System modeling and fault detection method and system for hot strip rolling looper

Abstract

The present invention discloses a system modeling and fault detection method and system for a hot strip rolling looper, which relates to the technical field of industrial steel rolling process modeling and fault detection, including: establishing a state space model of the hot strip rolling looper system at multiple stable working points based on least squares identification; fuzzifying the state space model based on preset fuzzy rules to obtain a T-S fuzzy model of the hot strip rolling looper system; optimizing the T-S fuzzy model based on a neural network to establish a target system model that combines a neural network and a fuzzy kernel representation; and performing fault detection on the hot strip rolling looper system based on the residual signal sum between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system. The present invention alleviates the technical problems of the existing model-based fault detection method, namely, the difficulty in establishing an accurate mathematical model of the looper, and the low accuracy and difficulty in fault interpretation of the data-driven fault detection method.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial steel rolling process modeling and fault detection, and in particular to a system modeling and fault detection method and system for a hot strip rolling looper. Background Art

[0002] Manufacturing is a key pillar of social development and the national economy. Process industries such as petrochemicals and steel metallurgy provide crucial support for economic growth and the construction of major national projects. With increasing demands for product quality, production efficiency, and process safety, modern process industries are becoming increasingly complex and large-scale, moving towards large-scale, multi-process collaboration. For example, the hot strip rolling process comprises numerous production steps, including heating, roughing, shearing, finishing, laminar cooling, and coiling. In the finishing process, the looper between adjacent stands of the finishing mill is crucial for stable hot rolling and ensuring steel quality. Looper height control maintains a constant looping volume, maintaining a constant metal flow rate and ensuring stable hot rolling. Looper tension control maintains a constant, low strip tension during rolling, preventing steel pulling and accumulation and mitigating the effects of various disturbances on tension. Therefore, proper monitoring of the looper's operating conditions during hot rolling is essential. If a fault is not detected and handled promptly, it can spread and even trigger a chain reaction, affecting the normal operation of the entire hot rolling process. This can impact product quality at the very least, and even lead to equipment damage, environmental pollution, property loss, and casualties at the worst. Therefore, designing a suitable fault detection method to monitor the operating status of the looper in real time is of great significance to the safe and reliable operation of the hot rolling process.

[0003] In recent years, driven by both technological advancement and market expansion, research in fault detection technology has yielded significant results. Currently, the main research approaches can be categorized into two types: model-based fault detection and data-driven fault detection. Model-based fault detection uses established models to describe and predict system behavior and compares the residual signal between the model's predictions and the system's actual output to assess and determine faults. Key methods include parameter estimation, state estimation, and Kalman filtering. Data-driven fault detection, on the other hand, utilizes methods such as data mining or machine learning to extract knowledge from historical data and analyze correlations between measured signals. These methods can be broadly categorized into signal processing, machine learning, information fusion, and multivariate statistical analysis. In contrast, model-based fault detection offers strong explanatory power, high accuracy, and the absence of extensive historical data. Consequently, it has gained widespread application in the industrial sector. However, today's complex industrial processes often exhibit severe nonlinear characteristics, such as strong coupling, hysteresis, and time-varying parameters. This makes it difficult to develop accurate mathematical models, hindering the application and development of these methods. Summary of the Invention

[0004] In order to solve the above technical problems existing in the prior art, the present invention provides a method and system for system modeling and fault detection of hot strip rolling loopers. The technical solution is as follows:

[0005] On the one hand, a system modeling and fault detection method for a hot strip rolling looper is provided, the method comprising: establishing a state space model of the hot strip rolling looper system at multiple stable working points based on least squares identification; fuzzifying the state space model based on preset fuzzy rules to obtain a TS fuzzy model of the hot strip rolling looper system; optimizing the TS fuzzy model based on a neural network to establish a target system model that combines a neural network and a fuzzy kernel representation; and performing fault detection on the hot strip rolling looper system based on a residual signal sum between a model output of the target system model within a preset time period and a true output of the hot strip rolling looper system.

[0006] Furthermore, the state space model includes:

[0007]

[0008] Among them, u and y are the system input vector and output vector respectively, are the estimates of the system state vector and output vector respectively, Ai, Bi, Ci, Di are the system matrices at the stable operating point, i is the stable operating point number, and r is the residual.

[0009] Furthermore, the TS fuzzy model includes:

[0010]

[0011] Among them, μ(t) is the antecedent variable of the preset fuzzy rule, h i (μ(t)) is the fuzzy weight, and

[0012]

[0013] Furthermore, the TS fuzzy model is optimized based on a neural network, and a target system model combining a neural network and a fuzzy kernel representation is established, including: constructing a SKR observer for the hot strip rolling looper system based on the TS fuzzy model and historical data of the hot strip rolling looper system; and establishing a target system model combining a neural network and a fuzzy kernel representation by continuously training and optimizing the neural network to approximate the feedback part of the SKR observer.

[0014] Furthermore, based on the residual signal sum between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system, fault detection is performed on the hot strip rolling looper system, including: constructing a fault detection statistic based on the residual signal sum between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system; judging whether the fault detection statistic exceeds a preset detection threshold; if so, judging that a fault has occurred in the hot strip rolling looper system.

[0015] Furthermore, the fault detection statistics include:

[0016]

[0017] Wherein, J is the fault detection statistic, and r is the residual between the model output and the true output.

[0018] Furthermore, the preset detection threshold includes: a maximum value of a detection statistic obtained by training the target system model based on fault-free data.

[0019] On the other hand, a system modeling and fault detection system for a hot strip rolling looper is also provided, comprising: an establishment module, a fuzzy module, an optimization module and a detection module; wherein the establishment module is used to establish a state space model of the hot strip rolling looper system at multiple stable working points based on least squares identification; the fuzzy module is used to fuzzify the state space model based on preset fuzzy rules to obtain a TS fuzzy model of the hot strip rolling looper system; the optimization module is used to optimize the TS fuzzy model based on a neural network to establish a target system model combining a neural network and a fuzzy kernel representation; the detection module is used to perform fault detection on the hot strip rolling looper system based on the residual signal sum between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system.

[0020] On the other hand, an electronic device is provided, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method provided by the present invention when executing the computer program.

[0021] On the other hand, a computer-readable storage medium is provided, in which a program code is stored. The program code can be called by a processor to execute the method provided by the present invention.

[0022] The embodiment of the present invention provides a system modeling and fault detection method and system for a hot strip rolling looper, which combines least squares identification, fuzzy dynamic modeling, neural network and other technologies, establishes a state space model of the system at different working points through least squares identification, and then fuzzifies it. With the help of the powerful nonlinear fitting ability of the neural network, the observer feedback part in the kernel representation (SKR) of the system stability is approximated to establish a more accurate mathematical model of the system; on this basis, a statistic is constructed using the residual signal between the model output and the system output, and the statistic is compared with the threshold value obtained by training without fault data to realize fault detection of the looper system, thereby alleviating the technical problems of the difficulty in establishing an accurate mathematical model of the looper in the model-based fault detection method of the existing technology and the low accuracy and difficulty in fault interpretation of the data-driven fault detection method. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 This is a flow chart of a system modeling and fault detection method for a hot strip rolling looper provided by an embodiment of the present invention;

[0025] Figure 2 This is a production process flow chart of a strip hot rolling looper system provided according to an embodiment of the present invention;

[0026] Figure 3 This is a neural network structure diagram and a schematic diagram of the cutting and splicing process provided by an embodiment of the present invention;

[0027] Figure 4 This is a schematic diagram of a looper object provided by an embodiment of the present invention;

[0028] Figure 5 This is a diagram showing the effect of the least squares identification of a working point provided by an embodiment of the present invention on the loop angle fitting.

[0029] Figure 6 This is a strip tension fitting effect diagram of the least squares identification of a working point provided by an embodiment of the present invention;

[0030] Figure 7 is a schematic diagram of fuzzy rules provided by an embodiment of the present invention;

[0031] Figure 8 This is a loop angle fitting effect diagram of the TS fuzzy model provided by an embodiment of the present invention;

[0032] Figure 9 TS fuzzy model provided in an embodiment of the present invention is a strip tension fitting effect diagram;

[0033] Figure 10 This is a looper angle fitting effect diagram of the TS fuzzy model and the kernel representation-based model after neural network optimization provided by an embodiment of the present invention;

[0034] Figure 11 1 is a strip tension fitting effect diagram of the TS fuzzy model and the kernel representation-based model after neural network optimization provided by an embodiment of the present invention;

[0035] Figure 12 1 is a looper angle fitting error diagram of the TS fuzzy model and the kernel representation-based model after neural network optimization provided by an embodiment of the present invention;

[0036] Figure 13 1 is a strip tension fitting error diagram of the TS fuzzy model and the kernel representation-based model after neural network optimization provided by an embodiment of the present invention;

[0037] Figure 14 This is a fault detection curve diagram of the strip tension sensor provided by an embodiment of the present invention after a fault is introduced;

[0038] Figure 15 This is a fault detection curve diagram for a rolling process with large fluctuations provided by an embodiment of the present invention;

[0039] Figure 16 It is a schematic diagram of a system modeling and fault detection system for a hot strip rolling looper provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0041] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0042] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0043] Example 1

[0044] Figure 1This is a flow chart of a system modeling and fault detection method for a hot strip mill looper according to an embodiment of the present invention, which is applied to a hot strip mill looper system. Figure 2 The present invention provides a production process flow chart of a hot strip rolling looper system according to an embodiment of the present invention.

[0045] like Figure 1 As shown, the method specifically includes the following steps:

[0046] Step S102: establishing a state space model of the hot strip rolling looper system at multiple stable working points based on least squares identification.

[0047] Step S104 , fuzzifying the state space model based on preset fuzzy rules to obtain a TS fuzzy model of the hot strip rolling looper system.

[0048] Step S106 , optimizing the TS fuzzy model based on the neural network, and establishing a target system model combining the neural network and the fuzzy kernel representation.

[0049] Step S108 , performing fault detection on the hot strip rolling looper system based on the residual signal sum between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system.

[0050] In step S102, considering that the looper angle output in the looper process data is stable at about 25°, the process data is divided into three stable working points according to the size of the strip tension, and the state space model of the process at different stable working points is established using least squares identification:

[0051]

[0052] Among them, u and y are the system input vector and output vector respectively, are the estimates of the system state vector and output vector, respectively. i 、B i 、C i 、D i is the system matrix at the stable operating point, i is the stable operating point number, and r is the residual.

[0053] Then, according to the state space model at different stable working points that has been established, the model is fuzzified using appropriate preset fuzzy rules to obtain the TS fuzzy model, including:

[0054]

[0055] Among them, μ(t) is the antecedent variable of the preset fuzzy rule, h i (μ(t)) is the fuzzy weight, and

[0056]

[0057] Specifically, step S106 further includes the following steps:

[0058] Step S1061, constructing a SKR observer for the hot strip mill looper system based on the TS fuzzy model and historical data of the hot strip mill looper system;

[0059] Step S1062 , by continuously training and optimizing the neural network to approximate the feedback part of the SKR observer, a target system model combining the neural network and the fuzzy kernel representation is established.

[0060] Specifically, for a complex industrial process, assume that its true model is:

[0061]

[0062] y(t)=g(x(t),u(t))+Δ g (x(t),u(t))

[0063] Among them, x, u, y are the system state vector, input vector, and output vector respectively, f and g are known model structures, Δ f , Δ g It's the unknown part of the process.

[0064] Based on the historical data obtained from the process operation, a SKR (Stable Kernel Representation) observer for nonlinear systems can be constructed, namely:

[0065]

[0066] in, is the state estimate vector, is an estimate of the system output.

[0067] Considering that traditional observers are mainly used to estimate the behavior of certain difficult-to-measure variables within the system, and one of the purposes of this invention is to establish a mathematical description of complex industrial processes such as loopers, the above-mentioned SKR observer can be rearranged to:

[0068]

[0069] At this time, the gap between the model and the actual situation of the system is mainly reflected in the residual r. The feedback part of the SKR observer of the process can be designed through methods such as robust control to establish an accurate mathematical model of complex industrial processes such as loopers. Since the output of the model is the true output y of the process, the model can theoretically fully characterize any complex industrial process. However, considering factors such as the time-varying parameters and process uncertainty of complex industrial processes, the known parts f and g in the observer are not enough to fully describe the process, that is, there are still some unknown factors in the model at this time. In addition, the establishment of the SKR observer for nonlinear processes often requires solving the Hamilton-Jacobi equation (HJE), which is very difficult or difficult to achieve. In order to solve the above problems, the present invention approximates the feedback part of the SKR observer through a neural network, and continuously optimizes the residual r through the neural network to establish an accurate analytical model for complex industrial processes such as loopers. The specific contents are as follows:

[0070] S1, using Taylor formula expansion, the SKR observer feedback part is refined into Approximation through neural networks

[0071] S2, slices the output vector of the neural network and stacks it row by row to match The matrix output is used as the approximation of the observer feedback gain. The specific structure is as follows Figure 3 As shown;

[0072] S3, taking into account the global accuracy of the model and the high accuracy of the model at the working point, the network loss function is set as follows:

[0073]

[0074] Among them, n represents the number of working points, y j Represents each working point, h represents a nonlinear function with a local function value, which is concretized as a Gaussian function in the present invention, namely:

[0075]

[0076] Among them, σ represents the degree of localization of the Gaussian function, and k represents the importance of the additional loss term.

[0077] S4, uses the fourth-order Runge-Kutta method to solve the state differential equation and gradually updates the state, that is:

[0078]

[0079] Where h represents the update step size.

[0080] S5, use the Xavier initialization method to assign initial weights to the network, the initial state of the system Assign 0 and use the gradient descent method to update the neural network until the loss drops to a low level, at which point the model matching is high enough.

[0081] Specifically, step S108 further includes the following steps:

[0082] Step S1081, constructing a fault detection statistic based on the residual signal sum between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system;

[0083] Step S1082, determining whether the fault detection statistic exceeds a preset detection threshold; if so, determining that a fault occurs in the hot strip rolling looper system.

[0084] Specifically, fault detection statistics include:

[0085]

[0086] Where J is the fault detection statistic and r is the residual between the model output and the true output.

[0087] In an optional implementation provided by an embodiment of the present invention, the preset detection threshold includes: the maximum value of the detection statistic obtained by training the target system model based on fault-free data, that is:

[0088]

[0089] Among them, J th is the preset detection threshold.

[0090] During the online detection process, the target system model is applied to the online data to observe whether the change of J(r) exceeds the preset detection threshold to determine whether a fault has occurred in the process, that is:

[0091]

[0092] The detection statistic adopted in the present invention is applicable to general complex industrial processes, avoids the assumption that the Q statistic and the T2 statistic require the process data to obey a certain distribution, and has greater application value.

[0093] From the above description, it can be seen that the present invention provides a system modeling and fault detection method for a hot strip rolling looper, which combines least squares identification, fuzzy dynamic modeling, neural network and other technologies, establishes a state space model of the system at different working points through least squares identification, and then fuzzifies it. With the help of the powerful nonlinear fitting ability of the neural network, the observer feedback part in the kernel representation (SKR) of the system stability is approximated to establish a more accurate mathematical model of the system; on this basis, the residual signal between the model output and the system output is used to construct a statistic, which is compared with the threshold obtained by training without fault data to realize fault detection of the looper system, alleviating the technical problems that the accurate mathematical model of the looper in the model-based fault detection method of the existing technology is difficult to establish and the data-driven fault detection method has low accuracy and difficult fault interpretation.

[0094] Example 2

[0095] This example uses the real process data of the looper system during the finishing rolling process to verify the effectiveness of the system modeling and fault detection method for the hot strip rolling looper proposed in this invention. This example takes the looper between the No. 1 rolling mill and the No. 2 rolling mill during the finishing rolling process as the research object. Figure 4 As shown in the figure, the main frame speed and the spool displacement of the hydraulic control valve are used as process inputs, the angle of the looper and the tension of the slab are used as process outputs, the sampling time is 0.1s, and stable data are selected to carry out the experiment:

[0096] I. Least squares identification. The process data is divided into three stable working points according to the variation range of the strip tension during the process. The system state space model established by least squares identification at one working point is:

[0097]

[0098] The least square identification fitting effect of one working point is as follows: Figure 5 、 Figure 6 As shown. Among them, Figure 5 This is a loop angle fitting effect diagram of the least squares identification of a working point provided by an embodiment of the present invention. Figure 6 This is a strip tension fitting effect diagram of the least squares identification at a working point provided by an embodiment of the present invention.

[0099] Ⅱ, TS fuzzy modeling. Figure 7 The triangle fuzzy rule shown fuzzifies the state space model at the three established working points. The fuzzy model fitting effect is as follows: Figure 8 、 Figure 9 As shown. Among them, Figure 8 1 is a loop angle fitting effect diagram of the TS fuzzy model provided by an embodiment of the present invention. Figure 91 is a strip tension fitting effect diagram of the TS fuzzy model provided according to an embodiment of the present invention.

[0100] III. Neural network optimization. Based on the TS fuzzy model, the neural network is used to optimize the model and a system model combining neural network and fuzzy kernel representation is established. The model fitting effect is as follows: Figure 10 、 Figure 11 As shown, the model fitting error is Figure 12 、 Figure 13 As shown. Among them, Figure 10 : This is a looper angle fitting effect diagram of the TS fuzzy model and the kernel representation-based model after neural network optimization provided by an embodiment of the present invention. Figure 11 : is a strip tension fitting effect diagram of the TS fuzzy model and the kernel representation-based model after neural network optimization provided by an embodiment of the present invention, Figure 12 is a looper angle fitting error diagram of the TS fuzzy model and the kernel representation-based model after neural network optimization provided by an embodiment of the present invention, Figure 13 1 is a strip tension fitting error diagram of a TS fuzzy model and a kernel representation-based model after neural network optimization provided by an embodiment of the present invention.

[0101] In order to quantify the accuracy of the model, the following formula is used to test the model:

[0102]

[0103] After testing, the model's fitting accuracy for the loop angle and strip tension were 93.59% and 94.09% respectively, and the model accuracy was relatively high.

[0104] IV. Fault detection. Since the looping process is more sensitive to faults, in order to test the method proposed in this invention, a fault is introduced on the strip tension sensor after the 400th sampling point. The detection effect is as follows: Figure 14 Similarly, after the 400th sampling point, the data with large fluctuations in the rolling process are considered as faults, and the detection effect is as follows: Figure 15 The fault detection effect is excellent, with an accuracy rate higher than 98%.

[0105] Example 3

[0106] Figure 16 Schematic diagram of a system modeling and fault detection system for a hot strip rolling looper according to an embodiment of the present invention. Figure 16 As shown, the system includes: an establishment module 10, a fuzzy module 20, an optimization module 30 and a detection module 40.

[0107] Specifically, a module 10 is established for establishing a state space model of the hot strip rolling looper system at multiple stable working points based on least squares identification.

[0108] The fuzzy module 20 is used to fuzzify the state space model based on preset fuzzy rules to obtain the TS fuzzy model of the strip hot rolling looper system.

[0109] The optimization module 30 is used to optimize the TS fuzzy model based on the neural network and establish a target system model that combines the neural network and the fuzzy kernel representation.

[0110] The detection module 40 is used to perform fault detection on the hot strip rolling looper system based on the residual signal sum between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system.

[0111] Specifically, the optimization module 30 is also used to: construct an SKR observer for the hot strip rolling looper system based on the TS fuzzy model and historical data of the hot strip rolling looper system; and establish a target system model that combines the neural network and fuzzy kernel representation by continuously training and optimizing the neural network to approximate the feedback part of the SKR observer.

[0112] Specifically, the detection module 40 is also used to: construct a fault detection statistic based on the residual signal sum between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system; determine whether the fault detection statistic exceeds a preset detection threshold; if so, determine that a fault has occurred in the hot strip rolling looper system.

[0113] The present invention also provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the method provided by the present invention is implemented when the processor executes the computer program.

[0114] The present invention also provides a computer-readable storage medium, in which program code is stored. The program code can be called by a processor to execute the method provided by the present invention.

[0115] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0116] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0117] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0118] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0119] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0120] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0121] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0122] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A system modeling and fault detection method for hot strip rolling loopers, characterized in that: The method comprises: The state space model of the hot strip rolling looper system at multiple stable working points is established based on least square identification. Fuzzifying the state space model based on preset fuzzy rules to obtain a TS fuzzy model of the strip hot rolling looper system; Optimizing the TS fuzzy model based on a neural network, and establishing a target system model combining a neural network and a fuzzy kernel representation; Performing fault detection on the hot strip rolling looper system based on a residual signal sum between a model output of the target system model within a preset time period and a true output of the hot strip rolling looper system; The TS fuzzy model is optimized based on a neural network, and a target system model combining a neural network and a fuzzy kernel representation is established, including: constructing a SKR observer for the hot strip rolling looper system based on the TS fuzzy model and historical data of the hot strip rolling looper system; By continuously training and optimizing the neural network to approximate the feedback part of the SKR observer, a target system model combining the neural network and the fuzzy kernel representation is established; Performing fault detection on the hot strip rolling looper system based on a residual signal sum between a model output of the target system model within a preset time period and a true output of the hot strip rolling looper system, including: constructing a fault detection statistic based on the sum of residual signals between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system; Determining whether the fault detection statistic exceeds a preset detection threshold; If yes, it is determined that the strip hot rolling looper system fails.

2. The method according to claim 1, characterized in that The state space model comprises: ； Among them, u and y are the system input vector and output vector respectively, are the estimates of the system state vector and output vector respectively, Ai, Bi, Ci, Di are the system matrices at the stable operating point, i is the stable operating point number, and r is the residual.

3. The method according to claim 2, characterized in that The TS fuzzy model includes: ； Among them, μ(t) is the antecedent variable of the preset fuzzy rule, h i (μ(t)) is the fuzzy weight, and .

4. The method according to claim 1, wherein The fault detection statistics include: ； Wherein, J is the fault detection statistic, and r is the residual between the model output and the true output.

5. The method according to claim 1, wherein The preset detection threshold includes: a maximum value of a detection statistic obtained by training the target system model based on fault-free data.

6. A system modeling and fault detection system for hot strip rolling loopers, characterized in that: include: Establishment module, fuzzy module, optimization module and detection module; among them, The establishment module is used to establish a state space model of the hot strip rolling looper system at multiple stable working points based on least squares identification; The fuzzy module is used to fuzzify the state space model based on preset fuzzy rules to obtain a TS fuzzy model of the strip hot rolling looper system; The optimization module is used to optimize the TS fuzzy model based on a neural network and establish a target system model that combines a neural network and a fuzzy kernel representation; The detection module is configured to perform fault detection on the hot strip rolling looper system based on a residual signal sum between a model output of the target system model within a preset time period and a true output of the hot strip rolling looper system; The optimization module is further used to: constructing a SKR observer for the hot strip rolling looper system based on the TS fuzzy model and historical data of the hot strip rolling looper system; By continuously training and optimizing the neural network to approximate the feedback part of the SKR observer, a target system model combining the neural network and the fuzzy kernel representation is established; The detection module is further used for: constructing a fault detection statistic based on the sum of residual signals between the model output of the target system model within a preset time period and the actual output of the hot strip rolling looper system; Determining whether the fault detection statistic exceeds a preset detection threshold; If yes, it is determined that the strip hot rolling looper system fails.

7. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, which can be called by a processor to execute the method according to any one of claims 1 to 5.

Images