Turntable system fault diagnosis method and system based on subspace technology
By constructing the data equations of the closed-loop system and designing the residual observer, and using subspace technology for parameter identification, the problem of missed fault reporting in the turntable control system was solved, and higher fault diagnosis accuracy was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-31
- Publication Date
- 2026-04-14
AI Technical Summary
Existing fault diagnosis methods for turntable control systems based on subspace technology are prone to missing faults in closed-loop systems, especially when considering feedback loops and controller effects, where the accuracy of existing methods is insufficient.
By collecting input and output data from the turntable system and combining it with known controller information, the data equations of the closed-loop system are constructed. A residual observer based on the system stability kernel is designed, subspace technology is used for parameter identification, and a fault evaluation function is designed to achieve online fault diagnosis.
It improves the accuracy of fault diagnosis in turntable control systems, especially in near-linear time-invariant systems with controller intervention, significantly improving the accuracy of fault diagnosis.
Smart Images
Figure CN121857635A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of turntable control system fault diagnosis technology, and more specifically, to a turntable system fault diagnosis method and system based on subspace technology. Background Technology
[0002] With the continuous development of industrial technology, turntable systems are widely used in hardware-in-the-loop simulation research and development experiments in industries such as aerospace, industrial manufacturing, and machine production. However, as a common type of precision electromechanical servo system, the turntable control system inevitably experiences various malfunctions during long-term operation. These malfunctions not only affect the progress of various industrial production processes, but some sudden failures can even damage components or the entire system, leading to serious safety accidents. Therefore, various industries have placed higher demands on the reliability and safety of turntable systems.
[0003] Existing fault diagnosis methods can be categorized into three types: model-driven, data-driven, and knowledge-driven. Model-driven fault diagnosis methods require the establishment of an accurate mathematical model of the system; however, due to the complexity of large electromechanical equipment, establishing a mathematical model that reflects the system is often quite difficult. Data-driven fault diagnosis methods do not require a known mathematical model of the object system, but only a large amount of available test data. However, their strong dependence on the quality and quantity of data leads to poor stability. Since both data-driven and model-driven fault diagnosis methods have their limitations, but their advantages are complementary, it is worthwhile to research fault diagnosis methods that combine data and model approaches.
[0004] Subspace technology provides a theoretical basis for this approach. Using this technology, only simple linear algebra calculations are needed to identify system parameters from process data such as system inputs and outputs, and to obtain the system's mathematical model. However, most existing fault diagnosis methods based on subspace identification are based on open-loop model process data, neglecting the role of the controller in the closed-loop system. Furthermore, the timely suppression effect of feedback loops on faults often leads to missed fault detections. Summary of the Invention
[0005] The technical problem to be solved by this invention is:
[0006] To address the issue of missed fault reports that occur during the fault diagnosis process of applying subspace technology to closed-loop turntable control systems.
[0007] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:
[0008] This invention provides a fault diagnosis method for a turntable system based on subspace technology, comprising the following steps:
[0009] S100: Collect input and output data of the motor system, and combine it with known controller information to construct the data equations of the closed-loop system;
[0010] This process includes constructing a state-space model, obtaining the system's data equations based on the system's recursive relationships, constructing a state model based on known controller information, obtaining the controller's data equations based on the controller model's recursive relationships, and finally obtaining the data equations for the closed-loop system.
[0011] S200. Based on the system stability kernel description, design the general expression of the residual observer to clarify the parameter objects that need to be identified;
[0012] This includes constructing the general form of residual observers and constructing a stable kernel description of a closed-loop system under fault-free and noise-free conditions.
[0013] S300. Based on the closed-loop system data equations constructed in step S100, the process data is projected into the relevant subspace using subspace technology to identify the target parameters in step S200, and a residual observer for the closed-loop system is designed.
[0014] This includes obtaining an extended state-space model by combining the data equations of the controller and the system, requiring that the projection basis selected in the projection is uncorrelated with the noise signal of the system but correlated with the reference signal of the controller; when identifying the data-driven stable kernel description of the closed-loop system, this includes collecting the system input and output data and constructing Hankel matrices for past and future data, constructing auxiliary variables based on the known information of the controller, performing LQ decomposition on the matrix group constructed by the auxiliary variables, obtaining the stable kernel description of the closed-loop system based on the expression of the auxiliary variables, and finally obtaining the residual observer of the closed-loop system;
[0015] S400. Based on the residual observer designed in step S300, collect residual information during the normal operation of the system, establish a fault evaluation function and set a fault threshold.
[0016] S500: Based on the residual observer designed in step S300, monitor the residual information of the turntable system in real time, and compare it with the fault threshold in step S400 to determine whether a fault has occurred in the system.
[0017] Compared with the prior art, the beneficial effects of the present invention are:
[0018] This invention aims to solve the fault detection problem of a turntable control system under controller operation, and is also applicable to solving problems of approximately linear time-invariant systems with known controller information. First, a structural block diagram of the control system is established, system input and output data are collected, and data equations for the closed-loop system are constructed based on the known controller information. Second, the concept of a system stability kernel description is given, and a general expression of the residual observer is designed based on this core, clarifying the parameter objects that need to be identified. Then, based on the equivalence relationship between LQ decomposition and spatial projection, auxiliary variables are constructed... and The system identifies the stable core description of the system; finally, it collects residual information under normal system conditions, sets fault evaluation functions and fault thresholds based on statistical methods, and completes online fault diagnosis of the control system.
[0019] This invention separates the controller model and the system model by combining known controller information and reselecting the projection subspace. Simultaneously, it does not require system model information, but directly identifies the system's stable kernel description using a data-driven approach, and designs a residual observer based on this. When applied to a turntable control system, this invention significantly improves the accuracy of fault diagnosis results compared to direct open-loop system identification methods. Attached Figure Description
[0020] Figure 1 This is a flowchart of a turntable system fault diagnosis method based on subspace technology in an embodiment of the present invention;
[0021] Figure 2 This is a block diagram of the turntable control system based on position feedback in an embodiment of the present invention;
[0022] Figure 3 This is a flowchart of the algorithm for identifying the stable kernel of a closed-loop system in an embodiment of the present invention. Detailed Implementation
[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0024] Specific Implementation Plan 1: Combining Figures 1 to 3 As shown, this invention provides a fault diagnosis method for a turntable system based on subspace technology, comprising the following steps:
[0025] S100. Establish the block diagram of the turntable control system. Taking a motor system failure as an example, collect the input and output data of the motor system, and combine it with the known controller information to construct the data equations of the closed-loop system. The specific process is as follows:
[0026] Combination Figure 2As shown, the turntable control system includes an industrial computer, a motor system, and a feedback system. The driver drives the motor based on the control signals from the industrial computer. The feedback system mainly refers to the system's angle measuring element, which is responsible for reading the motor's angle information and transmitting it to the industrial computer to form a closed-loop control system. The controller system and feedback system have low failure frequency and are easy to detect, while the motor system has a complex structure and many electrical components, making it the main area where common failures are concentrated. Moreover, failures in this part are not easily detected in the early stages, posing certain hidden dangers to the safety and reliability of the system. Therefore, the fault diagnosis method of the turntable control system is mainly designed and implemented for the motor part. Therefore, the input and output data collected are from this part, specifically the output voltage signal of the controller and the position signal measured by the feedback system.
[0027] The motor part can be approximated as a linear time-invariant system, therefore the state-space model can be described as:
[0028] (1)
[0029] in, These are the state variables of the system; Output signals to the controller; To measure the position signal; The coefficient matrix, which is compatible with the system dimension, is an unknown quantity in this invention; the process noise w(k) and the measurement noise v(k) are uncorrelated and are white noise signals with a mean of 0; according to... Given the system's recurrence relation for length, we can obtain the system's data equations:
[0030] （2）
[0031] in, Position output signals Controller output signal and state variables The Hankel matrix, For noise terms, Let be the Hankel matrix of the process noise signal and the measurement noise signal of the motor system, considered as zero-mean white noise signal, and independent of the controller reference signal; taking the input signal as an example, the data stacking length is selected as . Then the Hankel matrix of the corresponding signal is:
[0032] （3） （4）
[0033] in, It is usually an integer much larger than the system order; The input signal sequence;
[0034] It is the system's extended observable matrix:
[0035] （5）
[0036] in, , It is the lower triangular Topplitz matrix of the system, expressed as:
[0037] （6）
[0038] Known controller information can be described using a state-space model:
[0039] （7）
[0040] in, These are the state variables of the controller; It serves as the controller reference signal, satisfying the requirement of independence from both state variables and noise signals; The state-space parameter matrix of the controller is considered a known quantity in this invention since the controller information is known. The designed closed-loop system meets the requirements of well-posedness, internal stability, and sustainable excitation conditions.
[0041] Similarly, based on the recursive relationship of the controller model, the following controller data equations can be obtained:
[0042] （8）
[0043] in, These are the reference signals. and controller state variables The Hankel matrix; For the extended observable matrix of the controller:
[0044] （9）
[0045] The lower triangular Toplitz matrix of the controller:
[0046] （10）
[0047] Furthermore, the data equations for the closed-loop system are as follows:
[0048] （11）
[0049] S200. Based on the system stability kernel description, design the general expression of the residual observer to clarify the parameter objects that need to be identified;
[0050] This invention relates to a data-driven fault diagnosis method, primarily referring to a method for obtaining system residual information based on the motor input / output data collected in step S100. Therefore, the general form of the designed residual observer is as follows:
[0051] （12）
[0052] in, For data-driven system stable kernel description, It is a residual signal sequence. and The input and output signal sequence;
[0053] Formula (12) states that the input of the residual observer is the system input and output data, and the output of the residual observer is the residual signal. According to the actual requirements of fault detection, the residual signal should be approximately 0 when the system is fault-free and noise-free. Therefore, the matrix that satisfies this condition in the closed-loop system is called the matrix that satisfies this condition. For a stable kernel description of a closed-loop system, the following conditions must be met when the system is fault-free and noise-free:
[0054] （13）
[0055] The stability kernel description of the closed-loop system contains most of the system model information, which is the key to designing the residual observer and is also the parameter object that needs to be identified in this invention.
[0056] S300. Based on the closed-loop system data equations constructed in step S100, the process data is projected into the relevant subspace using subspace technology to identify the target parameters in step S200, and a residual observer for the closed-loop system is designed.
[0057] From the system's data equation formula (2), it can be seen that the system's basic information is contained in the extended observability matrix. or state Hankel matrix In the middle, that is, the first term on the right side of the equation; the core of subspace identification technology is to find suitable projection variables. By Projection to projection variable Eliminate by the subspace it is located in and The information is used to extract system information from the data matrix; based on this, the stable kernel description of the closed-loop system is described. Criteria for selecting projection variables in the middle:
[0058] Combining the data equations of the controller and the system, let The extended state-space model is obtained:
[0059] （14）
[0060] Where I is a unit vector;
[0061] Consider selecting projection variables Projecting formula (14) onto In the line space, in order to... The orthogonal subspaces of the column space contain only system model information, not controller model information. Therefore, the first block row of the second term on the right-hand side of the projected equation must be zero, and the second block row must be non-zero. That is:
[0062] （15）
[0063] This means that the projection substrate selected for projection must be uncorrelated with the system's noise signal but correlated with the controller's reference signal.
[0064] Based on this, combined Figure 3 The algorithm for identifying the stable kernel description of a closed-loop system is as follows:
[0065] S310, Data Acquisition System Input Output data Select the length of past data and future data length Construct the Hankel matrix of past data and future data Hankel matrix ;
[0066] To fully stimulate the system's characteristics, the reference signal needs to satisfy the sustainable excitation condition during the identification process, namely:
[0067] （16）
[0068] Sustainable incentive conditions require that the choice of data length meet certain conditions. To ensure the matrix It is the full term of the rank;
[0069] S320. Construct auxiliary variables based on the known information from the controller. and (The method is not unique). Based on the above requirements for the projection variable, the projection variable is selected as follows: and :
[0070] （17） （18）
[0071] S330, will Along Project to In the row space, numerical computation can be equivalent to processing a group of matrices. Perform LQ decomposition:
[0072] （19）
[0073] S340, Obtaining Matrix Blocks left null space matrix :
[0074] （20）
[0075] To reduce computational complexity, the left null space matrix can be calculated using singular value decomposition. The specific process is as follows:
[0076] （21） （22）
[0077] S350. Based on the expression (11) of the auxiliary variable, the relation can be obtained as follows:
[0078] （23）
[0079] in, This represents the noise signal contained in the system. When the system is noise-free and fault-free, this term is zero. Therefore, the stability kernel of the closed-loop system is described as follows:
[0080] （24）
[0081] Based on the identification results, the residual observer for the closed-loop system is designed as follows:
[0082] （25）
[0083] S400. Based on the residual observer designed in step S300, collect residual information during the normal operation of the system, establish a fault evaluation function and set an appropriate fault threshold.
[0084] For a system that may actually have faults, if both the process noise and measurement noise of the system follow a normal distribution, then the residual signal of the system contains two parts: fault information and noise information.
[0085] （26）
[0086] in, This is the system noise signal; This indicates a potential fault signal in the system; residual signals can be obtained by collecting input and output data during normal system operation, for reference. The test method establishes a fault evaluation function. for:
[0087] （27）
[0088] in, The covariance matrix of the residual signal during normal operation is calculated as follows:
[0089] （28）
[0090] Furthermore, based on the significance level required for fault detection... Set fault thresholds :
[0091] （29）
[0092] S500: Monitor the residual information of the turntable system in real time using the residual observer designed in step S300, and compare it with the fault threshold in step S400 to determine whether a fault has occurred in the system.
[0093] Real-time acquisition of system input and output data, acquisition of residual information through residual observer, calculation of system fault assessment value according to formula (27), comparison with fault threshold, and system fault detection according to the following formula:
[0094] （30）
[0095] Specific Implementation Method Two: The present invention provides a turntable system fault diagnosis system based on subspace technology. This system has a program module corresponding to the above steps, and executes the steps in the above-mentioned turntable system fault diagnosis method based on subspace technology when running.
[0096] The other combinations and connections in this implementation scheme are the same as in Specific Implementation Scheme 1.
[0097] Specific Implementation Method Three: This invention provides a computer-readable storage medium storing a computer program configured to, when invoked by a processor, implement the steps of a turntable system fault diagnosis method based on subspace technology. Other combinations and connections in this implementation method are the same as in Specific Implementation Method One.
[0098] While the present invention has been disclosed above, its scope of protection is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, and all such changes and modifications will fall within the scope of protection of the present invention.
Claims
1. A fault diagnosis method for a turntable system based on subspace technology, characterized in that, Includes the following steps: S100: Collect input and output data of the motor system, and combine it with known controller information to construct the data equations of the closed-loop system; This process includes constructing a state-space model, obtaining the system's data equations based on the system's recursive relationships, constructing a state model based on known controller information, obtaining the controller's data equations based on the controller model's recursive relationships, and finally obtaining the data equations for the closed-loop system. S200. Based on the system stability kernel description, design the general expression of the residual observer to clarify the parameter objects that need to be identified; This includes constructing the general form of residual observers and constructing a stable kernel description of a closed-loop system under fault-free and noise-free conditions. S300. Based on the closed-loop system data equations constructed in step S100, the process data is projected into the relevant subspace using subspace technology to identify the target parameters in step S200, and a residual observer for the closed-loop system is designed. This includes obtaining an extended state-space model by combining the data equations of the controller and the system, requiring that the projection basis selected in the projection is uncorrelated with the noise signal of the system but correlated with the reference signal of the controller; when identifying the data-driven stable kernel description of the closed-loop system, this includes collecting the system input and output data and constructing Hankel matrices for past and future data, constructing auxiliary variables based on the known information of the controller, performing LQ decomposition on the matrix group constructed by the auxiliary variables, obtaining the stable kernel description of the closed-loop system based on the expression of the auxiliary variables, and finally obtaining the residual observer of the closed-loop system; S400. Based on the residual observer designed in step S300, collect residual information during the normal operation of the system, establish a fault evaluation function and set a fault threshold. S500: Based on the residual observer designed in step S300, monitor the residual information of the turntable system in real time, and compare it with the fault threshold in step S400 to determine whether a fault has occurred in the system.
2. The method for fault diagnosis of a turntable system based on subspace technology according to claim 1, characterized in that: In step S100, the motor portion can be approximated as a linear time-invariant system, therefore the state-space model is described as follows: in, These are the state variables of the system; Output signals to the controller; To measure the position signal; The coefficient matrix is compatible with the system dimension; the process noise w(k) and the measurement noise v(k) are uncorrelated and are white noise signals with a mean of 0; According to Given the system's recurrence relation for length, we obtain the system's data equations: in, Position output signals Controller output signal and state variables The Hankel matrix, For noise terms, The Hankel matrix represents the process noise signal and the measurement noise signal of the motor system. It is considered to be a zero-mean white noise signal, and is independent of the controller reference signal. For the input signal, the data stacking length is selected as... Then the Hankel matrix of the corresponding signal is: in, It is an integer much larger than the system order; The input signal sequence; It is the system's extended observable matrix: in, , It is the lower triangular Topplitz matrix of the system, expressed as: Known controller information is described using a state-space model: in, These are the state variables of the controller; It serves as the controller reference signal, satisfying the requirement of independence from both state variables and noise signals; This is the state-space parameter matrix of the controller, since the controller information is known. Based on the recursive relationship of the controller model, the following controller data equations are obtained: in, These are the reference signals. and controller state variables The Hankel matrix; For the extended observable matrix of the controller: The lower triangular Toplitz matrix of the controller: The data equations for the closed-loop system are as follows:
3. The method for fault diagnosis of a turntable system based on subspace technology according to claim 2, characterized in that: Step S200 includes the following: The general form of the residual observer is as follows: in, For data-driven system stable kernel description, It is a residual signal sequence. and The input and output signal sequence; According to the actual requirements of fault detection, under the condition that the system is fault-free and noise-free, the residual signal should be approximately 0. Therefore, a matrix that satisfies this condition in a closed-loop system is called a matrix. For a stable kernel description of a closed-loop system, the following conditions must be met when the system is fault-free and noise-free:
4. The method for fault diagnosis of a turntable system based on subspace technology according to claim 3, characterized in that: Step S300 includes combining the data equations of the controller and the system, letting The extended state-space model is obtained: Where I is a unit vector; Consider selecting projection variables Projecting formula (10) onto In the line space, to allow The orthogonal subspaces of the column space contain only system model information, not controller model information. Therefore, the first block row of the second term on the right-hand side of the projected equation must be zero, and the second block row must be non-zero. That is: This means that the projection substrate selected for projection must be uncorrelated with the system's noise signal but correlated with the controller's reference signal.
5. The method for fault diagnosis of a turntable system based on subspace technology according to claim 4, characterized in that: The steps for constructing a stable kernel description for an identifiable closed-loop system include: S310, Data Acquisition System Input Output data Select the length of past data and future data length Construct the Hankel matrix of past data and future data Hankel matrix ; To fully stimulate the system's characteristics, the reference signal needs to satisfy the sustainable excitation condition during the identification process, namely: Sustainable incentive conditions require that the choice of data length meet certain conditions. To ensure the matrix It is the full term of the rank; S320. Construct auxiliary variables based on the known information from the controller. and Based on the above requirements for projection variables, the projection variables are selected as #imgpt60# and #imgpt61#: S330. Projecting #imgpt64# along #imgpt65# into the row space of #imgpt66# can be numerically equivalent to performing LQ decomposition on the matrix group #imgpt67#: S340, Find the left null space matrix #imgpt70# of matrix block #imgpt69#: The left null space matrix is calculated using the singular value decomposition method. The specific process is as follows: S350. Based on the expression (11) of the auxiliary variable, the relation can be obtained as follows: Here, #imgpt75# represents the noise signal contained in the system. This value is zero when the system is noise-free and fault-free. Therefore, the stable kernel of the closed-loop system is described as follows: Based on the identification results, the residual observer for the closed-loop system is designed as follows:
6. The method for fault diagnosis of a turntable system based on subspace technology according to claim 5, characterized in that: Step S400 includes the following: For a system with an actual fault, if both the process noise and measurement noise of the system follow a normal distribution, then the residual signal of the system contains two parts: fault information and noise information. Where #imgpt79# represents the system noise signal; #imgpt80# represents a potential fault signal in the system; and the fault assessment function #imgpt82# is established based on the verification method in #imgpt81#: Where #imgpt84# is the covariance matrix of the residual signal during normal operation, calculated as follows: Set the fault threshold according to the significance level required for fault detection (#imgpt86#). (#imgpt87#) 7. The method for fault diagnosis of a turntable system based on subspace technology according to claim 6, characterized in that: Step S500 includes: real-time acquisition of system input and output data, obtaining residual information through a residual observer, calculating the system fault assessment value according to formula (27), comparing it with the fault threshold, and performing system fault detection according to the following formula:
8. A fault diagnosis system for a turntable system based on subspace technology, characterized in that: The system has a program module corresponding to the steps described in any one of claims 1-7, and executes the steps in the above-described method for diagnosing turntable system faults based on subspace technology when it is run.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program configured to, when invoked by a processor, implement the steps of the turntable system fault diagnosis method based on subspace technology according to any one of claims 1-7.