to the water head x Unit frequency <![CDATA[e y > Transfer coefficient of the torque to the guide vane opening c Unit given frequency <![CDATA[e qx > Transfer coefficient of the flow rate to the rotational speed y Guide vane opening of the servomotor <![CDATA[e qh > Transfer coefficient of the flow rate to the water head u Control signal of the controller <![CDATA[e qy > Transfer coefficient of the flow rate to the guide vane opening h Water head

[0082] Establish an analytical model of the hydraulic turbine governing system at the equilibrium point of the hydraulic turbine governing system. According to Figure 2 the shown hydraulic turbine governing system, obtain the state - space expression of the system. The state - space expression is specifically:

[0083]

[0084]

[0085] Among them, a 32 = k i b p a 33 = - k i b p ,

[0086] c 11= 1, c 24 = 1, c 31 = e x , c 34 = e y , c 35 = e h , D = 0, e n = e x -e g , and the remaining parameters are 0.

[0087] S12. Analyze the faults f of the speed sensor x and the faults f of the servomotor displacement sensor y in the hydraulic turbine governing system, and let f T = [f x f y . Combine with the analytical model of the hydraulic turbine governing system to establish the fault distribution matrix:

[0088]

[0089]

[0090] Obtain the hydraulic turbine governing system model including the fault distribution matrix, which has the following form

[0091]

[0092] y = Cx + E s f

[0093] where, k p is the proportional gain, k i is the integral gain, k d is the derivative gain, b p is the permanent speed droop coefficient, e qy is the transfer coefficient of flow rate to guide vane opening, e qh is the transfer coefficient of flow rate to head, u is the control signal of the u controller, x is the unit frequency, c is the unit given frequency, T y is the main servomotor response time constant, T 1v is the derivative time constant, E s is the output fault distribution matrix, E a is the process fault distribution matrix, f is the system fault vector. Among them, f affects the system state vector through E a and f affects the system output vector through E s

[0094] Optionally, in the S2, it includes:

[0095] Perform a first-order Taylor expansion on the hydro-turbine governing system model of the fault distribution matrix, model the influence of parameter changes of each system matrix on the hydro-turbine governing system, and obtain the parameter uncertainty model of the hydro-turbine governing system; the state-space equation of the parameter uncertainty model is:

[0096]

[0097] y = (C + ΔC)x + Du + E s f

[0098] where, ΔA, ΔB, ΔC, ΔE a represent the uncertainty of the model, and ΔA = E A Σ A F A , ΔB = E B Σ B F B , ΔC = E C Σ C F C , where, E A , E B , E C , F A , F B , F C , are all known matrices and satisfy

[0099] Furthermore, since the parameter perturbation of the microcomputer governor is small, it is not considered in this embodiment. Only the errors caused by the uncertainty of the parameters T w , T a , T y , e x , e g , e y , e h , e qx , e qy , e qh are considered. T w0 , T a0 , T y0 , e x0 , e g0 , e y0 , e h0 , e qx0 , e qy0 , e qh0 represent the nominal values of the parameters. Specifically:

[0100] (1) E A , F A , Σ ARespectively:

[0101]

[0102]

[0103] Among them:

[0104]

[0105]

[0106]

[0107]

[0108]

[0109]

[0110] (2)E B 、F B 、Σ B Respectively:

[0111]

[0112]

[0113] Among them, e B51 =...=e B55 =1,

[0114] (3)E C 、F C 、Σ C Respectively:

[0115]

[0116]

[0117] Among them, e C31 =e C32 =e C33 =1, f C11 =e x0 ,f C24 =e y0 ,f C35 =e h0 。

[0118] (4) Respectively:

[0119]

[0120] Among them,

[0121] By establishing the above matrix, the parameter uncertainties of the system can be described mathematically. For example, if the parameter change is 10% of the nominal value, then

[0122] Optionally, in S3, it includes:

[0123] S31. For each fault detection channel, divide the fault vector of the hydraulic turbine governing system into two complementary sets, Set 1 and Set 2; among them, for the faults in Set 1, it is expected that the fault detection channel is sensitive to them; for the faults in Set 2, it is expected that the fault detection channel is insensitive to them.

[0124] S32. For the reference channel, set a reference unknown input observer UIO 0 , where Set 2 is the set containing all faults, and its state estimate is decoupled for all faults; for each fault detection channel except the reference channel, set the corresponding state unknown input observer UIO k , and its state estimate is decoupled for the faults in Set 2.

[0125] S33. According to the requirements of different fault detection channels for fault sensitivity, combined with the parameter uncertainty model and H ∞ theory, design the H 0 optimization performance index for the reference unknown input observer UIO ∞ , and design the H k / H_ optimization performance index for the state unknown input observer UIO ∞ .

[0126] S34. Use linear matrix inequalities to solve the optimization performance index in S33, obtain the optimal values of each performance index, and solve the feedback gain matrices of each unknown input observer to complete the construction of the dedicated observer.

[0127] S35. Use each sub-observer to perform state estimation on the hydraulic turbine governing system.

[0128] The state observers of each fault detection channel perform state estimation on the hydraulic turbine governing system to obtain the state estimation values of each state observer, and the reference observer of the reference channel obtains the actual state value of the hydraulic turbine governing system.

[0129] In an alternative embodiment, the above step S32 can also be replaced by:

[0130] The reference channel does not decouple any faults. For each fault detection channel except the reference channel, a corresponding state unknown input observer (UIO) is set up. k Its state estimate is decoupled from the faults in Set 1.

[0131] In the first embodiment, forward logic is adopted, that is, the reference channel decouples all faults, and other fault detection channels decouple Set 2. In an alternative embodiment, reverse logic is adopted. The reference channel does not perform decoupling operations, and other channels only decouple Set 1. A simpler observer structure and easier feedback gain configuration can be obtained, which means a faster convergence rate of the state estimate, that is, ultimately better real-time performance of fault diagnosis.

[0132] Using the above observer to construct a dedicated observer as shown in Figure 3 . A first-order low-pass filter for the output is constructed and extended to the original system to form an augmented system, so that the faults in the output channel are equivalent to virtual process faults, and the fault diagnosis of the output channel is transformed into the diagnosis of process faults.

[0133] The first-order low-pass filter is specifically:

[0134]

[0135] where φ ∈ R p is the filter state vector, and A s , B s satisfy: -A s is a stable matrix, and B s has full rank. Expanding the above formula into the state-space equation of the parameter-uncertain model, we can get:

[0136]

[0137]

[0138] where:

[0139]

[0140]

[0141] Optionally, in S32, it includes:

[0142] The faults are divided into f k (i.e., Set 1), f k (i.e., Set 2). For the reference channel, a reference unknown input observer (UIO) 0 is set up, and f 0 = φ, f 0 = f, where f is the fault vector of the system.

[0143] For the fault detection channel, for f respectively k Set up a state unknown input observer UIO with specific sensitivity k ;

[0144] Solve the matrix of the corresponding unknown input observer UIO k of The unknown input observer is in the form of:

[0145]

[0146]

[0147] The errors between the state estimated values and the actual state values of the unknown input observer and the residuals are respectively:

[0148] 1) For UIO 0 The state estimation error is:

[0149]

[0150] 2) For UIO k The state estimation error is:

[0151]

[0152] 3) The residual of UIO k is:

[0153]

[0154] where respectively represent the components in f k corresponding to in, represents the component in f k corresponding in, and n represents the dimension of the system state vector x. In this embodiment, the dimension of the preferred system state vector is 5.

[0155] Optionally, in S33, referring to the unknown input observer UIO 0 design H ∞ to optimize the performance index, and the optimized performance index is as follows:

[0156]

[0157] State unknown input observer UIO k design H ∞ / H -Optimize the performance index, and the optimized performance index is as follows:

[0158]

[0159] Optionally, in S34, it includes:

[0160] Use the bounded real lemma and Suchr complement lemma to solve the optimal solution of the optimization problem under the matrix inequality constraint corresponding to the optimized performance index, and obtain the corresponding feedback gain matrix Use the matrix Solve the matrix Complete the construction of the dedicated observer.

[0161] Optionally, in S4, it includes:

[0162] For each fault detection channel, subtract the state estimate value obtained from the UIO k from the state estimate value obtained from the UIO of the reference channel 0 to generate the residual sequence r k .

[0163] Adopt the threshold method for fault judgment. If the residual sequence value corresponding to the fault detection channel exceeds the set threshold, it is determined that the corresponding fault occurs, and the time information of the fault occurrence and the location information of the fault source are analyzed and obtained.

[0164] Use the threshold method to judge whether a fault occurs. When the threshold method judges a fault (or when the convergence time is included in the calculation, subtract the convergence time from the judgment time of the threshold method), it is considered that the fault occurs at that moment. Which channel judges the fault corresponds to the fault occurrence of that channel.

[0165] The technical solution of the embodiment of the present invention analyzes the propagation path of sensor faults in the hydro-turbine governing system and the influence of parameter changes on the system behavior, establishes a parameter uncertainty model of the hydro-turbine governing system including sensor faults, uses the unknown input observer (UIO) as a sub-observer to construct a dedicated observer, and designs corresponding H ∞ or H ∞ / H _ performance indexes to optimize the observer parameters according to the requirements of different channels for fault sensitivity, so as to achieve robust fault isolation in the case of multi-fault coupling in the governing system. It solves the technical problem that there are often deviations between the analytical model established for the system and the actual system behavior, and the system behavior cannot be accurately described, thereby deteriorating the performance of the fault diagnosis method. It realizes the beneficial effects of quickly obtaining relevant information such as the time of fault occurrence and the location of the fault source without a large amount of data-driven, low computational complexity and high efficiency, and designs corresponding H ∞ or H ∞ / H _ The performance indicators are used to optimize the observer parameters, effectively improving the robustness of the fault diagnosis method.

[0166] Embodiment 2

[0167] The present invention also provides a fault diagnosis system for a hydraulic turbine governing system, comprising:

[0168] A fault distribution matrix determination module for establishing a hydraulic turbine governing system model including a fault distribution matrix.

[0169] A parameter uncertainty model establishment module for constructing a parameter uncertainty model of the hydraulic turbine governing system according to the system behavior deviation caused by parameter uncertainty in the hydraulic turbine governing system model.

[0170] A state estimation module for establishing a dedicated observer according to the parameter uncertainty model, the dedicated observer including a reference channel and a fault detection channel, which are respectively composed of sub-observers with different fault sensitivities. According to the requirements of different fault detection channels for fault sensitivity, combined with the parameter uncertainty model and H ∞ theory, the parameters of the sub-observers for different fault detections are optimized, and each of the sub-observers performs state estimation on the hydraulic turbine governing system; wherein, the sub-observer adopts an unknown input observer.

[0171] A fault diagnosis module for comparing the state estimation values of each state observer with the state estimation value of a reference observer to generate a residual sequence, and judging whether the hydraulic turbine governing system has a fault according to the threshold method.

[0172] The fault diagnosis system for a hydraulic turbine governing system provided by the embodiment of the present invention can execute any embodiment of the present invention. The provided fault diagnosis method for a hydraulic turbine governing system has the corresponding functional modules and beneficial effects of the execution method.

[0173] Those skilled in the art can easily understand that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A fault diagnosis method for a hydraulic turbine governing system, characterized in that, comprising: S1. Establish a hydraulic turbine governing system model including a fault distribution matrix; S2. Construct a parameter uncertainty model of the hydraulic turbine governing system according to the system behavior deviation caused by parameter uncertainty in the hydraulic turbine governing system model; S3. Establish a dedicated observer based on the parameter uncertainty model. The dedicated observer includes a reference channel and a fault detection channel, which are respectively composed of sub-observers with different fault sensitivities. According to the requirements of different fault detection channels for fault sensitivity, combined with the parameter uncertainty model and H ∞ theory, optimize the parameters of the sub-observers for different fault detections, and each of the sub-observers performs state estimation on the hydro-turbine governing system; among them, the sub-observer adopts an unknown input observer. S4. Compare the state estimation values of each state observer with the state estimation value of the reference observer to generate a residual sequence, and judge whether the hydraulic turbine governing system has a fault according to the threshold method; wherein, in S3, it includes: S31. For each fault detection channel, divide the fault vector of the hydraulic turbine governing system into two complementary sets, Set 1 and Set 2; wherein, for the faults in Set 1, it is expected that the fault detection channel is sensitive to them; for the faults in Set 2, it is expected that the fault detection channel is not sensitive to them; S32. For the reference channel, set a reference unknown input observer (UIO). 0 , where Set 2 is the set containing all faults, and its state estimate is decoupled from all faults; for each fault detection channel except the reference channel, set the corresponding unknown input observer of the state (UIO). k , and its state estimate is decoupled from the faults in Set 2; alternatively, the reference channel is not decoupled from any fault, and for each fault detection channel except the reference channel, set the corresponding unknown input observer of the state (UIO). k , and its state estimate is decoupled from the faults in Set 1; S33. According to the requirements of different fault detection channels for fault sensitivity, combined with the parameter uncertainty model and H ∞ theory, design the reference unknown input observer UIO 0 to optimize the performance index. For the state unknown input observer UIO ∞ design H k to optimize the performance index; ∞ / H - Optimize the performance index; S34. Solve the optimization performance index in S33 by using linear matrix inequality to obtain the optimal values of each performance index, and solve the feedback gain matrix of each unknown input observer to complete the construction of the dedication observer; S35. Use each sub-observer to perform state estimation on the hydraulic turbine governing system.

2. The method according to claim 1, characterized in that, in S32, it includes: Divide the fault into f k and , f k i.e., Set1, i.e., Set2. For the reference channel, set a reference unknown input observer UIO 0 , and take f 0 = φ, ; where f is the fault vector of the system; For the fault detection channel, for f respectively k Set up a state unknown input observer UIO with specific sensitivity k ; Solve the corresponding unknown input observer (UIO) k of matrix. The form of the unknown input observer is as follows: The errors and residuals between the state estimation value of the unknown input observer and the actual state value are respectively: 1) For UIO 0 The state estimation error is: 2) For UIO k The state estimation error is: 3) UIO k The residual of Among them, C I =[I n×n 0], respectively represent the components of f k corresponding in, represents corresponding in the component, and n represents the dimension of the system state vector x.

3. The method according to claim 2, characterized in that, In S33, the reference unknown input observer UIO 0 Design H ∞ Optimize the performance index, and the optimized performance index is as follows: The state unknown input observer UIO k Design H ∞ / H - Optimize the performance index, and the optimized performance index is as follows:

4. The method according to claim 3, characterized in that, in S34, it includes: Use the bounded real lemma and the Suchr complement lemma to solve the optimal solution of the optimization problem under the matrix inequality constraints corresponding to the optimized performance index, and obtain the corresponding feedback gain matrix Use the matrix Solve the matrix Complete the construction of the dedicated observer 5. The method according to claim 1, characterized in that, in S1, it includes: S11. Establish an analytical model of the hydraulic turbine governing system at the equilibrium point of the hydraulic turbine governing system; S12. Analyze the speed sensor fault f x and the servomotor displacement sensor fault f y in the propagation path in the hydroturbine governing system, and let f T =[f x f y , and establish a fault distribution matrix in combination with the hydroturbine governing system analysis model: Obtain a hydraulic turbine governing system model including the fault distribution matrix, having the following form y = Cx + E s f where k p is the proportional gain, k i is the integral gain, k d is the derivative gain, b p is the permanent speed droop coefficient, e qy is the transfer coefficient of flow rate to guide vane opening, e qh is the transfer coefficient of flow rate to head, u is the control signal of the controller, x is the system state vector, T y is the main servomotor response time constant, T 1v is the derivative time constant, E s is the output fault distribution matrix, E a is the process fault distribution matrix, f is the system fault vector.

6. The method according to claim 5, characterized in that, in S2, it includes: Perform a first-order Taylor expansion on the hydraulic turbine governing system model with the fault distribution matrix, model the influence of the parameter changes of each system matrix on the hydraulic turbine governing system, and obtain the parameter uncertainty model of the hydraulic turbine governing system; the state space equation of the parameter uncertainty model is: y = (C + ΔC)x + Du + E s f Among them, ΔA, ΔB, ΔC, ΔE a represent the uncertainty of the model, and ΔA = E A Σ A F A , ΔB = E B Σ B F B , ΔC = E C Σ C F C , Among them, E A , E B , E C , F A , F B , F C , are all known matrices and satisfy 7. The method according to claim 1, characterized in that, in S4, it includes: The UIO of each fault detection channel k The obtained state estimate value and the UIO of the reference channel 0 The obtained state estimate values are subtracted to generate a residual sequence r k ; Adopt the threshold method for fault judgment. If the residual sequence value corresponding to the fault detection channel exceeds the set threshold, it is determined that the corresponding fault occurs, and the time information of the fault occurrence and the location information of the fault source are analyzed and obtained.

8. A fault diagnosis system for a hydraulic turbine governing system, characterized in that, comprising: A fault distribution matrix determination module for establishing a hydraulic turbine governing system model including a fault distribution matrix; A parameter uncertainty model establishment module for constructing a parameter uncertainty model of the hydraulic turbine governing system according to the system behavior deviation caused by parameter uncertainty in the hydraulic turbine governing system model; A state estimation module is configured to establish a dedicated observer according to the parameter uncertainty model. The dedicated observer includes a reference channel and a fault detection channel, which are respectively composed of sub-observers with different fault sensitivities. According to the requirements of different fault detection channels for fault sensitivity, combined with the parameter uncertainty model and H ∞ theory, the parameters of the sub-observers for different fault detections are optimized, and each of the sub-observers performs state estimation on the hydro-turbine governor system; wherein, the sub-observer adopts an unknown input observer. A fault diagnosis module, which is used to compare the state estimation values of each state observer with those of the reference observer to generate a residual sequence, and determine whether a fault occurs in the hydro turbine governing system according to the threshold method; Among them, the state estimation module is also used to perform the following steps: S31. For each fault detection channel, divide the fault vector of the hydro turbine governing system into two complementary sets, Set 1 and Set 2; among them, for the faults in Set 1, it is expected that the fault detection channel is sensitive to them; for the faults in Set 2, it is expected that the fault detection channel is not sensitive to them; S32. For the reference channel, set a reference unknown input observer (UIO). 0 , where Set 2 is the set containing all faults, and its state estimation value is decoupled from all faults; for each fault detection channel except the reference channel, set the corresponding unknown input observer (UIO) of the state k , whose state estimation value is decoupled from the faults in Set 2; or, the reference channel is not decoupled from any faults, and for each fault detection channel except the reference channel, set the corresponding unknown input observer (UIO) of the state k , whose state estimation value is decoupled from the faults in Set 1; S33. According to the requirements of the fault sensitivity of different fault detection channels, combined with the parameter uncertainty model and H ∞ theory, design the H 0 for the reference unknown input observer UIO ∞ to optimize the performance index. For the state unknown input observer UIO k design the H ∞ / H - to optimize the performance index; S34. Use linear matrix inequalities to solve the optimization performance index in S33, obtain the optimal values of each performance index, and solve the feedback gain matrix of each unknown input observer to complete the construction of the dedication observer; S35. Use each of the sub-observers to perform state estimation on the hydro turbine governing system.

Images