Engine internal damage identification method and system based on target detection

Abstract

The invention provides an engine internal damage identification method and system based on target detection, and the method comprises the steps: obtaining the distribution of thermal stress and mechanical stress under different working conditions through finite element analysis for typical working conditions of high temperature, high pressure and heat engine coupling load according to the structural characteristics and fault modes of an engine cylinder; processing an ultrasonic guided wave signal through a synthetic aperture focusing algorithm, reconstructing a three-dimensional shape of a crack by combining a three-dimensional geometric model of the cylinder body and guided wave propagation characteristics, and determining a crack propagation rate so as to analyze crack propagation of the cylinder body; the residual service life of the cylinder body under the given working condition is predicted according to the crack size, the crack growth rate and the high-risk working condition parameters in combination with the material mechanical property parameters of the fracture toughness; and according to the crack instability critical length, the cylinder body crack propagation analysis result and the high-risk working condition judgment result, the engine operation working condition is optimized, the occurrence frequency of the high-risk working condition is reduced, the service life of the cylinder body is prolonged, and the crack detection and recognition precision is improved.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method and system for identifying internal damage of an engine based on target detection. Background Art

[0002] The engine block is one of the core components of the engine. As the supporting structure of the combustion chamber and cylinder liner, it needs to withstand harsh working conditions such as high temperature and high pressure. During long-term use, cracks are prone to form inside the cylinder block, seriously affecting engine performance. The generation and expansion of cracks is a gradual process. The initial crack size is small and has little effect on engine performance. However, as the use time increases, the cracks will gradually expand under the action of alternating stress. When the crack size reaches a certain level, it will cause coolant leakage, cylinder pressure drop and other faults. In severe cases, it may even cause the cylinder block to break, causing devastating damage to the engine. In view of the characteristic of the gradual expansion of cylinder block cracks, it is urgent to establish a detection method that can quantitatively evaluate the degree of crack expansion to improve the accuracy of crack detection and identification. Summary of the invention

[0003] The present invention provides an engine internal damage identification method based on target detection, which mainly includes:

[0004] According to the structural characteristics and failure modes of the engine cylinder, the thermal stress and mechanical stress distribution under different working conditions are obtained through finite element analysis for typical working conditions of high temperature, high pressure and thermal-mechanical coupling loads;

[0005] A progressive damage quantification detection method is adopted. A piezoelectric sensor array is arranged on the surface of the engine cylinder block to excite and receive ultrasonic guided wave signals of different modes, and the amplitude of the acoustic emission signal during the crack propagation process is obtained. If the amplitude of the acoustic emission signal exceeds the preset threshold, it is judged that the crack is in an unstable propagation state, and the critical length of crack instability and the crack size are determined.

[0006] The ultrasonic guided wave signal is processed by synthetic aperture focusing algorithm, combined with the three-dimensional geometric model of the cylinder body and the guided wave propagation characteristics, the three-dimensional morphology of the crack is reconstructed, and the crack growth rate is determined to analyze the crack growth of the cylinder body;

[0007] Obtain engine speed, load, and temperature operating parameters, and obtain fatigue load spectra under different operating condition combinations based on the distribution of thermal stress and mechanical stress under different operating conditions, including stress amplitude, stress ratio, and number of cycles. If the crack growth rate exceeds a preset threshold, the operating condition combination is judged to be a high-risk condition that accelerates crack growth;

[0008] Combined with the material mechanical properties parameters of fracture toughness, the remaining service life of the cylinder under given working conditions is predicted based on crack size, crack growth rate and high-risk working condition parameters;

[0009] Combine the analysis results of the cylinder crack growth with the remaining service life of the cylinder to set the early warning thresholds for the crack size, growth rate and high-risk working conditions. When the monitoring parameters exceed the thresholds, an alarm message is generated to prompt the adoption of corresponding inspection and maintenance measures to prevent the crack from entering the rapid growth stage.

[0010] Based on the critical length of crack instability, cylinder crack propagation analysis results and high-risk operating condition judgment results, the engine operating conditions are optimized, the frequency of high-risk conditions is reduced, the service life of the cylinder is extended, and the accuracy of crack detection and identification is improved.

[0011] The present invention provides an engine internal damage identification system based on target detection, which mainly includes:

[0012] Finite element analysis module, used to analyze the thermal and mechanical stress distribution of the cylinder under different working conditions;

[0013] Ultrasonic guided wave detection module, used to monitor the amplitude of the acoustic emission signal during crack propagation and determine the crack state;

[0014] Crack reconstruction module, used to reconstruct the three-dimensional crack morphology, determine the crack growth rate and analyze the growth behavior;

[0015] Fatigue load spectrum module, used to obtain fatigue load spectrum according to working condition parameters and identify high-risk working conditions;

[0016] Life prediction module, used to predict the remaining service life of the cylinder body by combining the crack growth rate and operating parameters;

[0017] The early warning and optimization module is used to generate alarms when monitoring parameters exceed thresholds and optimize operating conditions to extend the life of the cylinder.

[0018] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0019] The present invention discloses a method and system for identifying internal damage of an engine based on target detection. The method obtains the stress distribution of the cylinder under different working conditions through finite element analysis, collects ultrasonic guided wave and acoustic emission signals using a piezoelectric sensor array, reconstructs the three-dimensional morphology of the crack and analyzes its expansion process. Combining the working condition parameters and stress distribution, a fatigue load spectrum is generated and high-risk working conditions are identified. Based on the crack size, expansion rate and material properties, the remaining life of the cylinder is predicted. By setting an early warning threshold, timely prompts for inspection and maintenance are given. Finally, the operating conditions are optimized, the frequency of high-risk working conditions is reduced, and the service life is extended. The present invention realizes comprehensive monitoring and life prediction of engine cylinder cracks, improves the crack detection accuracy and cylinder reliability, and provides a strong guarantee for the safe operation of the engine. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1The present invention is a flow chart of an engine internal damage identification method based on target detection. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] like Figure 1 In this embodiment, a method for identifying internal engine damage based on target detection may specifically include:

[0023] S101. According to the structural characteristics and failure modes of the engine cylinder, the thermal stress and mechanical stress distribution under different working conditions are obtained through finite element analysis for typical working conditions of high temperature, high pressure and thermal-mechanical coupling loads.

[0024] According to the pre-established three-dimensional model of the engine cylinder, meshing is performed, and the model is meshed using a tetrahedral meshing algorithm, and mesh encryption is performed for stress concentration areas. The thermophysical performance parameters and mechanical performance parameters of the cylinder material at different temperatures are obtained from the preset material database, and the parameters are associated with the mesh model to obtain a calculation model containing material property information. For high temperature and high pressure conditions, thermal loads and mechanical loads are applied to the boundary of the calculation model, the convection heat transfer coefficient of the contact surface between the cylinder and the coolant is set, and the finite difference method is run to perform transient heat conduction analysis to calculate the unsteady temperature field distribution of the cylinder. The temperature field distribution is used as the input of the thermal stress analysis, and the mechanical loads borne by the cylinder, including cylinder pressure and assembly preload, are superimposed. The thermal-structural coupling analysis is performed using the linear superposition principle to solve the stress-strain distribution cloud map and the maximum stress value of the cylinder under typical working conditions. The analysis results under different working conditions are compared and evaluated. By comparing the stress distribution under different temperature and pressure conditions, the weak areas and stress concentration positions of the cylinder are determined, providing a basis for the optimization design of the cylinder structure.

[0025] Specifically, based on the pre-established three-dimensional model of the engine cylinder, the model is meshed using the tetrahedral meshing algorithm, and the mesh is encrypted in the stress concentration area to obtain a finite element mesh model with a suitable number of units and node distribution. The thermophysical performance parameters and mechanical performance parameters of the cylinder material at different temperatures are obtained from the material database established by experimental tests and literature data collection, and these parameters are associated with the finite element mesh model to obtain a calculation model containing material property information. For high temperature and high pressure conditions, thermal loads and mechanical loads are applied to the boundaries of the finite element model, and the convective heat transfer coefficient of the contact surface between the cylinder and the coolant is set. The coefficient is determined by table lookup or empirical formula, and the finite difference method is run to perform transient heat conduction analysis to calculate the unsteady temperature field distribution of the cylinder. The temperature field distribution obtained in the previous step is used as the input of the thermal stress analysis, and the mechanical loads borne by the cylinder, including cylinder pressure and assembly preload, are superimposed. The thermal-structural coupling analysis is performed using the linear superposition principle to solve the stress-strain distribution cloud map and maximum stress value of the cylinder under typical working conditions. The analysis results under different working conditions are compared and evaluated. By comparing the stress distribution under different temperature and pressure conditions, the weak areas and stress concentration positions of the cylinder block are determined, providing a basis for the subsequent optimization design of the cylinder block structure. The three-dimensional model of the engine cylinder block is created using computer-aided design software. The model includes the cylinder block shell, water jacket, cylinder liner and bolt holes. The tetrahedral meshing algorithm sets the mesh size to 5mm, reduces the mesh size to 1mm around the inner wall of the cylinder liner and the bolt holes, and finally generates a finite element mesh model containing about 1 million units. The material database obtains the thermal conductivity, specific heat capacity and linear expansion coefficient of the cast iron cylinder block in the temperature range of 20℃ to 600℃ through experimental testing. For example, the thermal conductivity is 50W / (m·K) at 20℃ and drops to 35W / (m·K) at 600℃. Under high temperature and high pressure conditions, the inner surface temperature of the cylinder body is set to 300℃, the outer surface temperature is 80℃, and the cylinder pressure is 10MPa. The convective heat transfer coefficient between the cylinder block and the coolant contact surface is calculated by the Dittus-Boelter formula and is 10000 W / (m 2 ·K). The finite difference method was used for transient heat conduction analysis, the time step was set to 0.1s, and 100 time steps were calculated to obtain the temperature field distribution of each node of the cylinder body that changes with time. In the thermal stress analysis, the temperature field distribution was imported into the finite element software, and a cylinder pressure of 10MPa and a bolt preload of 50kN were applied at the same time. The linear superposition principle was used for thermal-structural coupling analysis, and the vonMises stress distribution cloud map of the cylinder body was calculated. The maximum stress value appeared at the upper end of the cylinder liner, which was 250MPa. By comparing the stress distribution under three different temperatures of 200℃, 300℃, and 400℃ and three different pressures of 8MPa, 10MPa, and 12MPa, it was found that the upper end of the cylinder liner and the area around the bolt hole are stress concentration areas, and the maximum stress increases linearly with the increase of temperature and pressure.

[0026] S102. Using a progressive damage quantification detection method, a piezoelectric sensor array is arranged on the surface of the engine cylinder block to excite and receive ultrasonic guided wave signals of different modes to obtain the amplitude of the acoustic emission signal during the crack propagation process. If the amplitude of the acoustic emission signal exceeds a preset threshold, it is determined that the crack is in an unstable propagation state, and the critical length of crack instability and the crack size are determined.

[0027] According to the geometric features and material properties of the engine cylinder, the Delaunay triangulation algorithm is used to determine the layout position and number of the piezoelectric sensor array, and the sensors are evenly distributed on the surface of the cylinder. The programmable signal generator generates an ultrasonic guided wave signal with a preset frequency range and a preset amplitude, which excites the transmitting sensor in the piezoelectric sensor array, and the receiving sensor captures the ultrasonic guided wave signal after propagation through the cylinder structure. The received signal is amplified by a preset gain and bandpass filtered in a preset frequency range, and digitized at a preset sampling rate using a preset bit analog-to-digital converter to obtain the digitized signal spectrum characteristics. The digitized acoustic emission signal is decomposed and denoised by a preset number of layers and feature extraction using a db4 wavelet transform, and the amplitude, energy and duration of the acoustic emission signal are calculated. The acoustic emission signal is divided into a noise signal and an effective crack signal by a K-means clustering algorithm. For the effective crack signal, it is compared with the crack size and acoustic emission signal feature database established in advance through finite element simulation and experimental data fitting. If the amplitude of the acoustic emission signal exceeds a preset threshold, it is judged that the crack is in an unstable expansion state. The critical length of crack instability and the current crack size are determined by using a cubic spline interpolation algorithm. The acoustic emission signal during crack propagation is monitored in real time, a preset number of data points are collected at preset time intervals, and the data is stored in a time series database for long-term trend analysis and crack propagation rate calculation.

[0028] Specifically, according to the geometric characteristics and material properties of the engine cylinder, the Delaunay triangulation algorithm is used to determine the layout and number of the piezoelectric sensor array, and the sensors are evenly distributed on the cylinder surface. Piezoelectric sensors with a frequency range of 20kHz-200kHz and a sensitivity of not less than -70dB are selected from the pre-established sensor performance database. The programmable signal generator generates an ultrasonic guided wave signal with a frequency range of 20kHz-200kHz and an amplitude of 1-10V to excite the transmitting sensor in the piezoelectric sensor array. The receiving sensor captures the ultrasonic guided wave signal after propagation through the cylinder structure, and the received signal is amplified by 50dB gain and 20kHz-200kHz bandpass filtered. The signal is digitized with a 16-bit analog-to-digital converter at a sampling rate of 1MHz to obtain the spectral characteristics of the digitized signal. The db4 wavelet transform is used to perform 5-layer decomposition and noise reduction and feature extraction on the digitized acoustic emission signal, and the amplitude, energy and duration of the acoustic emission signal are calculated. The acoustic emission signal is divided into a noise signal and an effective crack signal by the K-means clustering algorithm. For the effective crack signal, the crack size and acoustic emission signal feature database established in advance through finite element simulation and experimental data fitting are compared. If the acoustic emission signal amplitude exceeds the preset threshold of 85dB, the crack is judged to be in an unstable expansion state, and the critical length of crack instability and the current crack size are determined using the cubic spline interpolation algorithm. The acoustic emission signal during the crack expansion process is monitored in real time, 100 data points are collected per second, and the data is stored in the time series database for long-term trend analysis and crack growth rate calculation. On the surface of the engine cylinder, 16 piezoelectric sensors are arranged using the Delaunay triangulation algorithm to form a 4×4 matrix array with a sensor spacing of 100mm. Piezoelectric sensors with a frequency range of 20kHz-200kHz and a sensitivity of -65dB are selected from the database. The programmable signal generator generates a 100kHz, 5V sinusoidal ultrasonic guided wave signal, which is excited every 100ms. The signal captured by the receiving sensor is amplified by a 50dB gain amplifier, filtered by a 20kHz-200kHz bandpass filter, and digitized by a 16-bit analog-to-digital converter at a sampling rate of 1MHz to obtain a 16384-point digital signal. The digital signal is subjected to a fast Fourier transform to obtain the spectral characteristics in the range of 0-500kHz. The signal is decomposed into 5 layers using the db4 wavelet to extract the approximate coefficients and detail coefficients, and the signal amplitude, energy, and duration are calculated. The signal amplitude is defined as the maximum instantaneous amplitude, the energy is the sum of the squares of the signal, and the duration is the length of time that the signal amplitude exceeds the threshold. Using the K-means clustering algorithm, the number of clusters k=2 is set to divide the signal into two categories: noise and valid crack signals. The valid crack signal is compared with a pre-established database, which contains the acoustic emission characteristics of cracks of different lengths from 0.1mm to 10mm.If the signal amplitude exceeds 85dB, the crack is judged to be unstable and the crack length is estimated at 0.1mm intervals using the cubic spline interpolation algorithm. During real-time monitoring, 100 data points are collected per second and stored in the time series database. The crack growth rate within 10 minutes is calculated. If the rate exceeds 0.1mm / min, an early warning signal is triggered.

[0029] S103, processing the ultrasonic guided wave signal through a synthetic aperture focusing algorithm, combining the three-dimensional geometric model of the cylinder body and the guided wave propagation characteristics, reconstructing the three-dimensional morphology of the crack, and determining the crack growth rate to analyze the cylinder body crack growth.

[0030] Finite element analysis was performed using COMSOL Multiphysics software to calculate the propagation characteristics of ultrasonic guided waves in the cylinder, and to obtain the phase velocity and group velocity curves of the guided waves; the optimal excitation frequency was determined according to the slope of the group velocity dispersion curve; the ultrasonic guided wave signal was collected by a piezoelectric sensor array, and the collected original signal was subjected to bandpass filtering and Hilbert transform to extract the envelope characteristics of the signal; the signal was inverted and delayed by a time reversal algorithm to obtain a focused sound field distribution diagram; the focused sound field data was processed by a synthetic aperture focusing algorithm, Combining the three-dimensional geometric model of the cylinder body and the waveguide propagation characteristics, the arrival time method is used to perform triangulation to calculate the crack position coordinates; for the reconstructed three-dimensional crack morphology, the Otsu threshold segmentation algorithm is used to extract the crack edge contour, and the crack area change between two adjacent samplings is calculated; the crack growth rate is calculated using the finite difference method in the central difference format, and the cylinder crack growth process is quantitatively analyzed; a bilinear interpolation algorithm is used to generate a spatiotemporal distribution diagram of the crack growth rate, in which the horizontal axis represents time, the vertical axis represents crack length, and the color depth represents the growth rate.

[0031] Specifically, according to the three-dimensional geometric model and material parameters of the cylinder, COMSOL Multiphysics software was used for finite element analysis, the grid size was set to 1 / 10 of the wavelength, the propagation characteristics of ultrasonic guided waves in the cylinder were calculated, and the phase velocity and group velocity curves of the guided waves in the frequency range of 50kHz to 500kHz were obtained. The optimal excitation frequency was determined to be 200kHz through the slope of the group velocity dispersion curve, and the receiving time window was set to 500μs. The ultrasonic guided wave signal was collected by a piezoelectric sensor array, and the collected original signal was bandpass filtered and Hilbert transformed from 50kHz to 350kHz to extract the envelope characteristics of the signal. The signal was inverted and delayed by a time reversal algorithm. The delay time was calculated based on the distance difference between each sensor and the focusing point to obtain the sound field distribution map after focusing. The synthetic aperture focusing algorithm is used to process the focused acoustic field data. Combined with the three-dimensional geometric model of the cylinder body and the propagation characteristics of guided waves, the arrival time method is used to perform triangulation to calculate the crack position coordinates. The Levenberg-Marquardt algorithm is used for iterative least squares optimization. The number of optimization iterations is set to 100 times, and the convergence threshold is 0.001mm to reconstruct the three-dimensional morphology of the crack. For the reconstructed three-dimensional crack morphology, the Otsu threshold segmentation algorithm is used to extract the crack edge contour, and the crack area change between two adjacent samplings is calculated. Combined with the sampling time interval of 10ms, the finite difference method of the central difference format is used to calculate the crack growth rate, and the cylinder crack growth process is quantitatively analyzed. The bilinear interpolation algorithm is used to generate a 100×100 grid of crack growth rate spatiotemporal distribution diagram, with the horizontal axis representing time, the vertical axis representing crack length, and the color depth representing the growth rate. In COMSOL Multiphysics software, import the cylinder CAD model and set the material to gray cast iron, Young's modulus 170 GPa, Poisson's ratio 0.27, and density 7200 kg / m 3. The mesh is divided into tetrahedral units, and the maximum unit size is set to 1mm. The simulation frequency range is 50-500kHz, and the step size is 10kHz. The calculated A0 modal group velocity is 3200m / s at 200kHz, which is selected as the optimal excitation frequency. 16 piezoelectric sensors are used to form a 4×4 array with a spacing of 50mm. The sampling frequency is 2MHz and the sampling time is 500μs. The original signal is processed by a 100-order FIR bandpass filter with a cutoff frequency of 50-350kHz. After the envelope is extracted by Hilbert transform, the signal is time-reversed. The focus is set at the center of the cylinder, and the difference in distance from each sensor to the focus is divided by the speed of sound to obtain the delay time. After delay superposition, a focused sound field with a peak value of 2.5 times that of the original signal is obtained. In the synthetic aperture algorithm, the surface of the cylinder is divided into 1mm×1mm grids. The arrival time method uses the first three sensors that receive the signal for positioning. The initial damping factor λ of the Levenberg-Marquardt algorithm is 0.01, and the maximum iteration is 100 times. The reconstruction accuracy reaches 0.1mm. The Otsu algorithm automatically determines the threshold of 125 and extracts the crack contour. The central difference calculates the crack area increment within 10ms and divides it by the perimeter to get the expansion rate. Bilinear interpolation generates a 100×100 grid of time-space distribution diagrams, with a horizontal axis of 0-10s, a vertical axis of 0-10mm, and a color range of 0-1mm / s.

[0032] By modeling the vibration, impact and fatigue dynamic loads of the cylinder body, calculating the momentum distribution in the crack area, correlating the momentum distribution with the crack growth rate, and analyzing the influence of dynamic load on crack growth; establishing the relationship between momentum and crack growth rate, predicting the development trend of the crack, and determining the stress field distribution at the crack tip and the crack growth driving force in the crack initiation stage, stable crack growth stage and rapid crack growth stage to further analyze the progressive growth of the cylinder body crack.

[0033] A dynamic load model is established according to the geometric structure and material parameters of the cylinder body, and the stress field distribution and displacement field distribution of the crack area are calculated; a multivariate regression algorithm is used to establish the correlation between momentum distribution and crack growth rate, wherein the momentum size, direction and gradient are selected as independent variables, and the crack growth rate is the dependent variable; the stress intensity factor at the crack tip is calculated using the fracture mechanics theory, and the J integral is calculated by the virtual displacement method, wherein the virtual displacement field is defined, the strain energy density and the traction work are calculated; the stress field distribution at the crack tip is solved by the finite difference method, and the crack growth driving force is calculated by the stress intensity factor method; the crack growth rate is calculated in combination with the Paris formula da / dN=C(ΔK)^m, wherein da / dN is the crack growth rate, ΔK is the stress intensity factor range, and C and m are material constants; the crack growth amount is accumulated by numerical integration using the Runge-Kutta method to obtain the quantitative analysis results of the progressive growth of the cylinder body crack.

[0034] Specifically, according to the geometric structure and material parameters of the cylinder body, the vibration, impact and fatigue dynamic load models were established using ANSYS software. The grid size was set to 0.1 mm, and the boundary conditions included fixed support and load application points. The stress field distribution and displacement field distribution in the crack area were calculated. The momentum distribution in the crack area was obtained using the law of conservation of mass and the momentum theorem. The calculation results included the momentum vector of each node. The multivariate regression algorithm was used to establish the correlation between momentum distribution and crack growth rate. The momentum size, direction and gradient were selected as independent variables, and the crack growth rate was selected as the dependent variable. The regression coefficient was solved by the least squares method, and the determination coefficient R was used. 2 Evaluate the goodness of fit of the model, predict the development trend of the crack, and determine the expansion stage of the crack. For different crack expansion stages, the stress intensity factor at the crack tip is calculated using fracture mechanics theory, and the J integral is calculated using the virtual displacement method. The specific steps include defining the virtual displacement field, calculating the strain energy density and traction work, and combining the crack expansion criterion to determine the critical conditions of the crack initiation stage, stable expansion stage, and rapid expansion stage. The finite difference method is used to solve the stress field distribution at the crack tip, and the stress intensity factor method is used to calculate the crack expansion driving force. Combined with the Paris formula da / dN=C(ΔK)^m, where da / dN is the crack expansion rate, ΔK is the stress intensity factor range, C and m are material constants, the crack expansion rate is calculated, and the Runge-Kutta method is used to perform numerical integration to accumulate the crack expansion amount, realizing the quantitative analysis of the progressive expansion of the cylinder crack. The fatigue crack expansion threshold ΔKth of the material is determined by fatigue testing as the applicable lower limit of the Paris formula. The variance analysis method was used to compare the crack propagation behaviors under different dynamic load conditions to determine the main influencing factors of vibration, impact and fatigue loads on crack propagation. The cylinder CAD model was imported into ANSYS, and the material was set to ductile iron, with an elastic modulus of 170GPa and a Poisson's ratio of 0.275. The mesh was divided into tetrahedral units with a unit size of 0.1mm and a total number of approximately 2 million units. The boundary condition settings included a fixed constraint on the bottom of the cylinder, a 10MPa periodic pressure load on the top, and a frequency of 50Hz to simulate engine operation. The vibration load was set to a 0-100Hz sweep frequency with an amplitude of 0.5g; the impact load was a half-sine wave with a peak value of 500g and a duration of 0.5ms. The momentum distribution within 50μm of the crack tip was calculated. The multivariate regression model selected five independent variables, including momentum size, direction angle and gradient, and the stepwise regression method was used to screen the variables. The final model R 2 In the J integral calculation, the virtual displacement field is taken as the displacement field within 1 mm of the crack tip, four integral paths are selected, and the average value is taken. In the Paris formula, C = 2.3 × 10 -11, m = 3.2, obtained by CT test. When solving the stress field by finite difference, the grid size is 0.01mm and the time step is 0.1μs. The Runge-Kutta integration adopts the fourth-order method with an integration step of 0.01mm. The fatigue test was carried out on an MTS fatigue testing machine with a loading ratio of R = 0.1 and a frequency of 10Hz, and ΔKth = 5.8MPa√m was obtained. The results of variance analysis show that at a confidence level of 95%, the impact load has the most significant effect on the crack growth rate, with an F value of 78.3, followed by F = 45.6 fatigue load and F = 23.1 vibration load.

[0035] S104, obtaining engine speed, load, and temperature operating parameters, and obtaining fatigue load spectra under different operating condition combinations, including stress amplitude, stress ratio, and number of cycles, based on the distribution of thermal stress and mechanical stress under different operating conditions. If the crack growth rate exceeds a preset threshold, the operating condition combination is judged to be a high-risk operating condition that accelerates crack growth.

[0036] The engine control unit acquires engine speed, load and temperature operating parameters, and the operating parameters are converted from analog signals to digital signals by a data acquisition card; the thermal stress and mechanical stress distribution of the cylinder under different operating conditions are calculated using the finite element method according to the operating parameters, and the finite element method adopts a thermal-structural coupling analysis method; the stress time history is cycle counted using the rain flow counting method, and the rain flow counting method includes extracting stress peak and valley values, forming stress-time pairs, pairing to eliminate small cycles, and counting residual stress amplitudes; the fatigue damage accumulation value is calculated using Miner linear cumulative damage theory, and the fatigue damage accumulation value is obtained from the fatigue load spectrum under different operating condition combinations; the crack growth rate under the different operating condition combinations is calculated using the Paris formula, and if the crack growth rate exceeds a preset threshold, the fuzzy comprehensive evaluation method is used to perform risk assessment on the operating condition combination; the evaluation indicators of the fuzzy comprehensive evaluation method include stress amplitude, stress ratio and number of cycles, and the weights of the evaluation indicators are 0.5, 0.3 and 0.2 respectively.

[0037] Specifically, the engine speed, load and temperature parameters are collected through the engine control unit, and the analog signal is converted into a digital signal using a data acquisition card. The sampling frequency is 1kHz and the precision is 16 bits. The collected data is denoised using a Butterworth low-pass filter, and the cutoff frequency is set to 100Hz to obtain the engine operating parameters under different working conditions. According to the acquired working condition parameters, the finite element method is used to calculate the thermal stress and mechanical stress distribution of the cylinder under different working conditions. The mesh type uses tetrahedral units with a minimum size of 0.5mm. The thermal-structural coupling analysis method is used to consider the influence of the temperature field on the stress field, and the stress tensor field, including the principal stress and shear stress components, is calculated. Based on the calculated stress field, the rain flow counting method is used to count the stress time history cycle. The specific steps include extracting stress peak and valley values, forming stress-time pairs, pairing to eliminate small cycles, and counting the residual stress amplitude. The stress amplitude, stress ratio and number of cycles are extracted to generate the fatigue load spectrum under different working condition combinations. The Miner linear cumulative damage theory is used to calculate the fatigue damage accumulation value, and the damage critical value is set to 1. The crack growth rate under different working conditions is calculated using the Paris formula. The preset threshold of the crack growth rate is determined by fatigue test and is set to 10 -7 mm / cycle. If the crack growth rate exceeds the preset threshold, the fuzzy comprehensive evaluation method is used to conduct risk assessment on the working condition combination. The evaluation indicators include stress amplitude, stress ratio and number of cycles, with weights of 0.5, 0.3 and 0.2 respectively. The membership function uses Gaussian function to judge that the working condition combination is a high-risk working condition that accelerates crack growth. The orthogonal test method is used for sensitivity analysis to determine the influence of speed, load and temperature on the crack growth rate. The engine control unit collects working condition parameters with a speed range of 800-6000rpm, a load of 0-100%, and a temperature of 20-200℃. The data acquisition card has a sampling frequency of 1kHz and an accuracy of 16 bits. The Butterworth low-pass filter has an order of 4 and a cutoff frequency of 100Hz, and is designed by the Matlab function butter. The finite element analysis uses ANSYS software, with about 2 million tetrahedral mesh units and a minimum size of 0.5mm. In the thermal-structural coupling analysis, the thermal boundary conditions are 300°C for the inner surface of the cylinder, 80°C for the outer surface, and 5000W / (m 2 ·K). In the implementation of the rain flow counting method, the stress peak and valley value extraction adopts the three-point method, and the minimum stress amplitude is set to 10MPa. In Miner's linear cumulative damage theory, the damage amount of a single cycle Di=ni / Ni, ni is the actual number of cycles, and Ni is the fatigue life at this stress level, which is obtained by looking up the SN curve. The material constant C=2.3×10 -11 , m = 3.2, obtained through CT test. Crack growth rate preset threshold 10 -7mm / cycle is determined based on the fact that the crack increment does not exceed 0.1 mm after 106 cycles. In the fuzzy comprehensive evaluation, the membership functions of stress amplitude, stress ratio and number of cycles all use Gaussian functions, and the parameters μ are 200 MPa, 0.5 and 105 times respectively. The orthogonal test adopts L9(3 4 ) orthogonal table, with three levels of speed, load and temperature, and the influence of the three factors on the crack growth rate was obtained through variance analysis. The whole process was automated through Python scripts, and the calculation time was about 30 minutes.

[0038] S105. Combined with the material mechanical property parameters of fracture toughness, the remaining service life of the cylinder under given working conditions is predicted according to the crack size, crack growth rate and high-risk working condition parameters.

[0039] According to the material mechanical properties parameters and fracture toughness data, the critical crack length is calculated using fracture mechanics theory, and finite element analysis is performed using ANSYS software to obtain stress intensity factors under different crack sizes. If the stress intensity factor has been determined, the Paris formula is used to establish the relationship between the crack growth rate and the stress intensity factor range, and the range of stress intensity factors under given conditions is calculated by combining high-risk working condition parameters and the influence of temperature on material properties. According to the range of stress intensity factors, the Monte Carlo method is used to simulate the crack growth process, and the random number generation uses the Mersenne Twister algorithm. Combined with the randomness of material parameters and loads, the material property fluctuations are described by normal distribution, and the load fluctuations are described by Weibull distribution, generating a large number of crack growth paths. If the crack growth path has been generated, the simulation results are smoothed using the kernel density estimation method, using the Gaussian kernel function, and the bandwidth selection uses the Silverman rule. Based on the smoothed crack propagation path and cycle number distribution, the remaining life probability distribution function is fitted using Weibull distribution. The shape parameters and scale parameters are determined by maximum likelihood estimation method. The expected value and 95% confidence interval of the remaining service life of the cylinder block under given working conditions are calculated.

[0040] Specifically, according to the mechanical properties parameters and fracture toughness data of the material, the critical crack length was calculated using fracture mechanics theory. Finite element analysis was performed using ANSYS software. The mesh type selected was an 8-node hexahedral unit with a minimum unit size of 0.1 mm. The stress intensity factor under different crack sizes was calculated, and the relationship curve between the crack size and the stress intensity factor was established. The Paris formula was used to establish the relationship between the crack growth rate and the range of the stress intensity factor. Combined with the high-risk working condition parameters, the influence of temperature on material properties was considered, and the range of the stress intensity factor under a given working condition was calculated to obtain the corresponding crack growth rate. The Monte Carlo method was used to simulate the crack growth process. The Mersenne Twister algorithm was used to generate random numbers. The sample size was set to 10,000. Considering the randomness of material parameters and loads, the material property fluctuations were described by normal distribution, and the load fluctuations were described by Weibull distribution. A large number of crack growth paths were generated, and the distribution of the number of cycles when the crack reached the critical size was statistically analyzed. The simulation results were smoothed using the kernel density estimation method, using the Gaussian kernel function, and the bandwidth selection used the Silverman rule. Based on the crack propagation path and cycle number distribution after smoothing, the Weibull distribution is used to fit the remaining life probability distribution function. The shape parameters and scale parameters are determined by the maximum likelihood estimation method, and the expected value and 95% confidence interval of the remaining service life of the cylinder block under given working conditions are calculated. The prediction results are subjected to sensitivity analysis, and the tornado diagram is used to show the influence of each parameter on the remaining life, including the initial crack size, material fracture toughness, and Paris formula parameters. The cylinder CAD model is imported into ANSYS, and the material properties are set to gray cast iron, elastic modulus 170GPa, Poisson's ratio 0.25, and fracture toughness 30MPa·m^0.5. The mesh is divided into 8-node hexahedral elements SOLID185, the minimum element size is 0.1mm, and the total number of elements is about 1 million. The critical crack length is calculated to be 5mm. The displacement extrapolation method is used to calculate the stress intensity factor, and a 1 / 4 node singular element is set at the crack tip. The Paris formula parameters C=5×10^-12, m=3, considering the temperature effect, the C value increases by 20% for every 100℃ increase. The high-risk working condition is set as a speed of 4000rpm, a load of 80%, and a temperature of 180℃. In the Monte Carlo simulation, the elastic modulus of the material follows a normal distribution with a mean of 170GPa and a standard deviation of 5GPa, and the load follows a Weibull distribution with a shape parameter of 2.5 and a scale parameter of 100. The MersenneTwister algorithm has a period of 2^19937-1, and generates 10,000 sets of random numbers. The kernel density estimation uses a Gaussian kernel function with a bandwidth of h=1.06σn^(-1 / 5), where σ is the standard deviation and n is the number of samples. The Weibull distribution is fitted using the maximum likelihood estimation method, and the shape parameter is 2.8 and the scale parameter is 5×10^6.The expected value of the remaining life is 4.43×10^6 cycles, and the 95% confidence interval is [3.82×10^6,5.05×10^6] cycles. The Tornado diagram analysis shows that the initial crack size has the greatest impact on the remaining life, with a sensitivity of 0.85, followed by the Paris formula parameter m, with a sensitivity of 0.62.

[0041] The crack size is characterized by ultrasonic testing or X-ray testing to obtain the geometric size information of the crack. Combined with the fracture toughness of the material, the relationship between the crack size and the remaining service life of the cylinder block is established to predict the remaining service life of the cylinder block under given working conditions.

[0042] The cylinder body is scanned by phased array ultrasonic detection technology, and the echo signal obtained is processed by time reversal algorithm to obtain three-dimensional morphological information of the crack; based on the three-dimensional morphological information, the cylinder body is imaged by X-ray computed tomography technology, the crack contour is extracted by Otsu threshold segmentation algorithm, and three-dimensional reconstruction is performed in combination with FDK algorithm to obtain the volume and shape characteristics of the crack; the Dempster-Shafer evidence theory is used to fuse the ultrasonic detection result and the X-ray detection result to determine the precise size parameters of the crack; based on fracture mechanics theory, the finite element method is used to calculate the stress intensity factor of the crack at different sizes, and the relationship between the crack propagation rate and the stress intensity factor is established by Paris formula; the support vector regression algorithm is used to construct the correlation model between the crack size, working condition parameters and the remaining life of the cylinder body, and if the prediction accuracy of the correlation model meets the preset threshold, the remaining life prediction result of the cylinder body under the given working condition is output.

[0043] Specifically, the cylinder body was scanned by phased array ultrasonic detection technology, the probe frequency was set to 5MHz, the number of array elements was 64, and the echo signal was processed by time reversal algorithm, including three steps of time domain inversion, spatial focusing and iterative optimization, to reconstruct the three-dimensional morphology of the crack and obtain the length, depth and width of the crack. The cylinder body was imaged by X-ray computed tomography technology, the spatial resolution was set to 50μm, the tube voltage was 120kV, and the tube current was 200mA. The crack contour was extracted by Otsu threshold segmentation algorithm, and three-dimensional reconstruction was performed in combination with FDK algorithm to construct a three-dimensional model of the crack to obtain the volume and shape characteristics of the crack. The Dempster-Shafer evidence theory was used to fuse the ultrasonic and X-ray detection results to improve the accuracy of crack size measurement. Based on the fracture mechanics theory and combined with the fracture toughness parameters of the material, the finite element method was used to calculate the stress intensity factor under cracks of different sizes. The mesh type selected was a 20-node hexahedral unit with a minimum unit size of 0.1mm. The relationship between the crack growth rate and the stress intensity factor was established by the Paris formula. Monte Carlo simulation was used to consider the randomness of material parameters and loads, and 10,000 crack propagation paths were generated. The support vector regression algorithm was used to construct the correlation model between crack size, operating parameters and the remaining life of the cylinder block. The RBF kernel function was selected, and the kernel parameters and penalty factors were optimized by the grid search method to achieve the prediction of the remaining life of the cylinder block under given working conditions. The Olympus5L64-A12 phased array probe was used for ultrasonic testing, with a frequency of 5MHz, 64 array elements, and a scanning step of 0.5mm. In the time reversal algorithm, the time domain reversal used a 30μs time window, the spatial focusing used the MUSIC algorithm, and the iterative optimization was set 10 times. The X-ray CT scan used a NikonXTH225 industrial CT with a tube voltage of 120kV, a tube current of 200mA, a spatial resolution of 50μm, and 360° rotation to collect 900 projection images. The Otsu threshold method was used to binarize the 8-bit grayscale image, and the threshold was set to 128. FDK reconstruction algorithm uses Hamming filter, voxel size 50×50×50μm 3 In the Dempster-Shafer theory, the basic probability distribution function is determined based on the detection accuracy, which is 0.8 for ultrasound and 0.9 for X-ray. The combination rule adopts the Dempster combination rule. The finite element analysis uses ANSYS software, 20-node hexahedral unit SOLID186, the minimum unit size is 0.1mm, and the total number of units is about 1 million. Paris formula parameter C = 5×10 -12 , m = 3. In the Monte Carlo simulation, the elastic modulus of the material follows the normal distribution N (170 GPa, 5 GPa), and the load follows the Weibull distribution (λ = 100, k = 2). The support vector regression uses the RBF kernel function, and the kernel parameter γ and the penalty factor C are optimized by grid search, γ∈[2-15 ,2 3 ],C∈[2 -5 ,2 15 ], with a step size of 2 and 5-fold cross validation. The final model R 2 The average absolute error is less than 5%.

[0044] S106. Combine the analysis results of the cylinder crack growth and the remaining service life of the cylinder to set the early warning thresholds for the crack size, growth rate and high-risk working conditions. When the monitoring parameters exceed the threshold, an alarm message is generated to prompt the adoption of corresponding inspection and maintenance measures to prevent the crack from entering the rapid growth stage.

[0045] A support vector machine algorithm is used to establish a relationship model between crack size, expansion rate, operating parameters and remaining life. The model uses RBF kernel function and optimizes parameters through grid search method. According to the relationship model, the Monte Carlo simulation method is used to generate sample data, the crack expansion path and remaining life distribution under different conditions are calculated, and the warning thresholds of crack size, expansion rate and high-risk working conditions are determined. For the warning threshold, a fuzzy comprehensive evaluation method is used to construct a multi-parameter warning indicator system, and the triangular membership function of crack size, expansion rate and operating parameters is set. The weight of each indicator is determined by hierarchical analysis method, and the comprehensive warning index is calculated. Past historical data is obtained, and the sliding time window method is used to dynamically update the threshold according to the historical data to improve the adaptability of the early warning system. Based on the InfluxDB time series database, real-time monitoring data is stored, and the Apache Flink stream processing framework is used to perform real-time data analysis. When the monitoring parameter exceeds the threshold or the comprehensive early warning index exceeds the preset value, the alarm information generation module is triggered to push the early warning information to the relevant system.

[0046] Specifically, based on the analysis results of cylinder crack propagation and the remaining service life prediction data, the support vector machine algorithm was used to establish the relationship model between crack size, propagation rate, working condition parameters and remaining service life. The RBF kernel function was used to optimize the parameters through the grid search method, and the optimal parameter combination was determined by 5-fold cross validation. The Monte Carlo simulation method was used to generate 10,000 sample data. Considering that the material elastic modulus obeys the normal distribution with a mean of 170GPa and a standard deviation of 5GPa, and the load amplitude obeys the normal distribution with a mean of 100MPa and a standard deviation of 10MPa, the crack propagation path and remaining service life distribution under different conditions were calculated, and the crack size, propagation rate and warning thresholds of high-risk working conditions were determined. The crack size threshold was set to 5mm, the propagation rate threshold was set to 10^-6mm / cycle, and the high-risk working condition threshold was set to 4000rpm, 80% load, and 200℃ temperature. A multi-parameter early warning indicator system was constructed using the fuzzy comprehensive evaluation method. The triangular membership functions of crack size, expansion rate and operating parameters were set. The weights of each indicator were determined by the 9-point scaling method of the hierarchical analysis method, and the comprehensive early warning index was calculated. The sliding time window method was used to dynamically update the threshold based on the historical data of the past 24 hours to improve the adaptability of the early warning system. The real-time monitoring data was stored based on the InfluxDB time series database, and the Apache Flink stream processing framework was used for real-time data analysis. When the monitoring parameters exceeded the threshold or the comprehensive early warning index exceeded 0.8, the alarm information generation module was triggered and the early warning information was pushed to the relevant system. The ROC curve was used to analyze the accuracy of the early warning, and the false alarm rate and missed alarm rate were calculated. When the AUC value was greater than 0.9, it was determined that the performance of the early warning system met the requirements. In the construction of the support vector machine model, the RBF kernel function parameter γ and the penalty factor C were optimized by grid search, γ∈[2 -15 ,2 3 ],C∈[2 -5 ,2 15 ], the step size is a power of 2, and the best parameter combination is selected by 5-fold cross validation. In the Monte Carlo simulation, the MersenneTwister algorithm is used to generate random numbers, the material elastic modulus N (170GPa, 5GPa), the load amplitude N (100MPa, 10MPa), and the crack propagation path is obtained by 10,000 simulations. In the fuzzy comprehensive evaluation, the crack size membership function is set to a triangular function of (0, 3, 5) mm, and the expansion rate is (10 -7 ,5×10 -7 ,10 -6)mm / cycle, the working condition parameters are normalized and the triangular function of (0.6, 0.8, 1) is used. In the hierarchical analysis, the 1-9 scaling method is used to construct the judgment matrix, and the eigenvector is calculated to obtain the weight. The sliding time window length is set to 24 hours, the threshold is updated once an hour, the exponential weighted moving average method is used, and the smoothing coefficient α=0.2. The data retention policy is set to 30 days in InfluxDB, and continuous queries are used to aggregate data every 5 minutes. The parallelism of ApacheFlink tasks is set to 4, event time processing is used, and the watermark delay is 10 seconds. In the ROC curve analysis, the threshold is changed with a step size of 0.1, the true positive rate and false positive rate are calculated, and the AUC value is calculated using the trapezoidal rule. When AUC>0.9, it is considered that the performance of the early warning system meets the requirements.

[0047] S107. Based on the critical length of crack instability, the analysis results of cylinder crack propagation, and the judgment results of high-risk working conditions, optimize the engine operating conditions, reduce the frequency of high-risk working conditions, extend the service life of the cylinder, and improve the accuracy of crack detection and identification.

[0048] According to the operating parameters and crack growth rate data, a relationship model is established by using a support vector regression algorithm. The relationship model uses an RBF kernel function and optimizes parameters through a grid search method. The output results of the relationship model are obtained, and a genetic algorithm is used to perform multi-objective optimization on the engine operating conditions. The multi-objective optimization takes reducing the frequency of high-risk operating conditions and the crack growth rate as the optimization goal. The multi-objective optimization results are analyzed, and the variance analysis method is used to perform a sensitivity analysis of the operating parameters. The variance analysis method calculates the F value of each parameter to identify the operating parameters that have the most significant impact on crack growth. For the optimized operating parameters, the Monte Carlo method is used to simulate the crack growth process of the cylinder under the new operating conditions. The Monte Carlo method generates random samples and considers the randomness of material properties and loads. Based on the simulation results of the Monte Carlo method, a crack detection and identification model is constructed using a convolutional neural network. The convolutional neural network takes the optimized operating parameters and the crack growth characteristics under the new operating conditions as input, and fine-tunes the pre-trained model through a transfer learning method.

[0049] Specifically, according to the analysis results of the critical length of crack instability and the crack propagation of the cylinder body, the support vector regression algorithm was used to establish the relationship model between the operating parameters and the crack propagation rate. The RBF kernel function was used to optimize the parameters γ and C through the grid search method. The optimal parameter combination was determined by 5-fold cross validation, and the weight of the influence of the operating parameters on the crack propagation was obtained. The genetic algorithm was used to perform multi-objective optimization on the engine operating conditions. The population size was set to 100, the crossover rate was 0.8, and the mutation rate was 0.1. The optimization goal was to reduce the frequency of high-risk operating conditions and the crack propagation rate. The constraints of the speed range of 800-6000rpm, the load of 0-100%, and the temperature of 20-200℃ were set to obtain the optimized operating parameter combination. The variance analysis method was used to perform the sensitivity analysis of the operating parameters, calculate the F value of each parameter, and identify the operating parameters that have the most significant impact on crack propagation. Based on the optimized operating parameters, the Monte Carlo method is used to simulate the crack propagation process of the cylinder under the new operating conditions, and 10,000 random samples are generated. Considering the randomness of material properties and loads, the remaining service life of the cylinder is calculated. By comparing the life difference before and after optimization, the effect of the operating condition optimization is evaluated. A crack detection and recognition model is constructed using a convolutional neural network. The optimized operating parameters and the crack propagation characteristics under the new operating conditions are used as input. The pre-trained model is fine-tuned through the transfer learning method, and the existing detection data and simulation data are integrated to improve the detection and recognition accuracy of the model under the new working conditions. The optimization effect is evaluated using real-time monitoring data, and a feedback mechanism is established. When the actual crack propagation rate deviates from the predicted value by more than 10%, the operating parameter re-optimization process is triggered. In the support vector regression model, the RBF kernel function parameter γ and the penalty factor C are optimized by grid search, γ∈[2 -15 ,2 3 ],C∈[2 -5 ,2 15], the step size is a power of 2, and the best parameter combination is selected by 5-fold cross validation. The genetic algorithm uses roulette wheel selection, single-point crossover and uniform mutation, and iterates for 500 generations. In the variance analysis, the F value of the working condition parameters is calculated to obtain speed F = 78.3, load F = 45.6, and temperature F = 23.1. The Monte Carlo simulation uses the MersenneTwister algorithm to generate random numbers, material elastic modulus N (170GPa, 5GPa), and load amplitude N (100MPa, 10MPa). The convolutional neural network uses the ResNet50 architecture, transfer learning retains the weights of the first 48 layers, fine-tunes the last two layers, the learning rate is set to 0.001, the batch size is 32, and the training is 100 rounds. The pre-trained model is trained on the ImageNet dataset, and the fine-tuning dataset contains 1000 crack images and 500 non-crack images. The real-time monitoring data is collected once an hour and smoothed using the exponentially weighted moving average method with a smoothing coefficient α = 0.2. When the average crack growth rate for 24 consecutive hours deviates from the predicted value by more than 10%, the operating parameter re-optimization process is triggered, and the optimization cycle is set to 7 days.

[0050] The above is only a preferred embodiment of the present invention. It should be pointed out that ordinary technicians in this technical field can make several improvements and supplements without departing from the principle of the present invention. These improvements and supplements should also be regarded as the scope of protection of the present invention.

Claims

1. A method for identifying internal engine damage based on target detection, characterized in that: The method comprises: According to the structural characteristics and failure modes of the engine cylinder, the thermal stress and mechanical stress distribution under different working conditions are obtained through finite element analysis for typical working conditions of high temperature, high pressure and thermal-mechanical coupling loads; A progressive damage quantification detection method is adopted. A piezoelectric sensor array is arranged on the surface of the engine cylinder block to excite and receive ultrasonic guided wave signals of different modes, and the amplitude of the acoustic emission signal during the crack propagation process is obtained. If the amplitude of the acoustic emission signal exceeds the preset threshold, it is judged that the crack is in an unstable propagation state, and the critical length of crack instability and the crack size are determined. The ultrasonic guided wave signal is processed by synthetic aperture focusing algorithm, combined with the three-dimensional geometric model of the cylinder body and the guided wave propagation characteristics, the three-dimensional morphology of the crack is reconstructed, and the crack growth rate is determined to analyze the crack growth of the cylinder body; Obtain engine speed, load, and temperature operating parameters, and obtain fatigue load spectra under different operating condition combinations based on the distribution of thermal stress and mechanical stress under different operating conditions, including stress amplitude, stress ratio, and number of cycles. If the crack growth rate exceeds a preset threshold, the operating condition combination is judged to be a high-risk condition that accelerates crack growth; Combined with the material mechanical properties parameters of fracture toughness, the remaining service life of the cylinder under given working conditions is predicted based on crack size, crack growth rate and high-risk working condition parameters; Combine the analysis results of the cylinder crack growth with the remaining service life of the cylinder to set the early warning thresholds for the crack size, growth rate and high-risk working conditions. When the monitoring parameters exceed the thresholds, an alarm message is generated to prompt the adoption of corresponding inspection and maintenance measures to prevent the crack from entering the rapid growth stage. Based on the critical length of crack instability, cylinder crack propagation analysis results and high-risk operating condition judgment results, the engine operating conditions are optimized, the frequency of high-risk conditions is reduced, the service life of the cylinder is extended, and the accuracy of crack detection and identification is improved.

2. The method according to claim 1, characterized in that According to the structural characteristics and failure modes of the engine cylinder, the thermal stress and mechanical stress distribution under different working conditions are obtained through finite element analysis for typical working conditions of high temperature, high pressure and thermal-mechanical coupling loads, including: Mesh the engine block three-dimensional model established in advance, mesh the model using a tetrahedral meshing algorithm, and perform mesh encryption on stress concentration areas; Acquire the thermophysical performance parameters and mechanical performance parameters of the cylinder material at different temperatures from a preset material database, associate the parameters with the grid model, and obtain a calculation model containing material property information; For high temperature and high pressure working conditions, heat load and mechanical load are applied to the boundary of the calculation model, the convection heat transfer coefficient of the contact surface between the cylinder and the coolant is set, and the finite difference method is used to perform transient heat conduction analysis to calculate the unsteady temperature field distribution of the cylinder; The temperature field distribution is used as the input of thermal stress analysis, and the mechanical loads borne by the cylinder body, including cylinder pressure and assembly preload, are superimposed. The thermal-structural coupling analysis is performed using the linear superposition principle to obtain the stress-strain distribution cloud diagram and the maximum stress value of the cylinder body under typical working conditions; The analysis results under different working conditions are compared and evaluated. By comparing the stress distribution under different temperature and pressure conditions, the weak areas and stress concentration locations of the cylinder body are determined, providing a basis for the optimal design of the cylinder body structure.

3. The method according to claim 1, characterized in that The progressive damage quantification detection method is used to arrange a piezoelectric sensor array on the surface of the engine cylinder block, excite and receive ultrasonic guided wave signals of different modes, obtain the amplitude of the acoustic emission signal during the crack propagation process, and if the amplitude of the acoustic emission signal exceeds a preset threshold, it is judged that the crack is in an unstable propagation state, and the critical length of crack instability and the crack size are determined, including: According to the geometric features and material properties of the engine cylinder, the Delaunay triangulation algorithm is used to determine the layout position and quantity of the piezoelectric sensor array, and the sensors are evenly distributed on the surface of the cylinder; An ultrasonic guided wave signal having a frequency range of a preset range and an amplitude of a preset value is generated by a programmable signal generator, a transmitting sensor in the piezoelectric sensor array is excited, and a receiving sensor captures the ultrasonic guided wave signal after propagating through the cylinder structure; The received signal is amplified by a preset gain and band-pass filtered within a preset frequency range, and digitized using an analog-to-digital converter with a preset bit number at a preset sampling rate to obtain the spectral characteristics of the digitized signal; Using db4 wavelet transform to perform preset layer decomposition, noise reduction and feature extraction on the digitized acoustic emission signal, and calculating the amplitude, energy and duration of the acoustic emission signal; The acoustic emission signal is divided into a noise signal and an effective crack signal by using a K-means clustering algorithm; The effective crack signal is compared with a crack size and acoustic emission signal feature database established in advance through finite element simulation and experimental data fitting. If the amplitude of the acoustic emission signal exceeds a preset threshold, it is determined that the crack is in an unstable expansion state; Determine the critical length of crack instability and the current crack size using a cubic spline interpolation algorithm; The acoustic emission signal during the crack propagation process is monitored in real time, a preset number of data points are collected at preset time intervals, and the data are stored in a time series database for long-term trend analysis and crack propagation rate calculation.

4. The method according to claim 1, characterized in that: The method processes the ultrasonic guided wave signal by a synthetic aperture focusing algorithm, reconstructs the three-dimensional morphology of the crack by combining the three-dimensional geometric model of the cylinder body and the guided wave propagation characteristics, and determines the crack growth rate to analyze the crack growth of the cylinder body, including: The finite element analysis is performed using COMSOL Multiphysics software to calculate the propagation characteristics of the ultrasonic guided wave in the cylinder, and the phase velocity and group velocity curves of the guided wave are obtained; The optimal excitation frequency is determined based on the slope of the group velocity dispersion curve; The ultrasonic guided wave signal is collected by a piezoelectric sensor array, and the collected original signal is subjected to bandpass filtering and Hilbert transformation to extract the envelope characteristics of the signal; A time reversal algorithm is used to invert and delay the signal to obtain a focused sound field distribution diagram; The focused sound field data is processed by a synthetic aperture focusing algorithm, and the coordinates of the crack position are calculated by triangulation using an arrival time method in combination with a three-dimensional geometric model of the cylinder and the propagation characteristics of the guided wave; Based on the reconstructed three-dimensional crack morphology, the Otsu threshold segmentation algorithm is used to extract the crack edge contour, and the crack area change between two adjacent samplings is calculated; The crack growth rate is calculated by the finite difference method with central difference format, and the crack growth process of the cylinder block is quantitatively analyzed. A bilinear interpolation algorithm is used to generate a spatiotemporal distribution diagram of crack growth rate, wherein the horizontal axis of the spatiotemporal distribution diagram represents time, the vertical axis represents crack length, and the color depth represents the growth rate; It also includes: by modeling the vibration, impact and fatigue dynamic loads of the cylinder body, calculating the momentum distribution of the crack area, correlating the momentum distribution with the crack growth rate, and analyzing the influence of the dynamic load on the crack growth; establishing the relationship between momentum and crack growth rate, predicting the development trend of the crack, and determining the stress field distribution and crack growth driving force at the crack tip in the crack initiation stage, the crack stable growth stage and the crack rapid growth stage to further analyze the progressive growth of the cylinder body crack, specifically including: establishing a dynamic load model according to the geometric structure and material parameters of the cylinder body, calculating the stress field distribution and displacement field distribution in the crack area; using a multivariate regression algorithm to establish the correlation between momentum distribution and crack growth rate, where The momentum, direction and gradient are selected as independent variables, and the crack growth rate is the dependent variable; the stress intensity factor at the crack tip is calculated using the fracture mechanics theory, and the J integral is calculated using the virtual displacement method, in which the virtual displacement field is defined, the strain energy density and the traction work are calculated; the stress field distribution at the crack tip is solved using the finite difference method, and the crack growth driving force is calculated using the stress intensity factor method; the crack growth rate is calculated in combination with the Paris formula da / dN=C(ΔK)^m, in which da / dN is the crack growth rate, ΔK is the stress intensity factor range, and C and m are material constants; the crack growth amount is accumulated by numerical integration using the Runge-Kutta method to obtain the quantitative analysis results of the progressive growth of the cylinder crack.

5. The method according to claim 1, characterized in that The engine speed, load, and temperature operating condition parameters are obtained, and fatigue load spectra under different operating condition combinations are obtained according to the distribution of thermal stress and mechanical stress under different operating conditions, including stress amplitude, stress ratio, and number of cycles. If the crack growth rate exceeds a preset threshold, the operating condition combination is judged to be a high-risk operating condition for accelerated crack growth, including: The engine control unit acquires engine speed, load and temperature operating parameters, and the operating parameters are converted from analog signals to digital signals by a data acquisition card; According to the working condition parameters, the thermal stress and mechanical stress distribution of the cylinder under different working conditions are calculated by using the finite element method, wherein the finite element method adopts a thermal-structural coupling analysis method; The rain flow counting method is used to count the cycles of the stress time history, wherein the rain flow counting method includes extracting stress peak and valley values, forming stress-time pairs, pairing to eliminate small cycles and counting residual stress amplitudes; The fatigue damage accumulation value is calculated using Miner's linear cumulative damage theory, and the fatigue damage accumulation value is obtained from the fatigue load spectrum under different working condition combinations; The crack growth rate under the different working condition combinations is calculated using the Paris formula. If the crack growth rate exceeds a preset threshold, a fuzzy comprehensive evaluation method is used to conduct a risk assessment on the working condition combination. The evaluation indexes of the fuzzy comprehensive evaluation method include stress amplitude, stress ratio and number of cycles, and the weights of the evaluation indexes are 0.5, 0.3 and 0.2 respectively.

6. The method according to claim 1, characterized in that The material mechanical property parameters combined with fracture toughness are used to predict the remaining service life of the cylinder under given working conditions according to crack size, crack growth rate and high-risk working condition parameters, including: According to the material mechanical properties and fracture toughness data, the critical crack length is calculated using fracture mechanics theory, and finite element analysis is performed using ANSYS software to obtain stress intensity factors under different crack sizes. If the stress intensity factor has been determined, the relationship between the crack growth rate and the range of the stress intensity factor is established using the Paris formula, and the range of the stress intensity factor under a given working condition is calculated by combining the high-risk working condition parameters and the influence of temperature on material properties; According to the range of stress intensity factors, the Monte Carlo method is used to simulate the crack propagation process, and the random number generation adopts the MersenneTwister algorithm. In combination with the randomness of material parameters and loads, the material property fluctuation is described by normal distribution, and the load fluctuation is described by Weibul l distribution, so as to generate a large number of crack propagation paths; If the crack propagation path has been generated, the simulation results are smoothed using the kernel density estimation method, using the Gaussian kernel function and the Silverman rule for bandwidth selection; Based on the crack propagation path and cycle number distribution after smoothing, the remaining life probability distribution function is fitted using Weibul distribution, the shape parameters and scale parameters are determined by maximum likelihood estimation method, and the expected value and 95% confidence interval of the remaining service life of the cylinder block under given working conditions are calculated; It also includes: characterizing the crack size by ultrasonic testing or X-ray testing to obtain the geometric size information of the crack, combining the fracture toughness of the material, building the relationship between the crack size and the remaining service life of the cylinder, and realizing the prediction of the remaining service life of the cylinder under given working conditions, specifically including: scanning the cylinder by phased array ultrasonic testing technology, processing the acquired echo signal by time reversal algorithm, and obtaining the three-dimensional morphology information of the crack; according to the three-dimensional morphology information, imaging the cylinder by X-ray computed tomography technology, extracting the crack contour by Otsu threshold segmentation algorithm, and combining FDK algorithm for three-dimensional reconstruction to obtain The volume and shape characteristics of the crack are obtained; the Dempster-Shafer evidence theory is used to fuse the ultrasonic detection results and the X-ray detection results to determine the precise size parameters of the crack; based on the fracture mechanics theory, the finite element method is used to calculate the stress intensity factor of the crack at different sizes, and the relationship between the crack growth rate and the stress intensity factor is established through the Paris formula; the support vector regression algorithm is used to construct a correlation model between the crack size, operating condition parameters and the remaining life of the cylinder body, and if the prediction accuracy of the correlation model meets the preset threshold, the remaining life prediction result of the cylinder body under the given working condition is output.

7. The method according to claim 1, characterized in that The combined analysis result of the cylinder crack propagation and the remaining service life of the cylinder are analyzed to set the early warning thresholds of the crack size, propagation rate and high-risk working conditions. When the monitoring parameters exceed the thresholds, an alarm message is generated to prompt the taking of corresponding repair and maintenance measures to avoid the crack from entering the rapid crack propagation stage, including: A support vector machine algorithm is used to establish a relationship model between crack size, growth rate, operating parameters and remaining life. The model uses an RBF kernel function and optimizes parameters through a grid search method. According to the relationship model, the Monte Carlo simulation method is used to generate sample data, calculate the crack propagation path and remaining life distribution under different conditions, and determine the crack size, propagation rate and warning threshold of high-risk working conditions; According to the warning threshold, a multi-parameter warning index system is constructed by using fuzzy comprehensive evaluation method, and the triangular membership function of crack size, expansion rate and working condition parameters is set. The weight of each index is determined by hierarchical analysis method, and the comprehensive warning index is calculated; Obtain past historical data, use the sliding time window method, dynamically update the threshold according to the historical data, and improve the adaptability of the early warning system; Real-time monitoring data is stored based on the InfluxDB time series database, and real-time data analysis is performed using the Apache Flink stream processing framework. When the monitoring parameters exceed the threshold or the comprehensive warning index exceeds the preset value, the alarm information generation module is triggered and the warning information is pushed to the relevant system.

8. The method according to claim 1, characterized in that The method optimizes the engine operating conditions, reduces the frequency of high-risk conditions, extends the service life of the cylinder, and improves the accuracy of crack detection and identification based on the critical length of crack instability, the cylinder crack extension analysis results, and the high-risk conditions judgment results, including: According to the working condition parameters and crack growth rate data, a relationship model is established by using a support vector regression algorithm, and the relationship model uses an RBF kernel function and optimizes parameters through a grid search method; Obtaining the output result of the relationship model, and using a genetic algorithm to perform multi-objective optimization on the engine operating conditions, wherein the multi-objective optimization takes reducing the frequency of occurrence of high-risk operating conditions and the crack growth rate as optimization goals; Analyzing the multi-objective optimization results, performing sensitivity analysis of operating parameters using variance analysis, wherein the variance analysis method calculates the F value of each parameter to identify the operating parameters that have the most significant impact on crack propagation; For the optimized working condition parameters, the Monte Carlo method is used to simulate the crack propagation process of the cylinder under the new working condition. The Monte Carlo method generates random samples and considers the randomness of material properties and loads. Based on the simulation results of the Monte Carlo method, a crack detection and identification model is constructed using a convolutional neural network. The convolutional neural network takes the optimized operating parameters and crack propagation characteristics under new operating conditions as input, and fine-tunes the pre-trained model through a transfer learning method.

9. An engine internal damage identification system based on target detection, characterized in that: The system comprises: Finite element analysis module, used to analyze the thermal and mechanical stress distribution of the cylinder under different working conditions; Ultrasonic guided wave detection module, used to monitor the amplitude of the acoustic emission signal during crack propagation and determine the crack state; Crack reconstruction module, used to reconstruct the three-dimensional crack morphology, determine the crack growth rate and analyze the growth behavior; Fatigue load spectrum module, used to obtain fatigue load spectrum according to working condition parameters and identify high-risk working conditions; Life prediction module, used to predict the remaining service life of the cylinder body by combining the crack growth rate and operating parameters; The early warning and optimization module is used to generate alarms when monitoring parameters exceed thresholds and optimize operating conditions to extend the life of the cylinder.

Images