Babbitt metal coating surfacing thin-wall layer defect detection method and device

Abstract

The invention relates to the technical field of defect detection, and discloses a Babbitt metal coating surfacing thin-wall layer defect detection method and device. The method comprises the following steps: carrying out multi-source detection data acquisition, multi-domain transformation and filtering enhancement on a babbitt metal coating, and obtaining preprocessed multi-source initial coating detection data to carry out interface-tissue-defect mapping, so as to obtain unified representation of an internal tissue structure of the coating; performing progressive fusion of a coating depth structure and extraction of coating defect morphology and structural features on the unified feature representation to obtain preliminary defect semantic features, and performing feature weighting calculation of defect morphology on the preliminary defect semantic features to obtain initial defect semantic features; and carrying out bearing strength and shear strength calculation and fatigue life prediction on the initial defect semantic features to obtain a final detection result. According to the method, the defects of the thin-wall layer coated and surfaced by the babbitt metal can be efficiently and accurately detected.

Description

Technical Field

[0001] The present invention relates to the technical field of defect detection, and in particular to a method and a device for detecting defects in a thin-wall layer of babbitt alloy coating and surfacing. Background Art

[0002] In the field of babbitt coating cladding technology, defect detection is an important part of product quality control and performance evaluation, and the real-time monitoring and evaluation of thin-wall layer defects is the most critical and challenging stage in the detection process. Accurate detection of the defect status of babbitt coating is crucial for manufacturing companies, R&D institutions and testing centers, as it directly affects the reliability evaluation, performance optimization and service life prediction of the coating. Therefore, an effective defect detection method to monitor the internal structure and defect evolution of the coating is essential to ensure the structural integrity and service safety of the babbitt coating.

[0003] Nowadays, single detection methods such as ultrasound, infrared thermal imaging, eddy current or X-ray are generally used for defect detection, or artificial intelligence technology is tried to be applied to the defect recognition process to improve the accuracy of detection. However, these methods still have challenges in integrating multi-source detection data, processing nonlinear characteristics and adapting to complex defect morphology, and these detection methods often ignore some unique characteristics of babbitt alloy coatings, such as interface bonding state, structural uniformity, residual stress distribution, etc., which may have a significant impact on the formation and development of defects. That is, the existing babbitt alloy coating defect detection method ignores the fusion of multi-source data and the unique characteristics of the coating, resulting in low reliability of the final defect detection results. Summary of the invention

[0004] The main purpose of the present invention is to solve the problem that the existing Babbitt alloy coating defect detection method ignores multi-source data fusion and the unique characteristics of the coating, resulting in low reliability of the final defect detection result.

[0005] The first aspect of the present invention provides a thin-wall layer defect detection method for babbitt alloy coating cladding, the thin-wall layer defect detection method for babbitt alloy coating cladding comprising: collecting multi-source initial detection data of a thin-wall cladding layer to be detected in layers, and performing multi-domain transformation and filtering enhancement on the multi-source initial detection data to obtain pre-processed multi-source initial coating detection data; extracting and enhancing the cladding layer characteristics of the pre-processed multi-source initial coating detection data to obtain multi-physical field characteristics of the coating, and performing interface-structure-defect correlation mapping on the multi-physical field characteristics of the coating to obtain a unified feature representation of the internal organizational structure of the coating corresponding to the thin-wall cladding layer to be detected; performing multi-domain transformation and filtering enhancement on the unified feature representation A progressive fusion of the coating depth structure is performed to obtain a deep fusion feature, and the coating defect morphology and structural features are extracted from the deep fusion feature to obtain a preliminary defect semantic feature; based on a plurality of preset defect hazard assessment parameters, a multi-level weight allocation is performed on the preliminary defect semantic feature to obtain a defect feature attention weight, and based on the defect feature attention weight, a feature weighted calculation of a plurality of preset coating safety detection parameters is performed on the preliminary defect semantic feature to generate an initial detection result; the bearing strength and shear strength of the initial detection result are calculated to obtain a coating stress distribution state, and fatigue life prediction is performed on the coating stress distribution state to obtain a final detection result.

[0006] Optionally, in a first implementation method of the first aspect of the present invention, the multi-source initial detection data includes ultrasonic reflection signals, infrared thermal imaging data, eddy current response signals and coating penetration attenuation signals, and the layered acquisition of multi-source initial detection data of the thin-walled weld layer to be detected includes: layered acquisition of the longitudinal wave signal emitted by a preset ultrasonic detection unit to the thin-walled weld layer to be detected, to obtain ultrasonic reflection signals at multiple depths, and to obtain infrared thermal imaging data at multiple depths acquired by a preset infrared camera detection unit, and to obtain multi-frequency eddy current response signals of multi-frequency scanning excitation of a preset eddy current probe detection unit, and to obtain multi-angle coating penetration attenuation signals acquired by a preset X-ray detection unit.

[0007] Optionally, in a second implementation of the first aspect of the present invention, the multi-source initial detection data is subjected to multiple domain transformations and filtering enhancements to obtain preprocessed multi-source initial coating detection data, including: performing wavelet transform on the ultrasonic reflection signal to obtain time-frequency domain decomposition data, and performing threshold filtering on the time-frequency domain data to obtain a high signal-to-noise ratio ultrasonic signal; performing Fourier transform processing on the infrared thermal imaging data to obtain frequency domain thermal features, and performing bandpass filtering on the frequency domain thermal features to obtain high-resolution temperature field data; performing Hilbert transform processing on the impedance data corresponding to the eddy current response signal to obtain instantaneous phase features, and performing adaptive filtering on the instantaneous phase features to obtain high-precision impedance distribution data; performing wavelet packet transform on the coating penetration attenuation signal to obtain multi-scale projection features, and performing Wiener filtering on the multi-scale projection features to obtain high-contrast density distribution; performing spatiotemporal registration and depth correction on the high signal-to-noise ratio ultrasonic signal, the high-resolution temperature field data, the high-precision impedance distribution data, and the high-contrast density distribution to generate preprocessed multi-source initial coating detection data.

[0008] Optionally, in a third implementation of the first aspect of the present invention, the preprocessed multi-source initial coating detection data is subjected to the extraction and enhancement of the cladding layering characteristics to obtain the multi-physical field characteristics of the coating, including: extracting the acoustic wave reflection peak and attenuation characteristics of the corresponding high signal-to-noise ratio ultrasonic signal in the preprocessed multi-source initial coating detection data to obtain the acoustic attenuation characteristics, and calculating the coating acoustic impedance and evaluating the bonding state of the acoustic attenuation characteristics to obtain the interface bonding characteristics; performing time series sampling and logarithmic slope calculation of the transient temperature response of the corresponding high-resolution temperature field data in the preprocessed multi-source initial coating detection data to obtain the thermal conductivity distribution, and performing temperature area comparison and temperature anomaly area positioning on the thermal conductivity distribution to obtain the heat flow anomaly characteristics. ; Perform amplitude and phase separation and conductivity calculation on the high-precision impedance distribution data corresponding to the preprocessed multi-source initial coating detection data to obtain conductivity distribution, and calculate the depth profile of the conductivity distribution and extract material tissue characteristics to obtain material tissue characteristics; Perform contrast calculation and density anomaly detection on the high-contrast density distribution corresponding to the preprocessed multi-source initial coating detection data to obtain density anomaly characteristics, and perform interlayer correlation and voxel reconstruction on the density anomaly characteristics to obtain coating internal structure characteristics; Based on the preset physical field weight coefficient, perform weighted superposition of the interface bonding characteristics, the heat flow anomaly characteristics, the material tissue characteristics, and the coating internal structure characteristics in the corresponding time and space coordinate areas to obtain the coating multi-physical field characteristics.

[0009] Optionally, in a fourth implementation of the first aspect of the present invention, the multi-physical field characteristics of the coating are subjected to interface-structure-defect correlation mapping to obtain a unified feature representation of the internal structure of the coating corresponding to the thin-walled cladding layer to be detected, including: extracting the acoustic reflection-thermal resistance-conductivity-density gradient characteristics of the bonding surface position corresponding to the multi-physical field characteristics of the coating, and performing anomaly detection on the acoustic reflection-thermal resistance-conductivity-density gradient characteristics to obtain the interface bonding surface defect distribution; performing acoustic impedance ratio calculation on the interface bonding surface defect distribution to obtain the interface bonding strength value, and performing fracture morphology reconstruction on the interface bonding strength value to obtain the bonding interface morphological characteristics; The interface morphology characteristics are combined to reconstruct the microstructure and calculate the growth direction angle to obtain the interface grain arrangement characteristics, and the defect positions of the interface grain arrangement characteristics are marked to obtain a defect position distribution map; the defect type is identified on the defect position distribution map to obtain interface defect characteristics, and based on a preset position-type-size-depth sequence, the interface defect characteristics are encoded with feature vectors and arranged in a spatial distribution matrix to obtain a multi-dimensional feature matrix; based on the interface bonding strength value, the feature correlation coefficients and feature sorting are calculated for the multi-dimensional feature matrix to obtain a main feature sequence, and the main feature sequence is numerically normalized to obtain a unified feature representation.

[0010] Optionally, in a fifth implementation of the first aspect of the present invention, the unified feature representation is subjected to progressive fusion of the coating depth structure to obtain a deep fusion feature, including: calculating the surface roughness of the unified feature representation to obtain the coating surface micromorphology, and extracting crack features from the coating surface micromorphology to obtain a surface defect distribution; calculating the gap depth and determining the penetration of crack extension on the surface defect distribution to obtain a surface-interface connectivity feature, and calculating the acoustic impedance difference and thermal conductivity on the surface-interface connectivity feature to obtain an interface conduction feature; performing deep stratification and organizational feature reconstruction of the internal layered structure of the coating on the interface conduction feature to obtain an internal organizational morphology, and performing interlayer correlation calculation and failure risk calculation of multiple depth layers on the internal organizational morphology to obtain a stratified feature importance; calculating a weighted coefficient for the stratified feature importance based on a preset coating failure weight coefficient to obtain a feature weight matrix, and performing stratified combination calculation and interlayer coupling calculation on the feature weight matrix to generate a deep fusion feature.

[0011] Optionally, in a sixth implementation manner of the first aspect of the present invention, the deep fusion feature is used to extract the coating defect morphology and structural features to obtain preliminary defect semantic features, including: performing crack edge detection and crack length and width calculation on the deep fusion feature to obtain crack geometric features, and extracting pore contours and calculating the volume fraction of pore shapes on the deep fusion feature to obtain pore density distribution, and extracting inclusion boundaries and calculating the area of ​​inclusion contours on the deep fusion feature to obtain inclusion size spectrum, and calculating interface gradients on the deep fusion feature to obtain interface transition features, and The stress field and the expansion rate of the crack geometric features are calculated to obtain the crack expansion rate, and the spatial distance calculation and the critical value of the pore spacing are determined for the pore density distribution to obtain the pore aggregation area, and the component characteristics of the inclusion size spectrum are extracted and the criticality of the inclusion type is calculated to obtain the inclusion risk level, and the bonding strength of the interface transition characteristics is calculated to obtain the interface bonding state; the crack expansion rate, the pore aggregation area, the inclusion risk level and the interface bonding state are characterized by quantification and feature defect semantic mapping to obtain preliminary defect semantic features.

[0012] Optionally, in a seventh implementation of the first aspect of the present invention, the preliminary defect semantic features are assigned multi-level weights based on the preset multiple defect hazard assessment parameters to obtain defect feature attention weights, including: performing fracture toughness calculation on the corresponding crack semantic feature parameters in the preliminary defect semantic features to obtain a critical stress intensity factor, performing bearing area calculation on the corresponding pore semantic feature parameters in the preliminary defect semantic features to obtain a strength loss coefficient, performing stress field distortion calculation on the corresponding inclusion semantic feature parameters in the preliminary defect semantic features to obtain a local stress concentration coefficient, and performing shear strength calculation on the corresponding interface semantic feature parameters in the preliminary defect semantic features to obtain an interface bearing capacity, A safety margin calculation is performed on the critical stress intensity factor to obtain a crack hazard degree value, a stress concentration calculation is performed on the strength loss coefficient to obtain a pore influence degree value, a fatigue life reduction calculation is performed on the stress concentration coefficient to obtain an inclusion hazard degree value, and a strength retention rate calculation is performed on the interface bearing capacity to obtain an interface influence degree value; the defect critical damage values ​​corresponding to the crack hazard degree value, the pore influence degree value, the inclusion hazard degree value and the interface influence degree value are respectively calculated, and based on a preset defect critical failure threshold, a threshold ratio calculation is performed on each of the defect critical damage values ​​to obtain a defect feature attention weight.

[0013] Optionally, in an eighth implementation of the first aspect of the present invention, the feature weighted calculation of the preset multiple coating safety detection parameters of the preliminary defect semantic feature is performed based on the defect feature attention weight to generate an initial detection result, including: based on the defect feature attention weight, critical stress calculation is performed on the crack semantic feature parameter to obtain crack tip stress intensity, and effective cross-section calculation is performed on the pore semantic feature parameter to obtain the load-bearing area loss, and stress concentration calculation is performed on the inclusion semantic feature parameter to obtain the local stress peak value, and shear force calculation is performed on the interface semantic feature parameter to obtain the interface bearing value, and fatigue life calculation is performed on the crack tip stress intensity, A crack growth rate prediction value is obtained, and a strength attenuation calculation is performed on the bearing area loss to obtain a pore stress redistribution value, and a damage evolution calculation is performed on the local stress peak to obtain an inclusion stress concentration increment, and a strength attenuation calculation is performed on the interface bearing value to obtain an interface bonding strength change; based on the preset service parameters of the babbitt alloy coating, the crack growth rate prediction value, the pore stress redistribution value, the inclusion stress concentration increment and the interface bonding strength change are subjected to strength reduction and superposition to obtain an actual bearing capacity value of the coating, and based on the preset coating design bearing value, a ratio calculation is performed on the actual bearing capacity value of the coating to obtain a preliminary detection result of the corresponding coating service safety status.

[0014] The second aspect of the present invention provides a thin-wall layer defect detection device for babbitt alloy coating and cladding, and the thin-wall layer defect detection device for babbitt alloy coating and cladding comprises: a preprocessing module, which is used to collect multi-source initial detection data of the thin-wall cladding layer to be detected in layers, and perform multi-domain transformation and filtering enhancement on the multi-source initial detection data to obtain preprocessed multi-source initial coating detection data; an association mapping module, which is used to extract and enhance the cladding layer characteristics of the preprocessed multi-source initial coating detection data to obtain the multi-physical field characteristics of the coating, and perform interface-organization-defect association mapping on the multi-physical field characteristics of the coating to obtain a unified feature representation of the internal organizational structure of the coating corresponding to the thin-wall cladding layer to be detected; a deep fusion module, which is used to extract and enhance the unified feature representation of the internal organizational structure of the coating corresponding to the thin-wall cladding layer to be detected; and a deep fusion module. The feature representation is used to perform progressive fusion of the coating depth structure to obtain the deep fusion feature, and the coating defect morphology and structural features are extracted from the deep fusion feature to obtain the preliminary defect semantic feature; the initial detection module is used to perform multi-level weight allocation on the preliminary defect semantic feature based on a plurality of preset defect hazard assessment parameters to obtain the defect feature attention weight, and based on the defect feature attention weight, perform feature weighted calculation of a plurality of preset coating safety detection parameters on the preliminary defect semantic feature to generate the initial detection result; the final detection module is used to calculate the bearing strength and shear strength of the initial detection result to obtain the coating stress distribution state, and perform fatigue life prediction on the coating stress distribution state to obtain the final detection result.

[0015] The above-mentioned thin-wall layer defect detection method and device of Babbitt alloy coating cladding. In an embodiment of the present invention, multi-source initial detection data of a thin-walled cladding layer to be detected is collected in layers, and multiple domain transformations and filtering enhancements are performed on the multi-source initial detection data to obtain pre-processed multi-source initial coating detection data; the cladding layering features of the pre-processed multi-source initial coating detection data are extracted and enhanced to obtain multi-physical field features of the coating, and the multi-physical field features of the coating are associated with interface-structure-defect mapping to obtain a unified feature representation of the internal organizational structure of the thin-walled cladding layer to be detected corresponding to the coating; the unified feature representation is progressively fused with the coating depth structure to obtain a deep fusion feature, and the coating defect morphology and structural features are extracted from the deep fusion feature to obtain a preliminary defect semantic feature; based on a plurality of preset defect hazard assessment parameters, a multi-level weight allocation is performed on the preliminary defect semantic feature to obtain a defect feature attention weight, and based on the defect feature attention weight, a feature weighted calculation of a plurality of preset coating safety detection parameters is performed on the preliminary defect semantic feature to generate an initial detection result; the bearing strength and shear strength are calculated on the initial detection result to obtain a stress distribution state of the coating, and fatigue life prediction is performed on the stress distribution state of the coating to obtain a final detection result. Compared with the prior art, this application acquires basic physical field characteristics through multi-source detection data collection and filtering enhancement of Babbitt alloy coating, and performs interface-organization-defect mapping on these characteristics to obtain a unified representation of the internal organizational structure of the coating; then obtains defect semantic characteristics through deep structure progressive fusion and defect morphology analysis; then performs weight distribution and stress calculation based on defect hazard assessment parameters, and outputs the fatigue life prediction results of the coating. Through hierarchical data fusion and feature extraction, accurate identification of Babbitt alloy coating defect detection is achieved, especially in terms of interface bonding state, organizational uniformity and defect evolution, the multi-physical field characteristics and defect morphology of the coating are fully considered, and the detection accuracy is effectively improved; and the multi-level feature mapping and hazard assessment strategy are adopted to realize the correlation analysis between physical field characteristics and enhance the reliability of defect detection; in addition, through coating stress distribution calculation and life prediction, the defect type and impact degree are accurately assessed, thereby realizing efficient and accurate detection of Babbitt alloy coating defects as a whole.

[0016] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of a first embodiment of a method for detecting defects in a thin-wall layer of babbitt alloy coating cladding according to an embodiment of the present invention; Figure 2 It is a schematic diagram of an embodiment of a device for detecting defects in a thin-wall layer of babbitt alloy coating and cladding in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device end including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or device ends.

[0021] To facilitate understanding of this embodiment, the specific process of the embodiment of the present invention is described below. Figure 1 The first embodiment of the method for detecting defects in a thin-wall layer of babbitt alloy coating cladding in the embodiment of the present invention comprises: 101. Collect multi-source initial detection data of the thin-walled cladding layer to be detected in layers, and perform multi-domain transformation and filtering enhancement on the multi-source initial detection data to obtain pre-processed multi-source initial coating detection data; The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0022] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0023] In this embodiment, the multi-source initial detection data includes ultrasonic reflection signals, infrared thermal imaging data, eddy current response signals and coating penetration attenuation signals; the longitudinal wave signals emitted by the preset ultrasonic detection unit to the thin-walled cladding layer to be detected are collected in layers to obtain ultrasonic reflection signals at multiple depths, and infrared thermal imaging data at multiple depths collected by the preset infrared camera detection unit are obtained, and multi-frequency eddy current response signals of multi-frequency scanning excitation of the preset eddy current probe detection unit are obtained, and multi-angle coating penetration attenuation signals collected by the preset X-ray detection unit are obtained; the ultrasonic reflection signal is subjected to wavelet transformation to obtain time-frequency domain decomposition data, and the time-frequency domain data is subjected to threshold filtering to obtain a high signal-to-noise ratio ultrasonic signal; The infrared thermal imaging data is processed by Fourier transform to obtain frequency domain thermal features, and the frequency domain thermal features are bandpass filtered to obtain high-resolution temperature field data; the impedance data corresponding to the eddy current response signal is processed by Hilbert transform to obtain instantaneous phase features, and the instantaneous phase features are adaptively filtered to obtain high-precision impedance distribution data; the coating penetration attenuation signal is transformed by wavelet packet transform to obtain multi-scale projection features, and the multi-scale projection features are Wiener filtered to obtain high-contrast density distribution; high signal-to-noise ratio ultrasonic signals, high-resolution temperature field data, high-precision impedance distribution data and high-contrast density distribution are subjected to time-space alignment and depth correction to generate pre-processed multi-source initial coating detection data. The above-mentioned ultrasonic detection unit refers to a detection unit composed of an ultrasonic transmitting probe and a receiving probe, wherein the transmitting probe generates ultrasonic longitudinal waves of a specific frequency, and the receiving probe collects echo signals reflected from different depths of the coating; the above-mentioned infrared camera detection unit refers to a detection unit composed of an infrared thermal imager and an excitation light source, wherein the excitation light source provides thermal excitation to the coating surface, and the thermal imager collects temperature distribution changes on the coating surface in real time; the above-mentioned eddy current probe detection unit refers to a detection unit composed of an excitation coil and a detection coil, wherein the excitation coil generates an alternating magnetic field of different frequencies, and the detection coil receives the induced current response generated by the coating at different depths; the above-mentioned X-ray detection unit refers to a detection unit composed of an X-ray source and an array detector, wherein the X-ray source emits X-rays, which are received by the detector after attenuation by the coating to form a projection image reflecting the internal structure.

[0024] In practical applications, firstly, a longitudinal wave signal is emitted to the thin-walled cladding layer to be detected through a preset ultrasonic detection unit. Since the coating has different depth levels, the ultrasonic wave will generate reflections at different depths during the propagation process, so as to collect ultrasonic reflection signals of multiple depths, and use a preset infrared camera detection unit to perform thermal imaging scanning on the coating surface. Since heat conduction has a delay and attenuation effect inside the coating, the thermal response at different times is collected to obtain infrared thermal imaging data reflecting the temperature distribution at different depths, and the coating is excited by multi-frequency scanning using a preset eddy current probe detection unit. Since eddy currents of different frequencies have different skin depths, the eddy current response signals reflecting the electromagnetic characteristics of different depths are obtained by changing the excitation frequency, and the coating is penetrated from multiple angles using a preset X-ray detection unit to obtain penetration attenuation signals reflecting the density distribution inside the coating; then, the ultrasonic reflection signal is subjected to wavelet transformation to decompose the signal into the time-frequency domain so as to simultaneously analyze the time domain and frequency domain characteristics of the signal, and then the noise component is removed by threshold filtering to obtain an ultrasonic signal with a high signal-to-noise ratio; and for infrared thermal The imaging data is converted to the frequency domain by Fourier transform, the frequency characteristics of the temperature field are extracted, and the high-frequency noise and low-frequency drift are removed by bandpass filtering to obtain the high-resolution temperature field distribution; and for the eddy current detection signal, its instantaneous phase characteristics are obtained by Hilbert transform, and then the interference components are suppressed by adaptive filtering to obtain high-precision impedance data reflecting the distribution of coating conductivity; and for the X-ray penetration attenuation signal, multi-scale decomposition is performed by wavelet packet transform, and the noise of each scale component is suppressed by Wiener filtering, and finally a density distribution image with high contrast is obtained; then the high signal-to-noise ratio ultrasonic signal, high-resolution temperature field data, high-precision impedance distribution data and high-contrast density distribution are temporally and spatially registered and depth corrected, that is, the data obtained by different detection methods are aligned to the same coordinate system, and according to the physical characteristics of each detection method, such as the speed of sound of ultrasound, the diffusion coefficient of heat, the skin effect of eddy current and the attenuation law of X-rays, the depth information is uniformly converted and corrected, so as to ensure that the data at different depth positions can accurately correspond, and finally generate standardized multi-source initial coating detection data that can be directly used for subsequent analysis.For example, to inspect a babbitt alloy coating (2mm thick), ultrasonic waves are emitted at a frequency of 5MHz, 100 layers of data are collected, and the wavelet transform uses the Morlet wavelet function with fb=1.5 and fc=1 to decompose the signal, and the noise is reduced by the soft threshold λj=σ√(2logN) (σ is the noise standard deviation, N is the number of sampling points); infrared imaging uses a 60Hz sampling rate to obtain 300 frames of temperature field images, and a 6th-order Butterworth filter (D0=0.4) is used for bandpass filtering; eddy current detection scans 20 frequency points in the range of 20-200kHz, and the impedance signal is processed by a 50th-order adaptive Wiener filter; X-rays collect 36 projection images at 5° intervals, and the density distribution is reconstructed through 4 layers of wavelet packet decomposition and Wiener filtering (K=0.01), and finally the least squares method is used to estimate the alignment parameters to achieve unified expression of multi-source data.

[0025] 102. Extract and enhance the cladding layer features of the preprocessed multi-source initial coating detection data to obtain the multi-physical field features of the coating, and perform interface-structure-defect correlation mapping on the multi-physical field features of the coating to obtain a unified feature representation of the internal structure of the coating corresponding to the thin-wall cladding layer to be detected; In this embodiment, the acoustic wave reflection peak value and attenuation feature extraction are performed on the corresponding high signal-to-noise ratio ultrasonic signal in the preprocessed multi-source initial coating detection data to obtain the acoustic attenuation feature, and the coating acoustic impedance calculation and bonding state evaluation are performed on the acoustic attenuation feature to obtain the interface bonding feature; the transient temperature response time sampling and logarithmic slope calculation are performed on the corresponding high-resolution temperature field data in the preprocessed multi-source initial coating detection data to obtain the thermal conductivity distribution, and the temperature area comparison and temperature abnormality area positioning are performed on the thermal conductivity distribution to obtain the heat flow abnormality feature; the preprocessed The corresponding high-precision impedance distribution data in the multi-source initial coating detection data are separated in amplitude and phase and the conductivity is calculated to obtain the conductivity distribution, and the depth profile of the conductivity distribution is calculated and the material organization characteristics are extracted to obtain the material organization characteristics; the corresponding high-contrast density distribution in the preprocessed multi-source initial coating detection data is subjected to contrast calculation and density anomaly detection to obtain the density anomaly characteristics, and the density anomaly characteristics are interlayer correlated and voxel reconstructed to obtain the internal structure characteristics of the coating; based on the preset physical field weight coefficient, the interface bonding characteristics, heat flow anomaly characteristics, material The material organization characteristics and the internal structure characteristics of the coating are weightedly superimposed in the corresponding time-space coordinate area to obtain the multi-physical field characteristics of the coating; the acoustic reflection-thermal resistance-conductivity-density gradient characteristics of the corresponding bonding surface position of the coating multi-physical field characteristics are extracted, and the acoustic reflection-thermal resistance-conductivity-density gradient characteristics are detected abnormally to obtain the interface bonding surface defect distribution; the acoustic impedance ratio of the interface bonding surface defect distribution is calculated to obtain the interface bonding strength value, and the fracture morphology of the interface bonding strength value is reconstructed to obtain the bonding interface morphological characteristics; the microstructure reconstruction and growth direction of the bonding interface morphological characteristics are performed The angle is calculated to obtain the interface grain arrangement characteristics, and the interface grain arrangement characteristics are marked as defect positions to obtain a defect position distribution map; the defect type is identified on the defect position distribution map to obtain the interface defect characteristics, and based on the preset position-type-size-depth sequence, the interface defect characteristics are encoded by feature vectors and arranged in a spatial distribution matrix to obtain a multi-dimensional feature matrix; based on the interface bonding strength value, the multi-dimensional feature matrix is ​​calculated for feature correlation coefficients and feature sorting to obtain a main feature sequence, and the main feature sequence is numerically normalized to obtain a unified feature representation. The pre-processed multi-source initial coating detection data includes pre-processed high signal-to-noise ratio ultrasonic signals, high-resolution temperature field data, high-precision impedance distribution data, and high-contrast density distribution; the above-mentioned physical field weight coefficient refers to the importance weight allocation of the four physical field characteristics in defect detection (such as the weight allocation of 0.4, 0.3, 0.2, and 0.1); the above-mentioned position-type-size-depth sequence refers to the priority sorting when encoding defect characteristics.

[0026] In practical applications, firstly, feature extraction is performed on high signal-to-noise ratio ultrasonic signals, and the reflection peaks of sound waves at various interfaces of the coating are identified and extracted, and the attenuation law of these reflected waves during propagation is analyzed to obtain the attenuation characteristics that characterize the acoustic properties inside the coating. Based on the difference in acoustic impedance when sound waves propagate in different media, the interface acoustic impedance between the coating and the substrate is calculated, and the integrity of the interface bonding state is evaluated in combination with the change in the reflection coefficient, thereby obtaining the characteristic parameters reflecting the interface bonding quality of the coating, namely the interface bonding characteristics; and for high-resolution temperature field data, the transient temperature response after thermal excitation is applied to the coating surface is sampled in time series, and the logarithmic slope of the temperature change with time is calculated to obtain the thermal conductivity distribution that reflects the thermal conductivity of the coating. Then, by comparing the temperature field distribution characteristics of different regions, the temperature abnormality areas are identified and located, thereby obtaining the characteristic parameters that characterize the thermal conduction abnormality inside the coating, namely the heat flow abnormality Features; and for high-precision impedance distribution data, the complex impedance signal measured by the eddy current probe is separated into two components, amplitude and phase, and the conductivity distribution at different positions of the coating is calculated based on the principle of electromagnetic induction, and then the parameters reflecting the material structure characteristics are extracted by analyzing the variation law of conductivity in the depth direction, and the characteristic data characterizing the microstructure characteristics of the coating, namely the material structure characteristics, are obtained; and for high-contrast density distribution, the density contrast of different areas is calculated and the density abnormal area is detected to obtain the characteristics reflecting the uneven density distribution inside the coating, and the three-dimensional voxel structure of the coating is reconstructed by establishing the correlation relationship between the density distributions of adjacent layers, so as to obtain the characteristic parameters characterizing the integrity of the internal structure of the coating, namely the internal structure characteristics of the coating; and then, taking into account the differences in the sensitivity and reliability of different physical field characteristics to defect detection, based on the pre-set physical field weight coefficient, through the multi-physical field feature fusion function: ; Where Φ is the fusion feature, is the weight coefficient, φᵢ is the characteristic of each physical field, is the attenuation coefficient, D is the spatial distance function, and the obtained interface bonding characteristics, heat flow anomaly characteristics, material organization characteristics and coating internal structure characteristics are weightedly superimposed in the corresponding time and space areas to achieve effective fusion of multi-physical field characteristics, thereby obtaining multi-physical field characteristics that can comprehensively reflect the quality status of the coating.

[0027] Secondly, a variety of physical field characteristics are extracted at the interface between the coating and the substrate, including the reflection characteristics of sound waves at the interface, the thermal resistance characteristics during heat transfer, the conductivity change caused by electromagnetic induction, and the density gradient characteristics reflected by X-ray attenuation. By setting the normal variation range of each physical quantity, the extracted characteristics are detected for abnormality, thereby obtaining a characteristic map reflecting the distribution of defects on the interface bonding surface; then, by calculating the acoustic impedance ratio of the media on both sides of the interface, the strength value of the interface bonding is evaluated, and the morphological characteristics of the interface fracture are reconstructed using this intensity value information to obtain the microscopic morphological characteristics of the bonding interface, namely, the bonding interface morphological characteristics; then, for the reconstructed interface morphological characteristics, the microstructure of the interface area is reconstructed and the direction angle of grain growth is calculated to obtain the grain arrangement at the interface. Regular characteristic parameters are obtained, and the specific positions of defects are marked in these grain arrangement characteristics to generate an intuitive defect position distribution map; then, based on the morphological characteristics and distribution characteristics of the defects, type identification is performed to obtain the characteristic description of the interface defects, and these interface defect characteristics are encoded into characteristic vectors in the order of pre-set position, type, size and depth, and arranged into a multi-dimensional characteristic matrix according to their spatial distribution law; then, with the interface bonding strength value as a reference benchmark, the correlation coefficients between the features in the multi-dimensional characteristic matrix are calculated, and the features are ranked in importance based on these correlation coefficients, and the main feature sequences that are most critical to defect characterization are screened out, and then through numerical normalization, the features of different dimensions are converted into a unified numerical range, and finally a standardized unified feature representation is obtained. Not only does it achieve an effective mapping from physical field features to defect features, but it also improves the accuracy of subsequent defect identification and evaluation through feature selection and normalization processing.

[0028] 103. Perform progressive fusion of the coating depth structure on the unified feature representation to obtain the deep fusion feature, and extract the coating defect morphology and structural features from the deep fusion feature to obtain the preliminary defect semantic feature; In this embodiment, the surface roughness of the unified feature representation is calculated to obtain the microscopic morphology of the coating surface, and the crack feature is extracted from the microscopic morphology of the coating surface to obtain the surface defect distribution; the gap depth is calculated and the penetration of the crack extension is determined for the surface defect distribution to obtain the surface-interface connectivity feature, and the acoustic impedance difference and thermal conductivity are calculated for the surface-interface connectivity feature to obtain the interface conduction feature; the interface conduction feature is subjected to the depth stratification and organizational feature reconstruction of the internal layered structure of the coating to obtain the internal organizational morphology, and the internal organizational morphology is subjected to the interlayer correlation calculation and failure risk calculation of multiple depth layers to obtain the stratified feature importance; based on the preset coating failure weight coefficient, the weighted coefficient of the stratified feature importance is calculated to obtain the feature weight matrix, and the feature weight matrix is ​​subjected to the stratified combination calculation and the interlayer coupling calculation to generate the deep fusion feature; the deep fusion feature is subjected to the crack edge detection and the crack length and width calculation , the crack geometry characteristics are obtained, and the pore contour is extracted and the volume fraction of the pore shape is calculated for the deep fusion characteristics to obtain the pore density distribution, and the inclusion boundary is extracted and the area of ​​the inclusion contour is calculated for the deep fusion characteristics to obtain the inclusion size spectrum, and the interface gradient is calculated for the deep fusion characteristics to obtain the interface transition characteristics, and the stress field and expansion rate calculations are performed on the crack geometry characteristics to obtain the crack expansion rate, and the spatial distance calculation and critical value judgment of the pore spacing are performed on the pore density distribution to obtain the pore aggregation area, and the component characteristics are extracted from the inclusion size spectrum and the hazard degree of the inclusion type is calculated to obtain the inclusion risk level, and the bonding strength calculation is performed on the interface transition characteristics to obtain the interface bonding state; the crack growth rate, pore aggregation area, inclusion risk level and interface bonding state are characterized by quantification and feature defect semantic mapping to obtain preliminary defect semantic characteristics. The above-mentioned coating failure weight coefficient refers to the importance weight of the effects of defects in different depth layers on coating performance (for example, depth distribution position: the surface layer has the highest weight, generally 0.4, because it directly bears external loads; the interface layer has the second highest weight, generally 0.3, because it is related to the coating bonding strength; the intermediate transition layer has the second highest weight, generally 0.2, which affects the overall performance; the internal matrix layer has the lowest weight, generally 0.1, which mainly plays a supporting role); the above-mentioned coating depth structure is the coating surface-interface-internal structure; In practical applications, the surface roughness parameters of the coating are first calculated to characterize the microscopic morphological characteristics of the surface, including the calculation of the root mean square value and peak-to-valley distribution of the surface height, and the surface crack information is extracted based on these microscopic morphological characteristics. The position and shape of the cracks are identified through edge detection and morphological analysis, thereby obtaining the spatial distribution characteristics of the surface defects. For the detected surface defects, the depth development of the cracks is analyzed in detail. The connectivity between the surface defects and the interface layer is determined by calculating the depth parameters of the crack gap and evaluating whether it penetrates the interface layer. The acoustic property differences and thermal conductivity changes of these connected areas are analyzed to obtain the characteristic parameters characterizing the interface conduction performance. Based on the interface conduction characteristics, the coating is analyzed. The layered structure inside the layer is reconstructed by the organizational characteristics of each layer to obtain a complete description of the internal organizational morphology, and the mutual influence between the layers is evaluated by calculating the degree of characteristic correlation between layers of different depths, and the failure risk of each layer is calculated in combination with the material failure theory, so as to obtain a quantitative index reflecting the importance of the characteristics of each layer; and then considering the different degrees of influence of different depth layers on the overall performance of the coating during service, based on the pre-set coating failure weight coefficient, the characteristic importance of each layer is weighted and calculated, and a weight matrix reflecting the depth distribution characteristics is constructed. Finally, by layering and combining these weighted features and considering the coupling effect between layers, the effective fusion of the depth direction characteristics is achieved to obtain the deep fusion characteristics. Not only the characteristics of the various hierarchical structures of the coating are considered, but also the overall performance of the coating is comprehensively evaluated by establishing interlayer correlation and failure risk assessment, and the detection sensitivity and accuracy of internal defects of the coating are improved through reasonable weight distribution and feature fusion strategies.

[0029] Secondly, the edge detection algorithm is used to accurately identify the boundary of the crack, and the geometric parameters such as the length and width of the crack are calculated to obtain the geometric feature description of the crack. The contours of the pores in the coating are extracted, the shape characteristics and volume of each pore are calculated, and the number of pores per unit volume is counted to obtain a characteristic map reflecting the pore distribution density, namely the pore density distribution. The boundary contours of the inclusions are extracted, the area and shape parameters of each inclusion are calculated, and a complete inclusion size distribution spectrum, namely the inclusion size spectrum, is established. The coating interface area is subjected to gradient analysis, and the variation characteristics of the material properties at the interface are calculated to obtain the characteristic description of the interface transition zone, namely the interface transition characteristics. Based on the geometric characteristics of the crack, the stress field distribution at the crack tip is calculated by finite element analysis, and the crack is predicted in combination with the fracture mechanics theory. The expansion rate is calculated, and the pore density distribution is spatially analyzed. The distance between adjacent pores is calculated. The degree of pore aggregation is determined based on the set critical spacing, and the high-risk pore aggregation area is determined. In addition, by analyzing the composition characteristics and size distribution of inclusions, the influence of different types of inclusions on coating performance is evaluated, and the hazard level of inclusions is determined. At the same time, the characteristics of the interface transition zone are analyzed, the bonding strength of the interface is calculated, and the reliability state of the interface bonding is evaluated. Then, for these different types of defect characteristics, a standardized feature description system is established, and various defect parameters are converted into a unified semantic feature representation through feature space mapping. This mapping process not only takes into account the physical properties of the defects, but also includes the degree of their influence on the coating performance, thereby forming a complete preliminary defect semantic feature.

[0030] 104. Based on the preset multiple defect hazard assessment parameters, perform multi-level weight allocation on the preliminary defect semantic features to obtain the defect feature attention weight, and based on the defect feature attention weight, perform feature weighted calculation of the preset multiple coating safety detection parameters on the preliminary defect semantic features to generate the initial detection result; In this embodiment, the fracture toughness calculation is performed on the crack semantic feature parameters corresponding to the preliminary defect semantic feature to obtain the critical stress intensity factor, and the bearing area calculation is performed on the pore semantic feature parameters corresponding to the preliminary defect semantic feature to obtain the strength loss coefficient, and the stress field distortion calculation is performed on the inclusion semantic feature parameters corresponding to the preliminary defect semantic feature to obtain the local stress concentration coefficient, and the shear strength calculation is performed on the interface semantic feature parameters corresponding to the preliminary defect semantic feature to obtain the interface bearing capacity, And perform safety margin calculation on critical stress intensity factor to obtain crack hazard degree value, perform stress concentration calculation on strength loss coefficient to obtain pore influence degree value, perform fatigue life reduction calculation on stress concentration coefficient to obtain inclusion hazard degree value, and perform strength retention rate calculation on interface bearing capacity to obtain interface influence degree value; respectively calculate the critical damage values ​​of defects corresponding to crack hazard degree value, pore influence degree value, inclusion hazard degree value and interface influence degree value, and calculate threshold ratio of each defect critical damage value based on preset defect critical failure threshold to obtain defect feature attention weight; based on defect feature attention weight, perform critical stress calculation on crack semantic feature parameters to obtain crack tip stress intensity, perform effective cross-section calculation on pore semantic feature parameters to obtain bearing area loss, and perform calculation on inclusion semantic feature parameters. Stress concentration calculation is performed to obtain the local stress peak value, and shear force calculation is performed on the interface semantic characteristic parameters to obtain the interface bearing value, and fatigue life calculation is performed on the stress intensity at the crack tip to obtain the crack growth rate prediction value, and strength attenuation calculation is performed on the bearing area loss to obtain the pore stress redistribution value, and damage evolution calculation is performed on the local stress peak value to obtain the inclusion stress concentration increment, and strength attenuation calculation is performed on the interface bearing value to obtain the interface bonding strength change; based on the preset babbitt alloy coating service parameters, the crack growth rate prediction value, pore stress redistribution value, inclusion stress concentration increment and interface bonding strength change are subjected to strength reduction and superposition to obtain the actual bearing capacity value of the coating, and based on the preset coating design bearing value, the actual bearing capacity value of the coating is calculated by ratio to obtain the preliminary test results of the corresponding coating service safety status.The above-mentioned various defect hazard assessment parameters refer to the key indicators for evaluating the impact of various defects on coating performance (such as crack defects: critical crack growth length and critical growth rate, pore defects: critical pore density and critical aggregation, inclusion defects: critical inclusion size and critical distribution density and interface defects: critical shear strength and critical bonding rate, etc.); the above-mentioned defect critical failure thresholds refer to the critical values ​​of various defects that cause coating failure (such as cracks: when the crack growth rate is >0.1mm / h, it is judged as failure; pores: when the pore volume fraction is >10%, it is judged as failure; inclusions: when the maximum inclusion size is >100μm, it is judged as failure; interface: when the bonding strength is <200MPa, it is judged as failure); the above-mentioned various coating safety detection parameters refer to the mechanical parameters for evaluating the service safety of the coating (such as: crack parameters: stress intensity factor, fracture toughness; pore parameters : effective bearing area, strength loss coefficient; miscellaneous parameters: stress concentration coefficient, fatigue life reduction coefficient; interface parameters: shear strength, bonding strength retention rate); the above-mentioned service parameters of the babbitt alloy coating refer to the actual working conditions of the coating (such as operating conditions: working temperature (normal temperature 20MPa); load characteristics: static load, cyclic load, impact load; environmental factors: corrosive media, wear conditions; service time: cumulative service time, number of cycles); the above-mentioned coating design bearing value refers to the standard design indicators of the babbitt alloy coating (such as tensile strength: ≥300MPa, shear strength: ≥150MPa, interface bonding strength: ≥200MPa, fatigue strength: ≥150MPa (107 cycles)); the preliminary defect semantic features include crack semantic feature parameters, pore semantic feature parameters, inclusion semantic feature parameters and interface semantic feature parameters.

[0031] In practical applications, the critical stress intensity factor of the crack under the current load conditions is first calculated based on the fracture mechanics theory, and the critical conditions for crack extension are evaluated by analyzing factors such as the shape, size and position of the crack in the crack semantic feature parameters; and for the pore semantic feature parameters, the effective bearing area occupied by the pores is calculated to evaluate the influence of the pores on the bearing capacity of the coating, and the coefficient characterizing the degree of strength loss is obtained; and for the inclusion semantic feature parameters, the degree of stress field distortion around the inclusions is calculated to evaluate the local stress concentration effect, and the coefficient reflecting the severity of stress concentration is obtained; and for the interface semantic feature parameters, the shear strength at the interface is calculated to evaluate the bearing capacity of the interface, and the coefficient characterizing the interface is obtained. The parameters of combined quality are combined; and the degree of damage of cracks is evaluated by calculating the safety margin between the critical stress intensity factor and the actual working stress; the influence of pores is evaluated by analyzing the stress redistribution caused by the strength loss coefficient; the degree of damage of inclusions is evaluated by calculating the influence of stress concentration factor on fatigue life; the influence of interface bonding state is evaluated by analyzing the strength retention rate of interface bearing capacity; and for these four types of defects, their corresponding critical damage values, that is, the critical state parameters that cause coating failure, are calculated respectively, and then based on the pre-set defect critical failure threshold, the ratio of the current state of each defect to the critical state is calculated, so as to obtain the attention weight reflecting the importance of different defects.

[0032] Secondly, based on these attention weights, first calculate the stress intensity at the crack tip, evaluate the stability of the crack under the current load, and obtain the stress intensity at the crack tip; calculate the effective bearing area loss caused by the pores, evaluate the degree of reduction in structural strength, and obtain the bearing area loss; calculate the local stress peak around the inclusions, evaluate the severity of stress concentration, and obtain the local stress peak; calculate the bearing capacity of the interface, evaluate the reliability of the interface bonding, and obtain the interface bearing value; then, based on these mechanical parameters, calculate the crack expansion behavior under cyclic loads through the fatigue mechanics model, predict the crack expansion rate, and analyze the influence of the bearing area loss caused by pores on the coating strength, and calculate the stress redistribution state. The state is evaluated, and the evolution law of the local stress concentration effect caused by inclusions over time is evaluated to obtain the incremental change of stress concentration, and the attenuation of the interface bonding strength during service is analyzed to obtain the change in interface strength; then the actual service conditions of the babbitt alloy coating, including parameters such as working temperature, load type, and environmental factors, are considered to comprehensively evaluate the performance degradation caused by various defects; then the actual bearing capacity value of the coating in the current state is calculated by weighted superposition of the crack growth rate prediction value, pore stress redistribution value, inclusion stress concentration increment, and interface bonding strength change, and then compared with the design bearing value of the coating to obtain the test results reflecting the current service safety state of the coating. This multi-level evaluation method not only considers the independent influence of various defects, but also realizes a comprehensive evaluation of the overall performance of the coating through weight distribution and superposition analysis.

[0033] 105. The bearing strength and shear strength of the initial test results are calculated to obtain the stress distribution state of the coating, and the fatigue life of the coating stress distribution state is predicted to obtain the final test results.

[0034] In this embodiment, the normal force calculation of the load-bearing area loss is performed on the preliminary test results to obtain the effective load-bearing strength distribution, and the tangential force calculation of the interface bonding strength change is performed on the preliminary test results to obtain the effective shear strength distribution, and the measured values ​​corresponding to the effective load-bearing strength and the shear strength distribution are stress superimposed to obtain the actual stress distribution state of the coating; the number of cyclic loading is calculated for the stress distribution state of the coating to obtain the fatigue cycle stress spectrum, and the fatigue cycle stress spectrum is linearly superimposed with the damage amount and the number of cycles to obtain the coating damage accumulation; the power function relationship between the stress intensity factor and the crack length is calculated for the coating damage accumulation to obtain the crack propagation rate, and the critical size calculation of the crack propagation rate is performed to obtain the remaining service life; based on the preset actual working load value of the coating, the confidence interval calculation and the safety factor calculation are performed on the remaining service life to obtain the failure risk probability, and the failure risk probability is graded by threshold calculation to generate the final test result.

[0035] In practical applications, the effective bearing strength distribution of the coating in the vertical direction is evaluated by calculating the influence of the loss of bearing area caused by defects on the normal force. At the same time, the influence of the change of interface bonding strength on the tangential force is analyzed, and the effective strength distribution of the coating in the shear direction is calculated. Then, the measured values ​​of these normal and tangential stresses are reasonably superimposed to obtain the actual stress distribution reflecting the current stress state of the coating. Based on this stress distribution state, a complete fatigue cyclic stress spectrum is generated by statistically analyzing the cyclic load characteristics under actual working conditions, including parameters such as load amplitude, frequency and waveform, and according to the damage mechanics theory, the damage caused to the coating by each cyclic load is calculated. The damage is accumulated by the linear superposition principle to obtain the total damage accumulation of the coating; then, by establishing the power function relationship between the stress intensity factor and the crack extension, the crack extension rate under the current stress state is calculated, and based on this extension rate, the time required for the crack to reach the critical size is evaluated, thereby predicting the remaining service life of the coating; then, considering the volatility of the actual working load of the coating, the confidence interval of the life prediction is calculated by statistical analysis methods, and combined with the requirements of the safety factor, the risk probability of failure of the coating is evaluated, and then according to the preset risk level standard, the failure risk probability is graded and evaluated, and finally a coating test result report is generated. This multi-level evaluation method not only takes into account the static strength, but also includes the cumulative effect of fatigue damage, providing a comprehensive basis for the safety evaluation of the coating.

[0036] In the embodiment of the present invention, the basic physical field characteristics are obtained by multi-source detection data collection and filtering enhancement of the babbitt alloy coating, and the interface-organization-defect mapping is performed on these characteristics to obtain a unified representation of the internal organizational structure of the coating; then, the semantic characteristics of the defects are obtained through deep structure progressive fusion and defect morphology analysis; then, weight distribution and stress calculation are performed based on the defect hazard assessment parameters, and the fatigue life prediction results of the coating are output. Through hierarchical data fusion and feature extraction, accurate identification of babbitt alloy coating defect detection is achieved, especially in terms of interface bonding state, organizational uniformity and defect evolution, the multi-physical field characteristics and defect morphology of the coating are fully considered, and the detection accuracy is effectively improved; and the multi-level feature mapping and hazard assessment strategy are adopted, which not only realizes the correlation analysis between physical field characteristics, but also enhances the reliability of defect detection; in addition, through coating stress distribution calculation and life prediction, the defect type and impact degree are accurately assessed, thereby realizing efficient and accurate detection of babbitt alloy coating defects as a whole.

[0037] The above is a description of the thin-wall layer defect detection method of babbitt alloy coating cladding in the embodiment of the present invention. The following is a description of the thin-wall layer defect detection device of babbitt alloy coating cladding in the embodiment of the present invention. Figure 2 , an embodiment of a thin-wall layer defect detection device for babbitt alloy coating cladding in an embodiment of the present invention comprises: The preprocessing module 201 is used to collect multi-source initial detection data of the thin-walled cladding layer to be detected in layers, and perform multi-domain transformation and filtering enhancement on the multi-source initial detection data to obtain preprocessed multi-source initial coating detection data; The correlation mapping module 202 is used to extract and enhance the cladding layer characteristics of the preprocessed multi-source initial coating detection data to obtain the multi-physical field characteristics of the coating, and perform interface-structure-defect correlation mapping on the multi-physical field characteristics of the coating to obtain a unified feature representation of the internal structure of the coating corresponding to the thin-wall cladding layer to be detected; A deep fusion module 203 is used to perform progressive fusion of the coating depth structure on the unified feature representation to obtain a deep fusion feature, and extract the coating defect morphology and structural features from the deep fusion feature to obtain a preliminary defect semantic feature; The initial detection module 204 is used to perform multi-level weight allocation on the preliminary defect semantic features based on a plurality of preset defect hazard assessment parameters to obtain defect feature attention weights, and perform feature weighted calculation of a plurality of preset coating safety detection parameters on the preliminary defect semantic features based on the defect feature attention weights to generate an initial detection result; The final detection module 205 is used to calculate the bearing strength and shear strength of the initial detection result to obtain the stress distribution state of the coating, and to predict the fatigue life of the coating stress distribution state to obtain the final detection result.

[0038] In the embodiment of the present invention, the basic physical field characteristics are obtained by multi-source detection data collection and filtering enhancement of the babbitt alloy coating, and the interface-organization-defect mapping is performed on these characteristics to obtain a unified representation of the internal organizational structure of the coating; then, the semantic characteristics of the defects are obtained through deep structure progressive fusion and defect morphology analysis; then, weight distribution and stress calculation are performed based on the defect hazard assessment parameters, and the fatigue life prediction results of the coating are output. Through hierarchical data fusion and feature extraction, accurate identification of babbitt alloy coating defect detection is achieved, especially in terms of interface bonding state, organizational uniformity and defect evolution, the multi-physical field characteristics and defect morphology of the coating are fully considered, and the detection accuracy is effectively improved; and the multi-level feature mapping and hazard assessment strategy are adopted, which not only realizes the correlation analysis between physical field characteristics, but also enhances the reliability of defect detection; in addition, through coating stress distribution calculation and life prediction, the defect type and impact degree are accurately assessed, thereby realizing efficient and accurate detection of babbitt alloy coating defects as a whole.

[0039] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0040] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or 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, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0041] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0042] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting defects in thin-walled layers of babbitt alloy coating and cladding, characterized in that: The thin-wall layer defect detection method of babbitt alloy coating cladding comprises: Collecting multi-source initial detection data of the thin-walled cladding layer to be detected in layers, and performing multi-domain transformation and filtering enhancement on the multi-source initial detection data to obtain pre-processed multi-source initial coating detection data; Extracting and enhancing the cladding layer features of the preprocessed multi-source initial coating detection data to obtain the multi-physical field features of the coating, and performing interface-structure-defect correlation mapping on the multi-physical field features of the coating to obtain a unified feature representation of the internal structure of the coating corresponding to the thin-wall cladding layer to be detected; The unified feature representation is progressively fused with the coating depth structure to obtain a deep fusion feature, and the coating defect morphology and structural features are extracted from the deep fusion feature to obtain a preliminary defect semantic feature; Based on a plurality of preset defect hazard assessment parameters, a multi-level weight allocation is performed on the preliminary defect semantic features to obtain defect feature attention weights, and based on the defect feature attention weights, a feature weighted calculation of a plurality of preset coating safety detection parameters is performed on the preliminary defect semantic features to generate an initial detection result; The bearing strength and shear strength of the initial test results are calculated to obtain the stress distribution state of the coating, and the fatigue life of the coating stress distribution state is predicted to obtain the final test results.

2. The method for detecting defects in thin-walled layers of babbitt alloy coating and cladding according to claim 1, characterized in that: The multi-source initial detection data includes ultrasonic reflection signals, infrared thermal imaging data, eddy current response signals and coating penetration attenuation signals. The multi-source initial detection data of the thin-walled cladding layer to be detected is collected in layers, including: The longitudinal wave signal emitted by the preset ultrasonic detection unit to the thin-walled weld layer to be detected is collected in layers to obtain ultrasonic reflection signals at multiple depths, and infrared thermal imaging data at multiple depths collected by the preset infrared camera detection unit is obtained, as well as multi-frequency eddy current response signals of multi-frequency scanning excitation of the preset eddy current probe detection unit, and multi-angle coating penetration attenuation signals collected by the preset X-ray detection unit are obtained.

3. The method for detecting defects in thin-walled layers of babbitt alloy coating and cladding according to claim 2, characterized in that: The performing of multiple domain transformation and filtering enhancement on the multi-source initial detection data to obtain pre-processed multi-source initial coating detection data includes: Performing wavelet transformation on the ultrasonic reflection signal to obtain time-frequency domain decomposition data, and performing threshold filtering on the time-frequency domain data to obtain a high signal-to-noise ratio ultrasonic signal; Performing Fourier transform processing on the infrared thermal imaging data to obtain frequency domain thermal features, and performing bandpass filtering on the frequency domain thermal features to obtain high-resolution temperature field data; Performing Hilbert transform processing on the impedance data corresponding to the eddy current response signal to obtain instantaneous phase characteristics, and performing adaptive filtering on the instantaneous phase characteristics to obtain high-precision impedance distribution data; Performing wavelet packet transformation on the coating penetration attenuation signal to obtain multi-scale projection features, and performing Wiener filtering on the multi-scale projection features to obtain high-contrast density distribution; The high signal-to-noise ratio ultrasonic signal, the high-resolution temperature field data, the high-precision impedance distribution data and the high-contrast density distribution are subjected to time-space registration and depth correction to generate preprocessed multi-source initial coating detection data.

4. The method for detecting defects in thin-walled layers of babbitt alloy coating and cladding according to claim 1, characterized in that: The extraction and enhancement of the cladding layer features of the preprocessed multi-source initial coating detection data to obtain the multi-physical field features of the coating includes: Extracting the acoustic wave reflection peak and attenuation characteristics of the corresponding high signal-to-noise ratio ultrasonic signal in the preprocessed multi-source initial coating detection data to obtain an acoustic attenuation characteristic, and calculating the coating acoustic impedance and evaluating the bonding state of the acoustic attenuation characteristic to obtain an interface bonding characteristic; Performing time series sampling and logarithmic slope calculation of transient temperature response on the high-resolution temperature field data corresponding to the preprocessed multi-source initial coating detection data to obtain thermal conductivity distribution, and performing temperature region comparison and temperature anomaly region positioning on the thermal conductivity distribution to obtain heat flow anomaly characteristics; Separating the amplitude and phase and calculating the conductivity of the high-precision impedance distribution data corresponding to the preprocessed multi-source initial coating detection data to obtain the conductivity distribution, and calculating the depth profile of the conductivity distribution and extracting the material organization characteristics to obtain the material organization characteristics; Performing contrast calculation and density anomaly detection on the corresponding high-contrast density distribution in the preprocessed multi-source initial coating detection data to obtain density anomaly features, and performing interlayer correlation and voxel reconstruction on the density anomaly features to obtain internal structural features of the coating; Based on the preset physical field weight coefficient, the interface bonding characteristics, the heat flow anomaly characteristics, the material organization characteristics, and the internal structure characteristics of the coating are weightedly superimposed in the corresponding time and space coordinate areas to obtain the multi-physical field characteristics of the coating.

5. The method for detecting defects in thin-walled layers of babbitt alloy coating and cladding according to claim 1, characterized in that: The method of performing interface-structure-defect correlation mapping on the multi-physical field characteristics of the coating to obtain a unified feature representation of the internal structure of the coating corresponding to the thin-walled cladding layer to be detected includes: Extracting the acoustic reflection-thermal resistance-conductivity-density gradient characteristics of the multi-physical field characteristics of the coating corresponding to the bonding surface position, and performing anomaly detection on the acoustic reflection-thermal resistance-conductivity-density gradient characteristics to obtain the interface bonding surface defect distribution; Calculating the acoustic impedance ratio of the defect distribution on the interface bonding surface to obtain the interface bonding strength value, and reconstructing the fracture morphology of the interface bonding strength value to obtain the morphological characteristics of the bonding interface; Performing microstructural reconstruction and growth direction angle calculation on the morphological features of the bonding interface to obtain interface grain arrangement features, and marking defect positions on the interface grain arrangement features to obtain a defect position distribution map; Identify the defect type on the defect position distribution map to obtain interface defect features, and perform feature vector encoding and spatial distribution matrix arrangement on the interface defect features based on a preset position-type-size-depth sequence to obtain a multi-dimensional feature matrix; Based on the interface binding strength value, the feature correlation coefficient and feature sorting of the multidimensional feature matrix are calculated to obtain a main feature sequence, and the main feature sequence is numerically normalized to obtain a unified feature representation.

6. The method for detecting defects in thin-walled layers of babbitt alloy coating and cladding according to claim 1, characterized in that: The step of performing progressive fusion of the coating depth structure on the unified feature representation to obtain a deep fusion feature includes: Calculating the surface roughness of the unified feature representation to obtain the coating surface micromorphology, and extracting crack features from the coating surface micromorphology to obtain surface defect distribution; Calculating the gap depth and determining the penetration of the crack extension on the surface defect distribution to obtain the surface-interface connectivity characteristics, and calculating the acoustic impedance difference and thermal conductivity on the surface-interface connectivity characteristics to obtain the interface conduction characteristics; The interface conduction feature is subjected to deep stratification and organizational feature reconstruction of the internal layered structure of the coating to obtain the internal organizational morphology, and the internal organizational morphology is subjected to interlayer correlation calculation and failure risk calculation of multiple depth layers to obtain the importance of the stratified feature; Based on the preset coating failure weight coefficient, the weighted coefficient of the hierarchical feature importance is calculated to obtain a feature weight matrix, and the feature weight matrix is ​​subjected to hierarchical combination calculation and inter-layer coupling calculation to generate a deep fusion feature.

7. The method for detecting defects in thin-walled layers of babbitt alloy coating and cladding according to claim 1, characterized in that: The extraction of coating defect morphology and structural features from the deep fusion features to obtain preliminary defect semantic features includes: The deep fusion feature is used to detect crack edges and calculate crack length and width to obtain crack geometric features, and the deep fusion feature is used to extract pore contours and calculate the volume fraction of pore shapes to obtain pore density distribution, and the deep fusion feature is used to extract inclusion boundaries and calculate the area of ​​inclusion contours to obtain inclusion size spectrum, and the deep fusion feature is used to calculate interface gradients to obtain interface transition features, and the crack geometric feature is used to calculate stress field and expansion rate to obtain crack expansion rate, and the pore density distribution is used to calculate spatial distance and determine the critical value of pore spacing to obtain pore aggregation area, and the inclusion size spectrum is used to extract component features and calculate the hazard of inclusion types to obtain inclusion risk level, and the interface transition feature is used to calculate bonding strength to obtain interface bonding state; The crack growth rate, the pore aggregation area, the inclusion risk level and the interface bonding state are characterized by quantification and feature defect semantic mapping to obtain preliminary defect semantic features.

8. The method for detecting defects in thin-walled layers of babbitt alloy coating and cladding according to claim 1, characterized in that: The method of performing multi-level weight allocation on the preliminary defect semantic features based on the preset multiple defect hazard assessment parameters to obtain defect feature attention weights includes: Performing fracture toughness calculation on the crack semantic feature parameters corresponding to the preliminary defect semantic feature to obtain a critical stress intensity factor, performing bearing area calculation on the pore semantic feature parameters corresponding to the preliminary defect semantic feature to obtain a strength loss coefficient, performing stress field distortion calculation on the inclusion semantic feature parameters corresponding to the preliminary defect semantic feature to obtain a local stress concentration coefficient, performing shear strength calculation on the interface semantic feature parameters corresponding to the preliminary defect semantic feature to obtain an interface bearing capacity, performing safety margin calculation on the critical stress intensity factor to obtain a crack hazard degree value, performing stress concentration calculation on the strength loss coefficient to obtain a pore influence degree value, performing fatigue life reduction calculation on the stress concentration factor to obtain an inclusion hazard degree value, performing strength retention rate calculation on the interface bearing capacity to obtain an interface influence degree value; The defect critical damage values ​​corresponding to the crack hazard degree value, the pore influence degree value, the inclusion hazard degree value and the interface influence degree value are calculated respectively, and based on the preset defect critical failure threshold, the threshold ratio of each defect critical damage value is calculated to obtain the defect feature attention weight.

9. The method for detecting defects in thin-walled layers of babbitt alloy coating and cladding according to claim 8, characterized in that: The method of performing a feature weighted calculation of a plurality of preset coating safety detection parameters on the preliminary defect semantic features based on the defect feature attention weight to generate an initial detection result includes: Based on the defect feature attention weight, the critical stress calculation is performed on the crack semantic feature parameter to obtain the crack tip stress intensity, and the effective cross-section calculation is performed on the pore semantic feature parameter to obtain the load-bearing area loss, and the stress concentration calculation is performed on the inclusion semantic feature parameter to obtain the local stress peak value, and the shear force calculation is performed on the interface semantic feature parameter to obtain the interface load value, and the fatigue life calculation is performed on the crack tip stress intensity to obtain the crack growth rate prediction value, and the strength attenuation calculation is performed on the load-bearing area loss to obtain the pore stress redistribution value, and the damage evolution calculation is performed on the local stress peak to obtain the inclusion stress concentration increment, and the strength attenuation calculation is performed on the interface load value to obtain the interface bonding strength change; Based on the preset service parameters of the babbitt alloy coating, the crack growth rate prediction value, the pore stress redistribution value, the inclusion stress concentration increment and the interface bonding strength change are subjected to strength reduction and superposition to obtain the actual bearing capacity value of the coating. Based on the preset design bearing value of the coating, the actual bearing capacity value of the coating is ratio-calculated to obtain the preliminary detection result of the corresponding coating service safety status.

10. A device for detecting defects in thin-walled layers of babbitt alloy coating and cladding, characterized in that: The thin-wall layer defect detection device of babbitt alloy coating cladding comprises: A preprocessing module is used to collect multi-source initial detection data of the thin-walled cladding layer to be detected in layers, and perform multi-domain transformation and filtering enhancement on the multi-source initial detection data to obtain preprocessed multi-source initial coating detection data; An association mapping module is used to extract and enhance the cladding layer characteristics of the preprocessed multi-source initial coating detection data to obtain the multi-physical field characteristics of the coating, and to perform interface-structure-defect association mapping on the multi-physical field characteristics of the coating to obtain a unified feature representation of the internal organizational structure of the coating corresponding to the thin-wall cladding layer to be detected; A deep fusion module is used to perform progressive fusion of the coating depth structure on the unified feature representation to obtain a deep fusion feature, and to extract the coating defect morphology and structural features from the deep fusion feature to obtain a preliminary defect semantic feature; An initial detection module, for performing multi-level weight assignment on the preliminary defect semantic features based on a plurality of preset defect hazard assessment parameters to obtain defect feature attention weights, and performing feature weighted calculation of a plurality of preset coating safety detection parameters on the preliminary defect semantic features based on the defect feature attention weights to generate an initial detection result; The final detection module is used to calculate the bearing strength and shear strength of the initial detection result to obtain the stress distribution state of the coating, and to predict the fatigue life of the coating stress distribution state to obtain the final detection result.

Images