A steel wire rope defect detection method and system based on feature fusion and dimension reduction optimization

Abstract

The embodiment of the application discloses a steel wire rope defect detection method and system based on feature fusion and dimension reduction optimization, which comprises the following steps: acquiring a double-field metal magnetic memory signal of a steel wire rope; constructing a defect response term and a baseline drift term for the double-field metal magnetic memory signal based on a composite model construction method to obtain a metal magnetic memory signal analysis model; the parameters in each metal magnetic memory signal analysis model include nonlinear parameters and linear parameters; the value of the nonlinear parameters is searched iteratively based on an optimization algorithm; based on the value of the nonlinear parameters, the corresponding metal magnetic memory signal analysis model is characterized as a corresponding base matrix; the base matrix and the metal magnetic memory signal component are subjected to matrix pseudo-inverse operation to obtain the linear parameter estimation value corresponding to the value of each nonlinear parameter; based on a preset joint fitness function, the value of each nonlinear parameter and the corresponding linear parameter estimation value are evaluated to obtain optimal nonlinear parameter values and optimal linear parameter values, and defect information is output.

Description

Technical Field

[0001] This application relates to the fields of nondestructive testing and signal processing technology, and relates to, but is not limited to, a method and system for detecting defects in wire ropes using feature fusion and dimensionality reduction optimization. Background Technology

[0002] In heavy industry, steel wire ropes are critical load-bearing components, and their health condition directly impacts the safe operation of equipment. The surface of steel wire ropes is prone to localized damage such as broken wires and abrasion, especially clusters of defects formed by multiple adjacent damage points, posing a significant challenge to damage identification and quantitative assessment. When the spacing between these defects is extremely small, the magnetic memory signals acquired by sensors exhibit spatial overlap and interference, merging into a difficult-to-analyze single-peak envelope shape. The strong repulsive force between like magnetic poles leads to an outward shift of the peak, making the magnetic spacing much larger than the geometric spacing. Furthermore, sensor lift-off fluctuations and stress concentration within the steel wire rope introduce nonlinear macroscopic baseline drift into the signal, masking defect characteristics.

[0003] To extract defects from the aforementioned complex signals, the following approaches are mainly used in related technologies: First, traditional signal processing methods based on frequency domain filtering combined with amplitude thresholding; second, blind deconvolution algorithms are used to attempt to reconstruct the defect signals; third, one-dimensional convolutional neural network models trained with a large amount of simulation data are used for end-to-end identification; and fourth, conventional heuristic optimization algorithms are used to perform a global search on all parameters.

[0004] However, traditional filtering methods introduce phase distortion when forcibly removing the baseline, destroying subtle defect characteristics. Blind deconvolution algorithms are extremely sensitive to measured signals containing fluctuating baselines, easily producing ill-conditioned reconstructions, such as incorrectly splitting peaks. Deep learning models rely on the distribution of training data, and when faced with diverse baseline drift in measured data, they suffer from severe domain shift problems and insufficient generalization ability. Conventional heuristic optimization algorithms face the curse of dimensionality when searching all parameters simultaneously, and the dimensional differences between the two field signals can lead to scale conflicts, making the algorithm prone to getting trapped in local optima. These shortcomings make it difficult for related technologies to achieve high-precision quantitative detection of clustered defects in wire ropes. Summary of the Invention

[0005] In view of this, embodiments of this application provide a method and system for detecting defects in steel wire ropes by feature fusion and dimensionality reduction optimization, which at least solves the problem of inaccurate detection of cluster defects in steel wire ropes.

[0006] The technical solution of this application embodiment is implemented as follows: In a first aspect, embodiments of this application provide a feature fusion and dimensionality reduction optimization method for detecting defects in steel wire ropes, applied to a feature fusion and dimensionality reduction optimization system for detecting defects in steel wire ropes; the system includes a signal acquisition module, a model construction module, a parameter solving module, a parameter optimization module, and a defect output module; the method includes: The dual-field metallic magnetic memory signal of the wire rope is acquired using the signal acquisition module. Using the model building module, based on the composite model building method, a defect response term and a baseline drift term are constructed for each metal magnetic memory signal component in the dual-field metal magnetic memory signal, respectively, to obtain the corresponding analytical model of the metal magnetic memory signal; the parameters in each analytical model of the metal magnetic memory signal include nonlinear parameters and linear parameters; Using the parameter solving module, the value of the nonlinear parameter is iteratively searched based on the optimization algorithm; in each iteration search, based on the current value of the nonlinear parameter, the corresponding analytical model of the metal magnetic memory signal is represented as the corresponding basis matrix; matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metal magnetic memory signal component to obtain the linear parameter estimate value corresponding to the value of each nonlinear parameter; Using the parameter optimization module, based on a preset joint fitness function, the value of each nonlinear parameter and the corresponding estimated value of the linear parameter are evaluated to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value; Using the defect output module, defect information is output based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value.

[0007] Secondly, embodiments of this application provide a wire rope defect detection system based on feature fusion and dimensionality reduction optimization. The system includes: a signal acquisition module, a model construction module, a parameter solving module, a parameter optimization module, and a defect output module. The signal acquisition module is used to acquire the dual-field metal magnetic memory signal of the wire rope; The model building module is used to construct a defect response term and a baseline drift term for each metal magnetic memory signal component in the dual-field metal magnetic memory signal based on the composite model building method, thereby obtaining the corresponding metal magnetic memory signal analytical model; the parameters in each metal magnetic memory signal analytical model include nonlinear parameters and linear parameters. The parameter solving module is used to iteratively search for the value of the nonlinear parameter based on an optimization algorithm; in each iteration search, based on the current value of the nonlinear parameter, the corresponding analytical model of the metal magnetic memory signal is represented as the corresponding basis matrix; the matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metal magnetic memory signal component to obtain the linear parameter estimate value corresponding to the value of each nonlinear parameter; The parameter optimization module is used to evaluate the value of each nonlinear parameter and the corresponding estimated value of the linear parameter based on a preset joint fitness function, so as to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value. The defect output module is used to output defect information based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value.

[0008] The beneficial effects of the technical solutions provided in this application include at least the following: This application obtains a signal containing environmental noise and nonlinear baseline drift by acquiring the dual-field metallic magnetic memory signal of a steel wire rope. Based on the composite model construction method, a defect response term and a baseline drift term are constructed for each component of the dual-field metallic magnetic memory signal, resulting in a corresponding analytical model of the metallic magnetic memory signal. The parameters in each analytical model include both nonlinear and linear parameters. Decomposing the dual-field metallic magnetic memory signal into a defect response term corresponding to the target signal and a baseline drift term corresponding to the interference signal at the model level is a prerequisite for adaptively removing interference signals and obtaining defect information. An optimization algorithm iteratively searches for the values ​​of the nonlinear parameters. In each iteration, based on the current value of the nonlinear parameter, the corresponding analytical model of the metallic magnetic memory signal is represented as the corresponding basis matrix. A matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metallic magnetic memory signal component to obtain the estimated value of the linear parameter corresponding to each nonlinear parameter. By separating the solution of nonlinear and linear parameters, the optimization algorithm iteratively searches only in the space of nonlinear parameters after dimensionality reduction, reducing the dimensionality of the high-dimensional optimization problem and avoiding the curse of dimensionality caused by simultaneous searching of nonlinear and linear parameters, thus mitigating the risk of getting trapped in local optima in high-dimensional optimization. In each iteration, the optimization algorithm constructs a basis matrix containing the basis of the defect response term and the basis of the baseline drift term, based on the value of the currently searched nonlinear parameter. It then uses matrix pseudo-inverse operations to parsely obtain the linear parameter estimate that matches the nonlinear parameter value. This achieves fast and accurate solution for the linear parameter, and subtracting the baseline drift term avoids signal phase distortion. Based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value, defect information is output. The obtained defect information provides a quantitative list of defects for assessing the safety status of the wire rope, providing an accurate data foundation for subsequent spacing compensation and wire rope life prediction, and achieving high-precision quantitative decoupling of overlapping cluster defects. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 A schematic flowchart illustrating a feature fusion and dimensionality reduction optimization method for detecting defects in wire ropes, provided in an embodiment of this application; Figure 2 This is a schematic diagram illustrating the first effect of the feature fusion and dimensionality reduction optimization process described in the embodiments of this application; Figure 3 This is a schematic diagram illustrating the second effect of the feature fusion and dimensionality reduction optimization process described in the embodiments of this application; Figure 4 This is a schematic diagram illustrating the third effect of the feature fusion and dimensionality reduction optimization process described in the embodiments of this application; Figure 5 This is a schematic diagram of a wire rope defect detection system based on feature fusion and dimensionality reduction optimization provided in an embodiment of this application. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. The following embodiments are used to illustrate this application, but are not intended to limit the scope of this application. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0011] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0012] It should be noted that the terms "first, second, and third" used in the embodiments of this application are merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, and third" can be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0013] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which the embodiments of this application pertain. It should also be understood that terms such as those defined in general dictionaries should be understood to have a meaning consistent with their meaning in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0014] This application provides a feature fusion and dimensionality reduction optimization method for steel wire rope defect detection, which is applied to a feature fusion and dimensionality reduction optimization steel wire rope defect detection system. The system includes a signal acquisition module, a model construction module, a parameter solving module, a parameter optimization module, and a defect output module. Figure 1 This is a flowchart illustrating a feature fusion and dimensionality reduction optimization method for detecting defects in wire ropes, as provided in an embodiment of this application. Figure 1 As shown, the method includes at least the following steps: Step S110: Use the signal acquisition module to acquire the dual-field metal magnetic memory signal of the wire rope; A high-resolution Hall sensor array was used to synchronously acquire dual-field metallic magnetic memory signals, including environmental noise and nonlinear baseline drift, along the axial direction of the steel wire rope, thus obtaining the measured signals of this application. During the acquisition process, tangential and normal metallic magnetic memory signal components were acquired simultaneously, with the acquisition directions of the tangential and normal metallic magnetic memory signal components being orthogonal.

[0015] The nonlinear baseline drift is mainly caused by the lift-off fluctuations of the sensor probe and the changes in magnetization state due to uneven stress distribution within the wire rope. This drift has a large amplitude and complex shape, severely masking the defect signal.

[0016] To correlate the acquired dual-field metallic magnetic memory signal with the actual position of the wire rope, a spatial scan coordinate sequence is recorded simultaneously during acquisition. This spatial scan coordinate sequence defines the position of each sampling point along the axial direction of the wire rope.

[0017] In some embodiments, sensors may be added or removed to acquire metallic magnetic memory signals in one or more directions. When additional sensors are added, the acquired additional signals can serve as supplementary signals, enhancing the characterization capability of complex defects. For example, multiple directions may include directions at a 45-degree angle to the axial direction.

[0018] Step S120: Using the model building module, based on the composite model building method, a defect response term and a baseline drift term are constructed for each metal magnetic memory signal component in the dual-field metal magnetic memory signal, respectively, to obtain the corresponding metal magnetic memory signal analytical model; the parameters in each of the metal magnetic memory signal analytical models include nonlinear parameters and linear parameters; The composite model construction method can combine different mathematical model components into analytical expressions. The acquired dual-field metallic magnetic memory signal includes at least the metallic magnetic memory signal generated by the defect itself and the background drift signal caused by factors such as lift-off fluctuations and stress inhomogeneity. Therefore, the analytical model of the metallic magnetic memory signal is used to fully describe the superimposed signal.

[0019] In the analytical model of metallic magnetic memory signals, the defect response term represents the mathematical expression of the effective metallic magnetic memory signal generated by the local defects in the wire rope itself. The construction principle of the defect response term is based on the physical mechanism of the defective metallic magnetic memory field, specifically using a magnetic dipole model for equivalence. The defect response term can be expressed as a Gaussian function, where the spatial distribution of the metallic magnetic memory field generated by a single local defect can be approximated by a Gaussian function. Therefore, when constructing the analytical model of the metallic magnetic memory signal of multiple superimposed defects, the Gaussian response of each defect is used as a nonlinear basis, with the peak position and width corresponding to the center position and geometric dimensions of the defect, respectively.

[0020] The baseline drift term is a mathematical expression used to characterize slowly changing background interference signals caused by non-defect factors, such as sensor lift-off fluctuations and material stress inhomogeneity. The principle behind constructing the baseline drift term is to use a simple function to fit the macroscopic trend of these non-defect factors, such as a first-order polynomial function.

[0021] In the analytical model of metallic magnetic memory signals, decomposing the dual-field metallic magnetic memory signal into the defect response term corresponding to the target signal and the baseline drift term corresponding to the interference signal is a prerequisite for achieving adaptive removal of the interference signal and obtaining defect information.

[0022] Since the tangential and normal metallic magnetic memory signal components contain complementary defect physical characteristics, corresponding analytical models of metallic magnetic memory signals are established for the tangential and normal metallic magnetic memory signal components, respectively.

[0023] The parameters to be determined in the analytical model of metallic magnetic memory signals include nonlinear parameters and linear parameters. Nonlinear parameters are unknowns in the model that exhibit a nonlinear relationship with the signal values; they are parameters describing the spatial topological characteristics of the defect, such as the center position and width of the defect. The nonlinear relationship is reflected in these parameters through exponential, fractional, and other nonlinear operations. Linear parameters are unknowns in the model that exhibit a linear relationship with the signal values. These include the signal amplitude characterizing the defect signal intensity and the coefficients of the baseline drift term; for example, the coefficients of the baseline drift term can include the slope and intercept.

[0024] Step S130: Using the parameter solving module, iteratively search the value of the nonlinear parameter based on the optimization algorithm; in each iterative search, based on the current value of the nonlinear parameter, the corresponding analytical model of the metal magnetic memory signal is represented as the corresponding basis matrix; perform matrix pseudo-inverse operation on the basis matrix and the corresponding metal magnetic memory signal component to obtain the linear parameter estimate value corresponding to the value of each nonlinear parameter; The parameter solving module primarily employs a variable projection strategy to address optimization problems where some parameters are linear and others are nonlinear. This is achieved by separating the linear and nonlinear parameters to achieve dimensionality reduction. Essentially, the variable projection strategy uses Separable Nonlinear Least Squares (SNLS) to separate the nonlinear and linear parameters. The nonlinear parameters are fixed, and the corresponding linear parameters are solved using matrix pseudo-inverse operations.

[0025] Based on the optimization algorithm, the specific values ​​of the nonlinear parameters characterizing the topological features of the defect space are iteratively tried and updated; this process is called iterative search. In each iterative search, based on the currently tried fixed nonlinear parameter values, the corresponding analytical model of the metallic magnetic memory signal is equivalently re-expressed as a basis matrix composed of specific mathematical functions as column vectors. The columns of this basis matrix simultaneously contain the basis of the defect response term and the basis of the baseline drift term.

[0026] A matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metal magnetic memory signal components to analyze the linear parameter estimates corresponding to the nonlinear parameters that best match the measured signal. The linear parameter estimates include the amplitude of the defect and the slope and intercept of the baseline drift term.

[0027] By iteratively searching for nonlinear parameters, the curse of dimensionality, which occurs when optimizing both nonlinear and linear parameters simultaneously, is avoided. This reduces the dimensionality of the high-dimensional optimization problem, improving computational efficiency and convergence. In each iteration, a basis matrix is ​​obtained by constructing a basis containing the defect response term and a basis containing the baseline drift term. Then, a pseudo-inverse operation is performed to parse the linear parameters, thereby removing the baseline drift term and preventing signal distortion.

[0028] Step S140: Using the parameter optimization module, based on a preset joint fitness function, evaluate the value of each nonlinear parameter and the corresponding estimated value of the linear parameter to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value. The preset joint fitness function (hereinafter referred to as the joint fitness function) is an objective function composed of the sum of the normalized fitting residuals of the tangential field and the normal field. The smaller the value of the preset joint fitness function, the better the values ​​of the nonlinear parameters and the corresponding linear parameter estimates of the currently evaluated set of parameters fit the dual-field metallic magnetic memory signal to the analytical model.

[0029] The joint fitness function calculates a score for each set of nonlinear and linear parameters, which is the joint fitness function value. Based on this score, the guesses for the nonlinear parameters are continuously adjusted to find a set of nonlinear and linear parameters that minimizes the joint fitness function value.

[0030] By separating the solutions for nonlinear and linear parameters, the optimization algorithm only needs to iteratively search within the space of nonlinear parameters after the dimensionality is reduced. This avoids the curse of dimensionality caused by mixed searches of nonlinear and linear parameters, thus reducing the dimensionality of high-dimensional optimization problems and avoiding the risk of getting trapped in local optima in high-dimensional optimization.

[0031] In each iteration, the optimization algorithm constructs a basis matrix containing the basis of the defect response term and the basis of the baseline drift term, based on the value of the currently searched nonlinear parameter. Then, it uses matrix pseudo-inverse operations to parsely obtain linear parameter estimates that match the value of the nonlinear parameter. This achieves fast and accurate solution for the linear parameter, and the baseline drift term is subtracted during the solution process to avoid signal phase distortion.

[0032] Step S150: Using the defect output module, output defect information based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value.

[0033] The optimal nonlinear parameter provides the location and width of each defect, while the corresponding optimal linear parameter provides the signal amplitude characterizing the severity of the defect, thus outputting information about each defect.

[0034] The obtained defect information provides a quantitative list of defects for assessing the safety status of wire ropes, and provides an accurate data basis for subsequent spacing compensation and wire rope life prediction, achieving high-precision quantitative decoupling of overlapping cluster defects.

[0035] This application obtains a signal containing environmental noise and nonlinear baseline drift by acquiring the dual-field metallic magnetic memory signal of a steel wire rope. Based on the composite model construction method, a defect response term and a baseline drift term are constructed for each component of the dual-field metallic magnetic memory signal, resulting in a corresponding analytical model of the metallic magnetic memory signal. The parameters in each analytical model include both nonlinear and linear parameters. Decomposing the dual-field metallic magnetic memory signal into a defect response term corresponding to the target signal and a baseline drift term corresponding to the interference signal at the model level is a prerequisite for adaptively removing interference signals and obtaining defect information. An optimization algorithm iteratively searches for the values ​​of the nonlinear parameters. In each iteration, based on the current value of the nonlinear parameter, the corresponding analytical model of the metallic magnetic memory signal is represented as the corresponding basis matrix. A matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metallic magnetic memory signal component to obtain the estimated value of the linear parameter corresponding to each nonlinear parameter. By separating the solution of nonlinear and linear parameters, the optimization algorithm iteratively searches only in the space of nonlinear parameters after dimensionality reduction, reducing the dimensionality of the high-dimensional optimization problem and avoiding the curse of dimensionality caused by simultaneous searching of nonlinear and linear parameters, thus mitigating the risk of getting trapped in local optima in high-dimensional optimization. In each iteration, the optimization algorithm constructs a basis matrix containing the basis of the defect response term and the basis of the baseline drift term, based on the value of the currently searched nonlinear parameter. It then uses matrix pseudo-inverse operations to parsely obtain the linear parameter estimate that matches the nonlinear parameter value. This achieves fast and accurate solution for the linear parameter, and subtracting the baseline drift term avoids signal phase distortion. Based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value, defect information is output. The obtained defect information provides a quantitative list of defects for assessing the safety status of the wire rope, providing an accurate data foundation for subsequent spacing compensation and wire rope life prediction, and achieving high-precision quantitative decoupling of overlapping cluster defects.

[0036] Optionally, the dual-field metallic magnetic memory signal includes a tangential metallic magnetic memory signal component and a normal metallic magnetic memory signal component; the composite model construction method includes the magnetic dipole equivalent superposition method and the first-order linear polynomial characterization method; the defect response term includes a tangential defect response term and a normal defect response term; the baseline drift term includes a tangential baseline drift term and a normal baseline drift term; the step of constructing a defect response term and a baseline drift term for each metallic magnetic memory signal component in the dual-field metallic magnetic memory signal based on the composite model construction method to obtain the corresponding analytical model of the metallic magnetic memory signal includes: constructing the tangential defect response term for the tangential metallic magnetic memory signal component based on the magnetic dipole equivalent superposition method; constructing the tangential baseline drift term for the tangential metallic magnetic memory signal component based on the first-order linear polynomial characterization method; constructing the normal defect response term for the normal metallic magnetic memory signal component based on the magnetic dipole equivalent superposition method; and constructing the normal baseline drift term for the normal metallic magnetic memory signal component based on the first-order linear polynomial characterization method.

[0037] The tangential and normal magnetic memory signal components are orthogonal. The defect response term is constructed using the magnetic dipole equivalent superposition method, a precise characterization based on the physical mechanism of the magnetic memory field, accurately simulating the spatial superposition of multiple defect magnetic fields. The baseline drift term is constructed using a first-order linear polynomial representation method, providing a mathematical description of background interference caused by lift-off fluctuations and stress inhomogeneities. Combining the magnetic dipole equivalent superposition method and the first-order linear polynomial representation method, analytical models of the magnetic memory signals, including both target and interference signals, are established for the tangential and normal magnetic memory signal components. This achieves the separation of defect signals from noise interference, laying the model foundation for subsequent high-precision quantitative decoupling.

[0038] Optionally, the analytical model of the metal magnetic memory signal includes a tangential metal magnetic memory signal analytical model and a normal metal magnetic memory signal analytical model; the nonlinear parameters include the tangential center position and the tangential width; the linear parameters include the tangential signal amplitude, the slope of the tangential baseline drift term, and the intercept of the tangential baseline drift term; the tangential defect response term is constructed for the tangential metal magnetic memory signal component based on the magnetic dipole equivalent superposition method; the tangential baseline drift term is constructed for the tangential metal magnetic memory signal component based on the first-order linear polynomial characterization method, as shown in the following formula (1): Formula (1); in, This is the analytical model for the tangential metal magnetic memory signal. The spatial position coordinates of the tangential metal magnetic memory signal component are given. For the tangential defect response term, The number of defects. The amplitude of the tangential signal, Let i be the tangential center position of the i-th defect. It is the tangential width of the i-th defect. The spatial coordinates of the defect. For the tangential baseline drift term, The slope of the tangential baseline drift term. The intercept of the tangential baseline drift term.

[0039] The nonlinear parameters in the analytical model of tangential metallic magnetic memory signals are expressed as follows: The linear parameter is expressed as: .

[0040] Distinguishing between nonlinear parameters describing the characteristics of defect signals and linear parameters describing the intensity and baseline changes of defect signals is a prerequisite for implementing variable projection dimensionality reduction optimization. The parameter separation structure of the tangential metal magnetic memory signal analytical model further enables subsequent algorithms to transform complex nonlinear optimization problems into more easily solvable dimensionality-reduced forms, thereby improving computational efficiency and solution stability.

[0041] Optionally, the nonlinear parameters further include the normal center position and the normal width; the linear parameters further include the normal signal amplitude, the slope of the normal baseline drift term, and the intercept of the normal baseline drift term; the normal defect response term is constructed for the normal metal magnetic memory signal component based on the magnetic dipole equivalent superposition method; the normal baseline drift term is constructed for the normal metal magnetic memory signal component based on the first-order linear polynomial characterization method, as shown in the following formula (2): Formula (2); in, This is the analytical model of the normal metal magnetic memory signal. The spatial position coordinates of the normal metal magnetic memory signal component are given. For the normal defect response term, The amplitude of the normal signal, For the normal baseline drift term, The slope of the normal baseline drift term. The intercept of the normal baseline drift term is given.

[0042] The nonlinear parameters in the analytical model of the normal metal magnetic memory signal are expressed as follows: The linear parameter is expressed as: .

[0043] The normal defect response term of the normal magnetic memory signal analytical model includes nonlinear parameters, causing the model to exhibit an odd-symmetric form, consistent with the actual distribution characteristics of the normal magnetic memory field. The normal magnetic memory signal analytical model clearly defines the nonlinear and linear parameters of the normal direction. The normal and tangential magnetic memory signal analytical models exhibit different magnetic field response forms, which is beneficial for subsequent dual-field normalization optimization. By simultaneously satisfying both normal and tangential constraints with the same set of defect parameters, the accuracy and reliability of the decoupling results are improved, avoiding potential misjudgments or omissions that may occur with single-field detection.

[0044] Optionally, the basis matrix includes a tangential basis matrix and a normal basis matrix; the analytical model of the metal magnetic memory signal is characterized as the corresponding basis matrix based on the current value of the nonlinear parameter, as shown in the following formulas (3) and (4): Formula (3); in, Let be the tangential basis matrix. The basis for the tangential defect response terms from the first defect to the i-th defect, 1 and 2 are the bases of the tangential baseline drift term; Furthermore, given the nonlinear parameters, the tangential basis matrix It can also be expressed as .

[0045] Formula (4); in, Let be the normal basis matrix. The basis for the normal defect response terms of the first to the i-th defects, 1 and 1 are the bases of the normal baseline drift term.

[0046] Furthermore, given the nonlinear parameters, the normal basis matrix can also be expressed as... .

[0047] In a single iteration of the particle swarm optimization algorithm, given a fixed set of nonlinear parameters Thus, the analytical model of metallic magnetic memory signals is transformed into matrix equations. Among them, among them, include and , As a basis matrix, Here is the tangential basis matrix. Let be the normal basis matrix.

[0048] The basis matrix incorporates two types of functional basis vectors: one is a basis vector for the defect response term calculated based on the current nonlinear parameters, used to fit the defect magnetic field; the other is a basis vector for the baseline drift term, containing coordinate vectors and constant vectors, used to fit the drift of background disturbances. Integrating these two types of functional basis vectors into the basis matrix enables optimal estimation and separation of the baseline when solving for linear parameters through pseudo-inverse, avoiding signal distortion.

[0049] Optionally, the linear parameter estimates include tangential linear parameter estimates and normal linear parameter estimates; the matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metal magnetic memory signal components to obtain the linear parameter estimates corresponding to the value of each nonlinear parameter, as shown in the following formulas (5) and (6): Formula (5); in, The estimated value of the tangential linear parameter is... Let be the tangential basis matrix. for The transpose of the matrix, The tangential metal magnetic memory signal component; Formula (6); in, The estimated value of the normal linear parameter is... Let be the normal basis matrix. for The transpose of the matrix, The normal direction of the metal magnetic memory signal component is referred to as .

[0050] The parameters are separated into nonlinear and linear parameters using the separable nonlinear least squares method, and a step-by-step solution strategy is adopted. The separable nonlinear least squares method performs iterative searches only in the nonlinear parameter space using an optimization algorithm. In each iteration, the current guessed value of the nonlinear parameter is fixed, transforming the nonlinear analytical model of the metal magnetic memory signal into a linear model of the linear parameters, i.e., a matrix equation. The optimal linear parameter estimates are then obtained through matrix pseudo-inverse operations.

[0051] Therefore, the separable nonlinear least squares method transforms the joint optimization problem of high-dimensional mixed parameters into an alternating process of low-dimensional nonlinear search and linear analytical solution, which not only avoids the curse of dimensionality but also completes the distortion-free removal of the baseline drift term.

[0052] Optionally, based on a preset joint fitness function, the value of each nonlinear parameter and the corresponding estimated value of the linear parameter are evaluated to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value, including: generating a dual-field metallic magnetic memory fitting signal based on the value of each nonlinear parameter and the corresponding estimated value of the linear parameter; calculating a dual-field normalized residual based on the dual-field metallic magnetic memory signal and the dual-field metallic magnetic memory fitting signal; adding the dual-field normalized residuals to obtain a joint fitness function value; and searching for the value of the nonlinear parameter and the corresponding estimated value of the linear parameter based on the optimization algorithm with the goal of minimizing the joint fitness function value to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value.

[0053] Since the tangential and normal sensitivities differ, resulting in dimensional differences, a pre-defined joint fitness function based on standard deviation normalization is constructed, as shown in formula (7): Formula (7); in, This is the tangential metal magnetic memory signal component. The normal direction of the metal magnetic memory signal component. The standard deviation of the tangential metal magnetic memory signal component. This represents the standard deviation of the normal direction of the metal magnetic memory signal component. The standard deviation reflects the intensity of the metal magnetic memory signal. This represents the joint fitness function value. Let L2 be the norm of the tangential residual. To the current nonlinear parameter Estimates of the tangential linear parameters under the given conditions. Let L be the norm of the normal residual. To the current nonlinear parameter Estimates of the normal linear parameters under the given conditions.

[0054] The L2 norm of the tangential residual is a holistic measure of the difference between the tangential theoretical signal reconstructed using the fixed values ​​of the nonlinear parameters and the corresponding linear parameter estimates, and the tangential metallic magnetic memory signal component, under the condition of a fixed set of nonlinear parameter values.

[0055] The L2 norm of the normal residual is a holistic measure of the difference between the theoretical normal signal reconstructed using the values ​​of the nonlinear parameters and the corresponding linear parameter estimates, and the normal metallic magnetic memory signal component, under the condition of fixing the values ​​of the same set of nonlinear parameters.

[0056] Dividing the L2 norm of the tangential residual by the standard deviation of the tangential metallic magnetic memory signal component, and dividing the L2 norm of the normal residual by the standard deviation of the normal metallic magnetic memory signal component, transforms the absolute residuals of the two fields into relative residuals based on their respective fluctuation amplitudes, thus achieving normalization. This ensures that the fitting errors of the two fields are on the same scale, eliminating the differences in physical dimensions and inherent fluctuation amplitudes between the two field signals.

[0057] The optimization algorithm iteratively adjusts the values ​​of nonlinear parameters to minimize the joint fitness function value. When the optimization algorithm converges and finds the set of nonlinear parameter values ​​that minimize the joint fitness function value, the optimal linear parameter values ​​corresponding to this set of nonlinear parameter values ​​are also determined.

[0058] Optionally, the step of outputting defect information based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value includes: decoupling the optimal nonlinear parameter value into the optimal center position and optimal width of each defect; decoupling the optimal linear parameter value into the optimal signal amplitude, the slope of the optimal baseline drift term, and the intercept of the optimal baseline drift term for each defect; and determining and outputting defect information based on the optimal center position, optimal width, optimal signal amplitude, the slope of the optimal baseline drift term, and the intercept of the optimal baseline drift term for each defect.

[0059] The optimal nonlinear parameter values ​​are decoupled. The optimal nonlinear parameter is broken down into the optimal center position and optimal width for each independent defect. The optimal center position is the coordinate of the defect along the length of the wire rope, and the optimal width characterizes the geometric dimensions of the defect or the range of influence of its magnetic field signal.

[0060] The optimal linear parameter values ​​corresponding to the aforementioned optimal nonlinear parameters are decoupled. The linear parameters contain both the intensity information of the target defect and the correction information for background interference. Therefore, the optimal linear parameters are decoupled into two parts: the first part is the optimal signal amplitude for each defect, quantifying the intensity of the metallic magnetic memory field caused by the defect, which is the core indicator for assessing the severity of the defect; the second part is the slope and intercept of the optimal baseline drift term of the signal in the current detection segment. The slope and intercept of the optimal baseline drift term describe the mathematical form of the background interference that was successfully estimated and ultimately removed during the solution process. The output results can be used for system status monitoring and data quality verification.

[0061] The parameters obtained from the above decoupling, namely the optimal center position, optimal width and optimal signal amplitude of each defect, are integrated to determine and output complete defect information.

[0062] The output defect information can be a structured defect list or an inspection report, where each record in the defect list or inspection report corresponds to the center position, width, and signal strength of an identified real defect.

[0063] In some embodiments, the method further includes: calculating the magnetic spacing between adjacent defects based on the optimal center position corresponding to each defect; and inverting the actual spacing between adjacent defects based on the magnetic spacing.

[0064] Specifically, after decoupling the defect parameters, the magnetic distance between two adjacent defects is obtained by subtracting their optimal center coordinates. This magnetic distance is directly measured from the magnetic field signal waveform. Due to the repulsive force between the like magnetic poles of adjacent defects, which causes a wave crest shift effect, the magnetic distance may be larger than the actual physical distance between the defects on the wire rope, i.e., the geometric distance. Using the obtained precise magnetic distance value, combined with a physical model describing the wave crest shift effect or a predetermined mapping relationship, the geometric distance between defects that is not directly observed can be inverted, calculated, and quantitatively evaluated, thereby achieving a more physically accurate characterization of the spatial distribution of defects.

[0065] In summary, this application addresses key challenges in the quantitative detection of cluster defects in wire ropes through three core aspects. First, it replaces traditional high-dimensional blind search with a variable projection dimensionality reduction strategy. This method distinguishes between nonlinear spatial topological parameters and linear amplitude and baseline parameters, allowing the optimization algorithm to search only within the significantly reduced-dimensional nonlinear parameter space. This avoids the risk of getting trapped in local optima in high-dimensional optimization, improving solution efficiency and global convergence. Second, it replaces traditional static pre-filtering with dynamic adaptive baseline stripping. By embedding the functional basis representing baseline drift into the basis matrix and utilizing matrix pseudo-inversion, the optimal baseline matching the current defect topology can be calculated in each iteration. This mechanism completes baseline estimation and subtraction during the solution process, achieving signal reconstruction with zero phase distortion and avoiding the destruction of weak defect characteristics by traditional independent filtering. Third, it replaces single-field fitting or simple weighted fusion with a dual-field normalization physical consensus mechanism. By independently normalizing the fitting residuals using the standard deviations of the tangential and normal magnetic memory signal components, the inherent differences in physical dimensions and sensitivity between the two-field signals are eliminated, enabling the joint fitness function to fairly evaluate the fitting effect of the two fields. This mechanism forces the optimization result to simultaneously satisfy the physical characteristics of both the tangential and normal magnetic memory signal components, thereby achieving a reliable physical consensus between the two fields and greatly improving the accuracy and robustness of defect localization.

[0066] Based on the above, this application has the following beneficial effects: First, the method of this application possesses baseline immunity and zero phase distortion characteristics. When facing measured signals of wire ropes with severe nonlinear drift such as tilt or sag, this method does not require any preset filtering operations that may destroy signal characteristics. By dynamically and accurately stripping baseline interference within a mathematical framework, the original morphology and topological features of secondary weak signals in overlapping defects are completely preserved, thereby overcoming the ill-conditioned reconstruction problem. Second, this application solves the domain offset vulnerability of pure data-driven artificial intelligence algorithms. The method of this application exhibits excellent adaptability and robustness when facing complex sensor lift-off fluctuations and material stress inhomogeneity interference in actual engineering. Its decoupling accuracy and reliability far exceed those of pure data-driven models that rely on specific training data distributions. Third, this application breaks through the physical resolution limit with high-precision quantitative detection capabilities and can accurately verify physical laws. The algorithm possesses extremely strong super-resolution reconstruction capabilities, successfully separating clustered broken wire defects with extremely small geometric spacing or where the original signal has fused into a single peak. It can also accurately and quantitatively measure the outward shift of magnetic peaks caused by the repulsive effect of like magnetic poles. This high-precision quantitative decoupling result provides a theoretical and technical basis for the accurate quantitative evaluation of non-destructive testing of steel wire ropes.

[0067] The following describes a feature fusion and dimensionality reduction optimization method for detecting defects in steel wire ropes, based on specific embodiments. These specific embodiments are merely for illustrating the invention and do not constitute an undue limitation thereof.

[0068] The testing object in Embodiment 1 of this application is a steel wire rope sample with two internally pre-fabricated broken wires with extremely small spacing. The actual geometric center distance between the two broken wires is 50.00 mm. During the testing process, the tangential and normal magnetic memory signal components acquired after probe scanning experienced severe mutual interference, visually merging into a single large signal envelope, and exhibiting a downward-sloping nonlinear macroscopic baseline drift. After the testing system is started, a variable projection dimensionality reduction optimization process is executed. The optimization algorithm in this embodiment adopts the particle swarm optimization algorithm. The particle swarm optimization algorithm searches only within a four-dimensional parameter space containing the center positions and widths of the two defects. In each iteration, the baseline reconstruction module of the particle swarm optimization algorithm spontaneously calculates the current optimal baseline slope and intercept by calling matrix operations, thereby eliminating baseline interference in the signal. After dozens of efficient iterations, the particle swarm optimization algorithm quickly converges and outputs the decoupling result.

[0069] Figure 2 This is a schematic diagram illustrating the first effect of the feature fusion and dimensionality reduction optimization process described in the embodiments of this application. Figure 2 As shown, Figure 2 Part (a) is "H" xThe "Extremum Joint Extraction" result figure illustrates the decoupling process of two adjacent defects along the tangential (x-direction) metallic magnetic memory signal component. The horizontal axis represents the "absolute physical position" in millimeters (mm), with a scale range of approximately 120mm to 200mm, indicating the scanning position along the length of the wire rope, i.e., the spatial coordinate variable of the tangential metallic magnetic memory signal component. The vertical axis represents the "H" value. x ", which is the intensity of the tangential metallic magnetic memory signal, measured in amperes per meter (A / m), with a value range of approximately -600 A / m to -350 A / m. Figure 2 The gray scatter points in the figure correspond to the legend "Original H". x The figure shows the measured tangential metallic magnetic memory signal component, including noise and baseline drift. The green dashed line corresponds to the legend "Baseline," representing the optimal baseline drift term calculated using the method of this application. The red and blue dashed lines correspond to the legends "Defect 1" and "Defect 2," respectively, representing the separated defect response terms corresponding to the two independent wire breakage defects, with shapes conforming to Gaussian distribution characteristics. The black solid line corresponds to the legend "Joint Fitting," which is the optimal fitting curve formed by superimposing the baseline drift term, the tangential defect response term of Defect 1, and the tangential defect response term of Defect 2. The high degree of agreement with the original tangential metallic magnetic memory signal component proves the accuracy of the model and parameter solution. The values ​​"153.5" and "188.1" (unit: mm) marked in the figure are the magnetic field response center positions of the two defects extracted by this method. The difference between the two, 51.78 mm, is the measured magnetic spacing, reflecting the physical effect of "peak outward shift" caused by the repulsion of like magnetic poles. "data1" and "data2" in the legend are the identifiers of the data areas shown in the figure.

[0070] Figure 2 Part (b) is "H" z The "zero-crossing joint extraction" result image provides a complementary analysis view of the same set of defects on the normal (z-direction) metallic magnetic memory signal components. The horizontal axis ranges from approximately 100 mm to 200 mm. The vertical axis H... z The intensity of the normal magnetic memory signal component is expressed in A / m, with a range of approximately -150 A / m to 300 A / m. Gray dots represent the measured normal magnetic memory signal component. The green dashed line corresponds to the "baseline" in the diagram, representing the estimated normal baseline drift term in the normal magnetic memory signal component. The red and blue solid lines correspond to the "Defect 1" and "Defect 2" in the diagrams, respectively, representing the independent defect response terms of the two decoupled defects in the normal direction, with their waveforms exhibiting zero-crossing characteristics near the defect center. The black solid line corresponds to the "Joint Fitting" in the diagram, representing the superposition and fitting result of the normal baseline drift term, the normal defect response term of Defect 1, and the normal defect response term of Defect 2 in the normal direction. Figure 2In part (b), the short red dashed line points vertically to the horizontal coordinate "153.5mm," marking the zero-crossing characteristic position of "Defect 1" in the normal metal magnetic memory signal component. This is consistent with... Figure 2 The extreme positions of the tangential metallic magnetic memory signal components in part (a) corroborate each other, demonstrating the validity of this application. There are also corresponding feature markings near "Defect 2".

[0071] Figure 2 part (a) and Figure 2 Part (b) demonstrates that the method of this application can remove baseline interference from the metal magnetic memory signal and separate the independent magnetic field response of two closely adjacent defects, thereby achieving high-precision extraction of the defect center position.

[0072] Therefore, the method in Embodiment 1 of this application successfully stripped the baseline and extracted the originally hidden secondary defects from the main wave peak. The calculated center-to-center distance between the two defects was 51.78 mm, a result consistent with the physical law of wave peak outward shift caused by magnetic pole repulsion.

[0073] In the comparative experiment, when processing the same measured data, the traditional blind deconvolution algorithm failed because it could not handle the downward-sloping baseline, and incorrectly split the main peak and calculated a defect spacing of 22.50 mm; while the unadjusted one-dimensional convolutional neural network model produced a domain shift due to the interference of the measured baseline, and its predicted defect spacing was 30.00 mm, resulting in serious distortion.

[0074] The test object in Embodiment 2 of this application is a second steel wire rope sample with two broken wires prefabricated inside, and the actual geometric center distance between the two broken wires is 30.00 mm. Figure 3 This is a schematic diagram illustrating the second effect of the feature fusion and dimensionality reduction optimization process described in the embodiments of this application. For example... Figure 3 As shown, Figure 3 part (a) and Figure 3 The legend, curve composition, and physical meaning of part (b) are related to Figure 2 The above-described results demonstrate the decoupling of two more closely spaced defects with a geometric center distance of 30.00 mm in both the tangential and normal metallic magnetic memory signal components. The center-to-center distance between the two defects calculated using the method of this application is 32.44 mm. This result conforms to the wave peak shift law caused by the repulsion of like magnetic poles. Therefore, the method of this application has high-precision decoupling capability for clustered defects with smaller geometric spacing.

[0075] The test object in Embodiment 3 of this application is a third wire rope sample with two broken wires prefabricated inside, the actual geometric center distance between the two broken wires is 50.00 mm, and the third wire rope sample is different from the first one. Figure 4This is a schematic diagram illustrating the third effect of the feature fusion and dimensionality reduction optimization process described in the embodiments of this application. Figure 4 As shown, Figure 4 part (a) and Figure 4 The legend, curve composition, and physical meaning of part (b) are related to Figure 2 and Figure 3 The above-described results demonstrate the decoupling of two other defects with a geometric center distance of 50.00 mm in the tangential and normal magnetic memory signal components. Using the method described in this application, the center-to-center distance between the two defects was calculated to be 55.34 mm. This result verifies that the proposed method can stably and reliably extract defect features from background interference and accurately measure the magnetic distance under different defect spacing conditions, demonstrating the generalization performance of the proposed method.

[0076] Figures 2-4 In the diagram, Hx on the horizontal axis represents the tangential metal magnetic memory signal component, and the corresponding analytical model for the metal magnetic memory signal is as described in the embodiments of this application. The H-axis label represents the normal direction of the metal magnetic memory signal component, and the corresponding analytical model of the metal magnetic memory signal is as described in the embodiments of this application. .

[0077] Figure 5 This is a schematic diagram of the structure of a wire rope defect detection system based on feature fusion and dimensionality reduction optimization provided in an embodiment of this application, as shown below. Figure 5 As shown, this application proposes a wire rope defect detection system based on feature fusion and dimensionality reduction optimization. The system 500 includes: a signal acquisition module 510, a model construction module 520, a parameter solving module 530, a parameter optimization module 540, and a defect output module 550. The signal acquisition module 510 is used to acquire the dual-field metal magnetic memory signal of the wire rope. The model building module 520 is used to construct a defect response term and a baseline drift term for each metal magnetic memory signal component in the dual-field metal magnetic memory signal based on the composite model building method, thereby obtaining a corresponding metal magnetic memory signal analytical model; the parameters in each metal magnetic memory signal analytical model include nonlinear parameters and linear parameters. The parameter solving module 530 is used to iteratively search for the value of the nonlinear parameter based on an optimization algorithm; in each iteration search, based on the current value of the nonlinear parameter, the corresponding analytical model of the metal magnetic memory signal is represented as the corresponding basis matrix; the matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metal magnetic memory signal component to obtain the linear parameter estimate value corresponding to the value of each nonlinear parameter; The parameter optimization module 540 is used to evaluate the value of each nonlinear parameter and the corresponding estimated value of the linear parameter based on a preset joint fitness function, so as to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value. The defect output module 550 is used to output defect information based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value.

[0078] It should be noted that the description of the system embodiments above is similar to the description of the method embodiments above, and has similar beneficial effects. For technical details not disclosed in the system embodiments of this application, please refer to the description of the method embodiments of this application for understanding.

[0079] It should be noted that, in the embodiments of this application, if the above-mentioned feature fusion and dimensionality reduction optimization method for steel wire rope defect detection is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, or the part that contributes to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware and software combination.

[0080] Correspondingly, embodiments of this application provide a computer-readable storage medium storing a computer program thereon. When executed by a processor, this computer program implements the steps in the wire rope defect detection method of feature fusion and dimensionality reduction optimization described in any of the above embodiments. Correspondingly, embodiments of this application also provide a computer program product. When executed by a processor of an electronic device, this computer program product is used to implement the steps in the wire rope defect detection method of feature fusion and dimensionality reduction optimization described in any of the above embodiments.

[0081] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential 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 this application. The sequence numbers of the above-described embodiments are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0082] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0083] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0084] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units may be selected to achieve the purpose of the embodiments of this application according to actual needs. In addition, each functional unit in the embodiments of this application may be fully integrated into one processing unit, or each unit may be a separate unit, or two or more units may be integrated into one unit; the integrated unit may be implemented in hardware or in the form of hardware plus software functional units.

[0085] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause the device automatic test line to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0086] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined to obtain new method embodiments without conflict. The features disclosed in the several method or device embodiments provided in this application can be arbitrarily combined to obtain new method embodiments or device embodiments without conflict.

[0087] The above description is merely an embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for detecting defects in steel wire ropes based on feature fusion and dimensionality reduction optimization, characterized in that, A wire rope defect detection system applied to feature fusion and dimensionality reduction optimization; the system includes a signal acquisition module, a model construction module, a parameter solving module, a parameter optimization module, and a defect output module; the method includes: The dual-field metallic magnetic memory signal of the wire rope is acquired using the signal acquisition module. Using the model building module, based on the composite model building method, a defect response term and a baseline drift term are constructed for each metal magnetic memory signal component in the dual-field metal magnetic memory signal, respectively, to obtain the corresponding analytical model of the metal magnetic memory signal; the parameters in each analytical model of the metal magnetic memory signal include nonlinear parameters and linear parameters; Using the parameter solving module, the value of the nonlinear parameter is iteratively searched based on the optimization algorithm; in each iteration search, based on the current value of the nonlinear parameter, the corresponding analytical model of the metal magnetic memory signal is represented as the corresponding basis matrix; matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metal magnetic memory signal component to obtain the linear parameter estimate value corresponding to the value of each nonlinear parameter; Using the parameter optimization module, based on a preset joint fitness function, the value of each nonlinear parameter and the corresponding estimated value of the linear parameter are evaluated to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value; Using the defect output module, defect information is output based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value.

2. The method according to claim 1, characterized in that, The dual-field metallic magnetic memory signal includes a tangential metallic magnetic memory signal component and a normal metallic magnetic memory signal component; the composite model construction method includes the magnetic dipole equivalent superposition method and the first-order linear polynomial characterization method; the defect response term includes a tangential defect response term and a normal defect response term; the baseline drift term includes a tangential baseline drift term and a normal baseline drift term; based on the composite model construction method, a defect response term and a baseline drift term are constructed for each metallic magnetic memory signal component in the dual-field metallic magnetic memory signal to obtain the corresponding analytical model of the metallic magnetic memory signal, including: Based on the magnetic dipole equivalent superposition method, the tangential defect response term is constructed for the tangential metal magnetic memory signal component; based on the first-order linear polynomial characterization method, the tangential baseline drift term is constructed for the tangential metal magnetic memory signal component. Based on the magnetic dipole equivalent superposition method, the normal defect response term is constructed for the normal metal magnetic memory signal component; based on the first-order linear polynomial characterization method, the normal baseline drift term is constructed for the normal metal magnetic memory signal component.

3. The method according to claim 2, characterized in that, The analytical model for the metallic magnetic memory signal includes a tangential metallic magnetic memory signal analytical model and a normal metallic magnetic memory signal analytical model; the nonlinear parameters include the tangential center position and the tangential width; the linear parameters include the tangential signal amplitude, the slope of the tangential baseline drift term, and the intercept of the tangential baseline drift term; the tangential defect response term is constructed for the tangential metallic magnetic memory signal component based on the magnetic dipole equivalent superposition method; the tangential baseline drift term is constructed for the tangential metallic magnetic memory signal component based on the first-order linear polynomial characterization method, as shown in the following formula: ； in, This is the analytical model for the tangential metal magnetic memory signal. The spatial position coordinates of the tangential metal magnetic memory signal component are given. For the tangential defect response term, The number of defects. The amplitude of the tangential signal, Let i be the tangential center position of the i-th defect. It is the tangential width of the i-th defect. The spatial coordinates of the defect. For the tangential baseline drift term, The slope of the tangential baseline drift term. The intercept of the tangential baseline drift term.

4. The method according to claim 3, characterized in that, The nonlinear parameters also include the normal center position and normal width; the linear parameters also include the normal signal amplitude, the slope of the normal baseline drift term, and the intercept of the normal baseline drift term; the normal defect response term is constructed for the normal metal magnetic memory signal component based on the magnetic dipole equivalent superposition method; the normal baseline drift term is constructed for the normal metal magnetic memory signal component based on the first-order linear polynomial characterization method, as shown in the following formula: ； in, This is the analytical model of the normal metal magnetic memory signal. The spatial position coordinates of the normal metal magnetic memory signal component are given. For the normal defect response term, The amplitude of the normal signal, For the normal baseline drift term, The slope of the normal baseline drift term. The intercept of the normal baseline drift term is given.

5. The method according to claim 4, characterized in that, The basis matrix includes a tangential basis matrix and a normal basis matrix; based on the current value of the nonlinear parameter, the corresponding analytical model of the metal magnetic memory signal is represented by the corresponding basis matrix, as shown in the following formula: ； in, Let be the tangential basis matrix. The basis for the tangential defect response terms from the first defect to the i-th defect, 1 and 2 are the bases of the tangential baseline drift term; ； in, Let be the normal basis matrix. The basis for the normal defect response terms of the first to the i-th defects, 1 and 1 are the bases of the normal baseline drift term.

6. The method according to claim 5, characterized in that, The linear parameter estimates include tangential linear parameter estimates and normal linear parameter estimates; the matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metal magnetic memory signal components to obtain the linear parameter estimates corresponding to the value of each nonlinear parameter, as shown in the following formula: ； in, The estimated value of the tangential linear parameter is... Let be the tangential basis matrix. for The transpose of the matrix, The tangential metal magnetic memory signal component; ； in, The estimated value of the normal linear parameter is... Let be the normal basis matrix. for The transpose of the matrix, The normal direction of the metal magnetic memory signal component is referred to as .

7. The method according to claim 1, characterized in that, The step of evaluating the value of each nonlinear parameter and the corresponding estimated value of the linear parameter based on a preset joint fitness function to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value includes: Based on the value of each of the nonlinear parameters and the corresponding estimated value of the linear parameter, a dual-field metallic magnetic memory fitting signal is generated. Based on the dual-field metallic magnetic memory signal and the dual-field metallic magnetic memory fitted signal, calculate the dual-field normalized residual; The combined fitness function value is obtained by summing the normalized residuals of the two fields. Based on the optimization algorithm, with the goal of minimizing the joint fitness function value, the values ​​of the nonlinear parameters and the corresponding estimated values ​​of the linear parameters are searched to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value.

8. The method according to claim 1, characterized in that, The step of outputting defect information based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value includes: The optimal nonlinear parameter value is decoupled into the optimal center position and optimal width of each defect; The optimal linear parameter value is decoupled into the optimal signal amplitude, the slope of the optimal baseline drift term, and the intercept of the optimal baseline drift term for each defect. Based on the optimal center position, optimal width, optimal signal amplitude, optimal baseline drift term slope, and optimal baseline drift term intercept of each defect, defect information is determined and output.

9. A wire rope defect detection system based on feature fusion and dimensionality reduction optimization, characterized in that, The system includes: a signal acquisition module, a model building module, a parameter solving module, a parameter optimization module, and a defect output module; The signal acquisition module is used to acquire the dual-field metal magnetic memory signal of the wire rope; The model building module is used to construct a defect response term and a baseline drift term for each metal magnetic memory signal component in the dual-field metal magnetic memory signal based on the composite model building method, thereby obtaining the corresponding metal magnetic memory signal analytical model; the parameters in each metal magnetic memory signal analytical model include nonlinear parameters and linear parameters. The parameter solving module is used to iteratively search for the value of the nonlinear parameter based on an optimization algorithm; in each iteration search, based on the current value of the nonlinear parameter, the corresponding analytical model of the metal magnetic memory signal is represented as the corresponding basis matrix; the matrix pseudo-inverse operation is performed on the basis matrix and the corresponding metal magnetic memory signal component to obtain the linear parameter estimate value corresponding to the value of each nonlinear parameter; The parameter optimization module is used to evaluate the value of each nonlinear parameter and the corresponding estimated value of the linear parameter based on a preset joint fitness function, so as to obtain the optimal nonlinear parameter value and the corresponding optimal linear parameter value. The defect output module is used to output defect information based on the optimal nonlinear parameter value and the corresponding optimal linear parameter value.

Images