Multi-sensor fused laser shock peening quality monitoring device and method

Abstract

The invention discloses a multi-sensor fusion laser shock strengthening quality monitoring device and method, and relates to the technical field of laser surface strengthening, and the method comprises the following steps: 1, collecting laser energy spectrum characteristics, flow matching curves, sound signal characteristic templates and standard surface quality images of different materials under different process conditions, constructing a process parameter reference database; 2, multi-source signals in the machining process are collected; 3, preprocessing the multi-source signal to generate a multi-dimensional feature matrix; and 4, performing comprehensive analysis according to the multi-dimensional feature matrix and the process parameter reference database to obtain a quality monitoring result. According to the invention, through deep integration of multi-sensor fusion and an intelligent control technology, omnibearing real-time monitoring of a laser shock peening process is realized.

Description

Technical Field

[0001] This invention relates to the field of laser surface strengthening technology, and more specifically to a multi-sensor fusion laser shock strengthening quality monitoring device and method. Background Technology

[0002] Laser shock peening technology uses high-energy pulsed laser-induced shock waves to induce plastic deformation on the surface of materials, thereby improving the fatigue resistance and wear resistance of parts. The effectiveness of this process is significantly influenced by multiple coupled parameters: In terms of processing parameters, laser energy directly determines the peak pressure of the shock wave, and its fluctuations will lead to deviations in the depth of the strengthened layer; the focusing quality and focal point position affect the spatial distribution of energy, and even slight deviations can cause distortion of the energy density gradient; the matching relationship between the shock frequency and the overlap rate of the laser spot directly determines the uniformity of plastic deformation, and excessively high frequencies can easily cause hysteresis in the recovery of the water film in the constraint layer, resulting in attenuation of the shock wave energy. In terms of environmental parameters, the thickness of the water film in the constraint layer needs precise control; too thick a film can easily induce a plasma shielding effect, while too thin a film leads to a sharp increase in the pressure wave propagation loss rate.

[0003] Currently, commercial laser shock blasting equipment generally suffers from three major technical bottlenecks: First, quality fluctuations are uncontrollable. Abnormal conditions such as laser energy drift, water film fluctuations in the constraint layer, and aluminum foil rupture in the absorption layer are difficult to detect in a timely manner, leading to uneven distribution of residual stress. Second, detection methods are outdated, relying on manual visual inspection or offline inspection after processing, which cannot interrupt the defect transmission chain, resulting in a high rework rate. Third, monitoring dimensions are limited. Existing systems mostly rely on a single sensor, lacking response to key quality characteristics such as acoustic emission signals and abrupt changes in surface morphology, resulting in a high misjudgment rate when processing complex curved workpieces. These shortcomings severely restrict the large-scale application of laser shock blasting technology in high-end manufacturing.

[0004] Therefore, how to achieve comprehensive quality inspection of laser shock peening, improve control accuracy, and avoid abnormal response lag is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of the above problems, the present invention is proposed to provide a multi-sensor fusion laser shock enhancement quality monitoring device and method to overcome or at least partially solve the above problems, realize real-time monitoring of the processing process, and help to achieve adaptive quality control.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, embodiments of the present invention provide a method for monitoring the quality of laser shock enhancement through multi-sensor fusion, comprising the following steps: Step 1: Collect laser energy spectrum characteristics, flow matching curves, acoustic signal feature templates and standard surface quality images under different materials and process conditions; set the set values ​​for laser energy, water film flow rate and acoustic spectrum; and construct a process parameter benchmark database in combination with the standard surface quality images. Step 2: Acquire multi-source signals during the processing; Step 3: Preprocess the multi-source signals to generate a multi-dimensional feature matrix; Step 4: Conduct a comprehensive analysis based on the multidimensional feature matrix and the process parameter benchmark database to obtain quality monitoring results.

[0008] Preferably, the multi-source signals include laser energy signals, flow signals, surface images, and acoustic signals; the laser energy signals include laser power and pulse frequency.

[0009] Preferably, the preprocessing employs a hierarchical denoising and multidimensional normalization strategy to achieve accurate time synchronization and spatial registration, generating a multidimensional feature matrix. The hierarchical denoising and multidimensional normalization strategy includes Kalman filtering, moving average filtering, median filtering, Gaussian filtering, bandpass filtering, and normalization.

[0010] Preferably, based on the laser energy signal, flow rate signal, and acoustic signal in the multidimensional feature matrix, and combined with the set values ​​of laser energy, water film flow rate, and acoustic spectrum in the process parameter benchmark database, the processing quality is calculated, including laser quality, flow rate quality, and noise level; based on the surface image and standard surface quality image, edge detection using the Canny operator and region growing algorithm is performed to identify defect regions in the surface image and calculate the geometric feature parameters of the defect regions; the workpiece surface quality score is calculated based on the geometric feature parameters; the processing quality and the workpiece surface quality score constitute the quality monitoring result.

[0011] Preferably, the deviation between the measured laser energy signal and the laser energy setpoint in the process parameter reference database is calculated based on ΔE=α×(P^β×f^γ-P_target^β×f_target^γ) to obtain the energy deviation ΔE, where α=0.85, β=0.6, γ=0.3, P_target represents the laser power setpoint, f_target represents the pulse frequency setpoint, P represents the measured laser power, and f represents the measured pulse frequency. When the energy deviation is less than the lower energy threshold, the laser quality is slightly deviated; when the energy deviation is within the range of the lower and upper energy thresholds, the laser quality is moderately deviated; and when the energy deviation is greater than the upper energy threshold, the laser quality is severely deviated. The flow deviation is calculated based on the collected flow signals and the water film flow setpoint in the process parameter benchmark database. When the flow deviation is less than the lower limit threshold, the flow quality is slightly deviated. When the flow deviation is within the range of the lower and upper limits thresholds, the flow quality is moderately deviated. When the flow deviation is greater than the upper limit threshold, the flow quality is severely deviated. Multi-domain feature extraction technology is used to extract the spectral features of acoustic signals in the 100-10kHz frequency band. The system status is determined based on the extracted spectral features and the spectral grading criteria constructed based on the acoustic spectrum setpoint. The spectral grading criteria are as follows: if the similarity between the spectral features and the acoustic spectrum setpoint is greater than 0.8, the noise quality is normal; if the similarity between the spectral features and the acoustic spectrum setpoint is 0.6-0.8, the noise quality is in warning mode; and if the similarity between the spectral features and the acoustic spectrum setpoint is less than 0.6, the noise quality is abnormal.

[0012] Preferably, the defect area includes breakage and peeling, and the geometric feature parameters include failure area, area ratio, aspect ratio and roundness, etc.; the workpiece surface quality score Q_score is expressed as: Q_score=100-(α×A_defect / A_total), where α represents the evaluation coefficient, A_defect represents the failure area, and A_total represents the total area. The total area is calculated based on the failure area and area ratio.

[0013] Preferably, the method further includes step 5, which generates a response action based on the quality monitoring results and a graded response strategy. The graded response strategy includes: when the laser quality and flow rate quality have slight deviations and the noise level is at the warning level, and the laser energy and water film flow rate cannot be adjusted to a steady state, a level 1 response is executed to adjust the signal; when the laser quality and flow rate quality have moderate deviations and the noise level is at the warning level, and the laser energy, water film flow rate, and acoustic signal cannot be adjusted to a steady state, a level 2 response is executed to suspend the alarm and initiate manual intervention; when the laser quality and flow rate quality have severe deviations, the noise level is abnormal, and the workpiece surface quality score is lower than the set surface threshold, a level 3 response is executed to initiate an emergency shutdown, and the adjustment signal is fed back to the process parameter benchmark database for recording.

[0014] Preferably, a Level 1 response includes: An improved fuzzy PID control is used to adjust the laser energy signal. The formula P_newtarget=P_current×(1+K1×ΔE+K2×ΔQ+K3×ΔS) is used for comprehensive compensation adjustment, where ΔE is the energy deviation, ΔQ is the flow deviation, ΔS is the acoustic signal deviation, K1=0.5, K2=0.3, K3=0.2, P_newtarget represents the target laser power, and P_current is the current laser power. If the adjustment still exceeds the specified range, it is judged as a laser malfunction and a shutdown protection is triggered. The flow signal is adjusted using a PID control loop, expressed as u(t)=Kp×e(t)+Ki×∫e(t)dt+Kd×de(t) / dt, where u(t) represents the target flow rate, e(t) represents the flow deviation, Kp represents the proportional gain, Ki represents the integral gain, and Kd represents the derivative gain.

[0015] Preferably, spectral features in the acoustic signal are extracted using multi-domain feature extraction technology, including time-domain features (RMS value, peak factor, impulse factor), frequency-domain features (power spectral density, frequency centroid, spectral entropy), and time-frequency features (wavelet packet decomposition coefficients, MFCC parameters). The similarity S between the spectral features and the acoustic spectrum setting is calculated using weighted Euclidean distance, with the formula S=Σ(wi×|Fi_current-Fi_standard|), where Fi_current represents the current spectral feature and Fi_standard represents the acoustic spectrum setting.

[0016] Preferably, if a defective area is identified and a broken or peeling area is found, the laser emission is immediately interrupted, the laser shock strengthening equipment is controlled to move out of the workpiece to a safe position, the surface reprocessing procedure is started, and abnormal data is recorded for system optimization.

[0017] Preferably, during laser shock peening, the workpiece is mounted on a robotic arm. During the laser shock peening process, displacement and angle sensors are used to collect the displacement and angle of the robotic arm fixing the workpiece. The robotic arm pose is corrected based on the displacement and angle to adjust the laser emission angle. The robotic arm pose correction steps are as follows: A dual closed-loop control strategy is used, employing algorithms X_correct = X_target + Kp_pos × (X_measured - X_target) and θ_correct = θ_target + Kp_angle × (θ_measured - θ_target) to achieve precise six-degree-of-freedom positioning. Here, X_correct is the displacement error, X_measured is the collected displacement measurement value, X_target is the set displacement target value, Kp_pos = 0.8, θ_correct is the angle error, θ_target is the angle target value, θ_measured is the angle measurement value, and Kp_angle = 0.6.

[0018] A multi-sensor fusion laser shock-enhanced quality monitoring device, comprising: Multi-sensor array module and intelligent control system; The system comprises a multi-sensor array module to acquire signals from multiple sources. The intelligent control system includes a database, a signal preprocessing unit, and a decision control unit. The database stores a benchmark database of process parameters. The signal preprocessing unit receives and processes the acquired multi-source signals in real time, generating a multi-dimensional feature matrix. The decision control unit performs comprehensive analysis based on the multi-dimensional feature matrix and the benchmark database of process parameters, establishing a process parameter matching model and an anomaly feature database to obtain quality monitoring results. The system receives and processes multi-source signals in real time, performs fusion analysis based on machine learning algorithms, and obtains comprehensive analysis results for quality monitoring.

[0019] Preferably, the multi-sensor array module includes a laser energy sensor, a flow rate sensor, a surface quality sensor, and an acoustic wave sensor, constructing a comprehensive quality monitoring system covering energy input, workpiece positioning, constraint layer status, surface quality, and impact effects; wherein, The laser energy sensor uses high-precision photoelectric detection technology to monitor laser power and pulse frequency in real time. It is installed inside the laser and uses a beam splitter to measure the laser energy. The flow velocity sensor is installed at key nodes of the water supply pipeline in the constraint layer to collect flow signals and detect water film flow fluctuations in real time. The surface quality sensor is equipped with a high-resolution industrial camera, consisting of a visible light sensor and an 850nm infrared light sensor, which are positioned at a 15° angle to the laser beam. Combined with image recognition algorithms, it monitors surface damage and defects in the absorption layer. The acoustic sensors employ directional sound pickup technology and are positioned at the four corners of the processing area to comprehensively collect acoustic signals during the impact process.

[0020] Preferably, the signal preprocessing unit employs a layered noise reduction and multi-dimensional normalization strategy to perform noise reduction and normalization processing on multi-source sensor signals. The noise reduction processing includes Kalman filtering, moving average filtering, median filtering, Gaussian filtering, and bandpass filtering; the normalization processing includes normalization processing.

[0021] Preferably, the decision control unit calculates the processing quality, including laser quality, flow rate quality, and noise level, based on the laser energy signal, flow rate signal, and acoustic signal in the multi-dimensional feature matrix, combined with the set values ​​of laser energy, water film flow rate, and acoustic spectrum in the process parameter benchmark database. It then uses Canny operator edge detection and region growing algorithms to detect and identify defective regions based on surface images and surface quality image standards, calculating the geometric feature parameters of these regions. Based on these geometric feature parameters, it calculates the workpiece surface quality score. The processing quality and the workpiece surface quality score together form the quality monitoring result, possessing intelligent process parameter matching analysis and accurate abnormal state identification functions.

[0022] Preferably, the intelligent control system also includes an execution output unit, which generates response actions based on quality monitoring results and a graded response strategy to achieve multi-level alarms, and uses a PID control algorithm to adjust the signal execution response. Key parameters such as laser output power, pulse frequency, and robotic arm motion trajectory are dynamically adjusted based on quality monitoring results, and a graded protection mechanism is triggered when abnormal operating conditions are detected.

[0023] Preferably, a displacement sensor and an angle sensor are also provided to collect the displacement and angle of the workpiece, respectively, and to correct the posture of the robotic arm that fixes the workpiece.

[0024] As can be seen from the above technical solution, compared with the prior art, this invention discloses a multi-sensor fusion laser shock strengthening quality monitoring device and method, applicable to real-time monitoring and control of the quality of high-precision parts processed by laser shock strengthening in aerospace, automotive manufacturing and other fields. Through the deep integration of multi-sensor fusion and intelligent control technologies, comprehensive real-time monitoring and adaptive quality control of the laser shock strengthening process are achieved. Compared with traditional single-sensor monitoring systems, this invention can promptly and accurately detect process parameter deviations, equipment malfunctions, and surface quality problems, significantly improving quality control accuracy and production efficiency, increasing quality detection accuracy by more than 35%, and shortening the anomaly response time to the millisecond level. A multi-level hierarchical protection mechanism effectively ensures processing safety, avoiding equipment damage and workpiece scrap; the self-learning function, through continuous optimization of the process parameter model, can adapt to the processing needs of different material properties and complex workpiece geometries. This technology fundamentally solves the key technical problems that have long existed in the field of laser shock strengthening, such as large quality fluctuations, detection lag, and insufficient control accuracy, providing a reliable and efficient quality assurance solution for high-end manufacturing industries such as aerospace, nuclear power, and marine engineering. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0026] Figure 1 This is a flowchart of a multi-sensor fusion laser shock enhancement quality monitoring method provided in an embodiment of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] This invention discloses a multi-sensor fusion method for monitoring the quality of laser shock enhancement, such as... Figure 1 As shown, it includes the following steps: S1: Collect laser energy spectrum characteristics, flow matching curves, acoustic signal feature templates and standard surface quality images under different materials and process conditions; set the set values ​​of laser energy, water film flow rate and acoustic spectrum; and construct a process parameter benchmark database in combination with the standard surface quality images. S2: Acquires multi-source signals during the processing; S3: Preprocess the multi-source signals to generate a multi-dimensional feature matrix; S4: Based on the multidimensional feature matrix and the process parameter benchmark database, a comprehensive analysis is conducted to obtain the quality monitoring results.

[0029] Furthermore, the multi-source signals include laser energy signals, flow signals, surface images, and acoustic signals; the laser energy signals include laser power and pulse frequency.

[0030] Furthermore, the preprocessing employs hierarchical denoising and multidimensional normalization strategies to achieve accurate time synchronization and spatial registration, generating a multidimensional feature matrix. The hierarchical denoising and multidimensional normalization strategies include Kalman filtering, moving average filtering, median filtering, Gaussian filtering, bandpass filtering, and normalization.

[0031] Furthermore, based on the laser energy signal, flow rate signal, and acoustic signal in the multidimensional feature matrix, and combined with the set values ​​of laser energy, water film flow rate, and acoustic spectrum in the process parameter benchmark database, the processing quality is calculated, including laser quality, flow rate quality, and noise level. Based on the surface image and standard surface quality image, edge detection using the Canny operator and region growing algorithm are performed to identify defect regions in the surface image and calculate the geometric feature parameters of the defect regions. The workpiece surface quality score is then calculated based on these geometric feature parameters. The processing quality and the workpiece surface quality score together constitute the quality monitoring result.

[0032] Furthermore, the deviation between the measured laser energy signal and the laser energy setting value in the process parameter reference database is calculated based on ΔE=α×(P^β×f^γ-P_target^β×f_target^γ), to obtain the energy deviation ΔE, where α=0.85, β=0.6, γ=0.3, P_target represents the laser power setting value, f_target represents the pulse frequency setting value, P represents the measured laser power, and f represents the measured pulse frequency. When the energy deviation is less than the lower energy threshold, the laser quality is slightly deviated; when the energy deviation is within the range of the lower and upper energy thresholds, the laser quality is moderately deviated; and when the energy deviation is greater than the upper energy threshold, the laser quality is severely deviated. The flow deviation is calculated based on the collected flow signals and the water film flow setpoint in the process parameter benchmark database. When the flow deviation is less than the lower limit threshold, the flow quality is slightly deviated. When the flow deviation is within the range of the lower and upper limits thresholds, the flow quality is moderately deviated. When the flow deviation is greater than the upper limit threshold, the flow quality is severely deviated. Multi-domain feature extraction technology is used to extract the spectral features of acoustic signals in the 100-10kHz frequency band. The system status is determined based on the extracted spectral features and the spectral grading criteria constructed based on the acoustic spectrum setpoint. The spectral grading criteria are as follows: if the similarity between the spectral features and the acoustic spectrum setpoint is greater than 0.8, the noise quality is normal; if the similarity between the spectral features and the acoustic spectrum setpoint is 0.6-0.8, the noise quality is in warning mode; and if the similarity between the spectral features and the acoustic spectrum setpoint is less than 0.6, the noise quality is abnormal.

[0033] Furthermore, the defect area includes breakage and peeling, and the geometric feature parameters include failure area, area ratio, aspect ratio, and roundness, etc.; the workpiece surface quality score Q_score is expressed as: Q_score=100-(α×A_defect / A_total), where α represents the evaluation coefficient, A_defect represents the failure area, and A_total represents the total area. The total area is calculated based on the failure area and area ratio.

[0034] Furthermore, the defect area includes breakage and peeling, and the geometric feature parameters include failure area, area ratio, aspect ratio, and roundness, etc.; the workpiece surface quality score Q_score is expressed as: Q_score=100-(α×A_defect / A_total), where α represents the evaluation coefficient, A_defect represents the failure area, and A_total represents the total area. The total area is calculated based on the failure area and area ratio.

[0035] Furthermore, it also includes S5, which generates response actions based on quality monitoring results and a graded response strategy. The graded response strategy includes: when the laser quality and flow rate quality have slight deviations and the noise level is at the warning level, and the laser energy and water film flow rate cannot be adjusted to a steady state, a level 1 response is executed to adjust the signal; when the laser quality and flow rate quality have moderate deviations and the noise level is at the warning level, and the laser energy, water film flow rate, and acoustic signal cannot be adjusted to a steady state, a level 2 response is executed to suspend the alarm and require manual intervention; when the laser quality and flow rate quality have severe deviations, the noise level is abnormal, and the workpiece surface quality score is lower than the set surface threshold, a level 3 response is executed to shut down the machine immediately and simultaneously feed the adjustment signal back to the process parameter benchmark database for recording.

[0036] Further Level 1 responses include: An improved fuzzy PID control is used to adjust the laser energy signal. The formula P_newtarget=P_current×(1+K1×ΔE+K2×ΔQ+K3×ΔS) is used for comprehensive compensation adjustment, where ΔE is the energy deviation, ΔQ is the flow deviation, ΔS is the acoustic signal deviation, K1=0.5, K2=0.3, K3=0.2, P_newtarget represents the target laser power, and P_current is the current laser power. If the adjustment still exceeds the specified range, it is judged as a laser malfunction and a shutdown protection is triggered. The flow signal is adjusted using a PID control loop, expressed as u(t)=Kp×e(t)+Ki×∫e(t)dt+Kd×de(t) / dt, where u(t) represents the target flow rate, e(t) represents the flow deviation, Kp represents the proportional gain, Ki represents the integral gain, and Kd represents the derivative gain.

[0037] Furthermore, based on the acoustic signal feature template, multi-domain feature extraction technology is used to analyze the spectral characteristics of the acoustic signal in the 100-10kHz frequency band, including time-domain features (RMS value, peak factor, impulse factor), frequency-domain features (power spectral density, frequency centroid, spectral entropy), and time-frequency features (wavelet packet decomposition coefficients, MFCC parameters). The similarity between the spectral characteristics and the acoustic spectrum setting value is calculated using weighted Euclidean distance S=Σ(wi×|Fi_current-Fi_standard|) for pattern matching analysis, where Fi_current represents the current spectral characteristics and Fi_standard represents the acoustic spectrum setting value.

[0038] Furthermore, if a damaged or peeling area appears, the laser emission is immediately interrupted, the laser shock strengthening equipment is controlled to move out of the workpiece to a safe position, the surface reprocessing procedure is started, and abnormal data is recorded for system optimization.

[0039] Furthermore, during the laser shock peening process, the workpiece is mounted on a robotic arm. During this process, displacement and angle sensors are used to collect the workpiece's displacement and angle, respectively. Based on the displacement and angle, the robotic arm's pose is corrected to adjust the laser emission angle. The robotic arm pose correction steps are as follows: A dual closed-loop control strategy is employed, using the algorithms X_correct = X_target + Kp_pos × (X_measured - X_target) and θ_correct = θ_target + Kp_angle × (θ_measured - θ_target) to achieve precise six-degree-of-freedom positioning. Here, X_correct is the displacement error, X_measured is the collected displacement measurement value, X_target is the set displacement target value, Kp_pos = 0.8, θ_correct is the angle θ error, θ_target is the angle θ target value, θ_measured is the angle measurement value, and Kp_angle = 0.6.

[0040] This invention employs multi-sensor fusion technology, integrating various detection units such as laser energy sensors, displacement sensors, flow velocity sensors, surface quality sensors, and acoustic wave sensors to construct a comprehensive quality monitoring system covering energy input, workpiece positioning, constraint layer status, surface quality, and impact effects. The system utilizes machine learning algorithms to deeply analyze the inherent correlations between the monitoring parameters of each sensor, establishing a precise process parameter matching model and a complete database of abnormal features. During laser shock peening processing, the control system collects multi-source sensor data in real time, accurately identifying process parameter deviations and abnormal operating conditions through advanced signal fusion analysis technology. When the system detects a severe abnormal state, it immediately triggers a graded protection mechanism, effectively ensuring processing safety and product quality stability. Furthermore, it can automatically adjust key parameters such as laser output power, pulse frequency, and robotic arm trajectory according to preset intelligent control strategies, achieving high-precision closed-loop feedback control.

[0041] On the other hand, a multi-sensor fusion laser shock enhancement quality monitoring device includes: Multi-sensor array module and intelligent control system; The multi-sensor array module collects signals from multiple sources; the intelligent control system includes a database, a signal preprocessing unit, and a decision control unit; the database stores a benchmark database of process parameters; the signal preprocessing unit receives the collected multi-source signals in real time and performs preprocessing to generate a multi-dimensional feature matrix; the decision control unit performs comprehensive analysis based on the multi-dimensional feature matrix and the benchmark database of process parameters, establishes a process parameter matching model and an anomaly feature database, and obtains quality monitoring results.

[0042] This is used to receive and collect multi-source signals in real time. Based on machine learning algorithms, the received signals are fused and analyzed to establish a process parameter matching model and an abnormal feature database for comprehensive analysis to obtain quality monitoring results.

[0043] Furthermore, the multi-sensor array module includes a laser energy sensor, a displacement sensor, a flow velocity sensor, a surface quality sensor, and an acoustic sensor, constructing a comprehensive quality monitoring system covering energy input, workpiece positioning, constraint layer status, surface quality, and impact effects; wherein, The laser energy sensor uses high-precision photoelectric detection technology to monitor laser power and pulse frequency in real time. It is installed inside the laser and uses a beam splitter to measure the laser energy to ensure the stability of laser output. The flow velocity sensor is installed at key nodes in the water supply pipeline of the constraint layer to collect flow signals, detect water film flow fluctuations in real time, and maintain the best constraint effect. The surface quality sensor is equipped with a high-resolution industrial camera, consisting of a visible light sensor and an 850nm infrared light sensor, which are positioned at a 15° angle to the laser beam. Combined with image recognition algorithms, it monitors surface damage and defects in the absorption layer. The acoustic sensors employ directional sound pickup technology and are positioned at the four corners of the processing area to comprehensively collect acoustic signals during the impact process.

[0044] Furthermore, the signal preprocessing unit performs noise reduction processing on multi-source signals. It adopts a hierarchical noise reduction and multi-dimensional normalization strategy to perform noise reduction and normalization processing on multi-source sensor signals. The noise reduction processing uses Kalman filtering, moving average filtering, median filtering, Gaussian filtering and bandpass filtering; the normalization processing uses a normalization formula for mapping. The laser energy signal is effectively eliminated by using an adaptive Kalman filter algorithm to remove random noise and system drift during the photoelectric conversion process; the displacement signal is filtered by a moving average filter with a window length of 10 sampling points to remove mechanical vibration interference; the flow velocity signal is filtered by median filtering combined with a low-pass filter with a cutoff frequency of 50Hz to eliminate the influence of pipeline pulsation; the visual signal is effectively removed by Gaussian filtering (σ=1.2) and morphological opening and closing operations to remove image noise and artifacts; the acoustic signal is filtered by a bandpass filter of 100Hz-10kHz combined with an adaptive noise cancellation algorithm to remove environmental interference. The normalization formula X_norm=(X-X_min) / (X_max-X_min) is used to map the data of various sensors to the standard interval [0,1]. The laser pulse trigger signal is used as the time reference to perform high-precision timestamp alignment on all sensor data, and all signals are resampled to a uniform frequency of 50Hz to ensure the spatiotemporal synchronization and comparability of multi-source data. Furthermore, the decision control unit calculates the processing quality, including laser quality, flow rate quality, and noise level, based on the laser energy signal, flow rate signal, and acoustic signal in the multi-dimensional feature matrix, combined with the set values ​​of laser energy, water film flow rate, and acoustic spectrum in the process parameter benchmark database. It also performs edge detection using the Canny operator and region growing algorithm based on surface images and surface quality image standards to identify defective regions and calculate their geometric feature parameters. The workpiece surface quality score is then calculated based on these geometric feature parameters. The processing quality and workpiece surface quality score together form the quality monitoring results, enabling intelligent matching analysis of process parameters and accurate identification of abnormal states.

[0045] Furthermore, the intelligent control system also includes an execution output unit. Based on the quality monitoring results from multi-sensor fusion analysis and combined with a hierarchical response strategy, it generates response actions and uses a PID control algorithm to adjust the signal execution response. By generating laser power adjustment commands and multi-level alarm signals, it achieves real-time control and adjustment, dynamically adjusting key parameters such as laser output power and pulse frequency. When abnormal operating conditions are detected, a hierarchical protection mechanism is triggered.

[0046] Furthermore, displacement sensors and angle sensors are also installed to collect the displacement and angle of the workpiece, respectively, and to correct the posture of the robotic arm that fixes the workpiece. The displacement sensor adopts non-contact photoelectric positioning technology and is installed coaxially with the servo motor to accurately monitor the workpiece positioning deviation and ensure processing accuracy.

[0047] Furthermore, the robotic arm pose correction employs a dual closed-loop control strategy. Position compensation uses the algorithm X_correct = X_target + Kp_pos × (X_measured - X_target), and angle compensation uses the algorithm θ_correct = θ_target + Kp_angle × (θ_measured - θ_target) to achieve precise six-degree-of-freedom positioning. Here, X_correct is the displacement error, X_measured is the acquired displacement measurement value, X_target is the set displacement target value, the position control gain Kp_pos = 0.8, θ_correct is the angle θ error, θ_target is the angle θ target value, θ_measured is the angle measurement value, and the angle control gain Kp_angle = 0.6.

[0048] Furthermore, the alarm signal generation adopts an intelligent multi-level alarm mechanism: Level 1 response triggers a yellow warning signal when the deviation of a single key parameter exceeds the set threshold by 80%; Level 2 response activates an orange alarm signal and suspends processing when the deviation of a single parameter exceeds the threshold or the deviation of multiple parameters exceeds 60% simultaneously; Level 3 response shutdown protection immediately executes a red emergency shutdown signal when the deviation of a key parameter exceeds the safety threshold or a serious surface defect is detected, comprehensively ensuring the safety of the processing process and the reliability of product quality.

[0049] On the other hand, in one specific embodiment, laser shock peening quality monitoring and surface damage treatment of aero-engine blades are carried out.

[0050] This embodiment is applied to the laser shock peening process of titanium alloy blades for aero-engines. The laser shock peening processing device consists of a pulsed laser, a six-degree-of-freedom robotic arm, a multi-sensor module, and a distributed control system.

[0051] The pulsed laser has a wavelength of 1064nm and a pulse energy of 2-25J; the six-degree-of-freedom robotic arm has a repeatability accuracy of ±0.02mm; the sensor array is specifically configured as follows: the laser energy sensor is integrated into the optical path reflector group with a sampling frequency of 1kHz; the laser displacement sensor is installed at the end of the robotic arm, measuring a distance of 30mm from the workpiece surface with an accuracy of 2μm; the electromagnetic flowmeter is installed in the water confinement layer supply pipeline with a flow range of 0.5-15L / min; the industrial camera with 5 megapixels, combined with an 850nm ring LED light source, constitutes a surface quality detection unit; the acoustic wave sensor has a frequency response range of 100Hz-20kHz and is arranged at the four corners of the processing platform.

[0052] Process parameters: Impact strengthening of TC4 titanium alloy blades, laser energy 12J, spot size 3mm, water film flow rate 0.9L / min, acquisition of corresponding acoustic wave characteristic spectrum, and reference image template of surface aluminum foil.

[0053] After the enhanced processing is initiated, the control system acquires acoustic signals and surface images in real time. After passing through the signal preprocessing unit and the decision control unit, the acoustic signal undergoes frequency domain analysis, showing a main frequency shift to 3kHz. Based on the surface image, a 2mm × 3mm hole is identified in the absorbing layer aluminum foil. Specifically, the signal preprocessing unit uses a 100Hz-10kHz bandpass filter on the acoustic signal and a Gaussian filter (σ=1.2) for noise reduction on the surface image, generating a multi-dimensional feature matrix. The decision control unit performs frequency domain analysis, showing a main frequency shift from the standard 5kHz to 3kHz. The similarity calculation result is 0.4, below the anomaly threshold of 0.6. The surface quality evaluation algorithm, using the Canny operator edge detection, identifies a 2mm × 3mm hole in the absorbing layer aluminum foil. The quality score Q_score = 100 - (0.6 × 600 / 10) = 64 points, below the safety threshold of 94 points.

[0054] In the control system, the decision control unit completes the multi-source data fusion analysis within 200ms and triggers a three-level response in the execution output unit: immediately interrupting laser emission, cutting off laser output, controlling the robotic arm to move the workpiece to a safe position and locking the robotic arm movement; activating the audible and visual alarm device with alarm code E03; and controlling the robotic arm to transfer the workpiece to the rework station.

[0055] Record abnormal data, including the coordinates of the damage location (15.2, 8.7) mm and the damage area of ​​6 mm², and feed the abnormal data back to the process parameter reference database for recording; Maintenance personnel discovered that the aluminum foil in the absorbent layer was damaged. After cleaning and reapplying new aluminum foil, the system returned to normal.

[0056] On the other hand, in one specific embodiment, laser shock peening of aero-engine blades is performed for quality monitoring and energy fluctuation processing. The processing apparatus is the same as in the specific embodiment described above.

[0057] Process parameters: Impact strengthening of TC17 titanium alloy blades, using laser energy of 25J, spot size of 5mm, water film flow rate of 1.5L / min, and collecting corresponding acoustic characteristic spectrum and laser energy.

[0058] After the enhanced processing is started, the control system acquires the acoustic signal and the current laser energy in real time. After passing through the signal preprocessing unit and the decision control unit, the decision control unit uses the laser energy meter to identify that the laser energy has decreased from the set value of 25J to 18J. According to ΔE=α×(P^β×f^γ-P_target^β×f_target^γ), the energy deviation is calculated to be -28%, which exceeds the ±20% threshold. The frequency domain analysis of the acoustic signal shows that the main frequency has shifted from the standard 4kHz to 7kHz, with a similarity of 0.5, which is judged as an abnormal state.

[0059] The decision control unit completes multi-source data fusion analysis within 200ms, triggers a three-level response in the execution output unit, and performs comprehensive compensation adjustment. Using the formula P_newtarget=P_current×(1+K1×ΔE+K2×ΔQ+K3×ΔS), where ΔE=-0.28, ΔQ=0.05, ΔS=-0.3, the target laser power P_newtarget=P_current×(1+0.5×(-0.28)+0.3×0.05+0.2×(-0.3))=0.875×P_current, the laser power is first automatically adjusted to increase the power to 1.14 times the original set value. If the laser energy sensor detects that the energy is still 18J after adjustment, which does not reach the set value, the laser will stop emitting light, the movement of the robotic arm will be locked, the audible and visual alarm device will be activated with alarm code E05, and the robotic arm will be controlled to transfer the workpiece to the installation station.

[0060] Maintenance personnel discovered that the laser focusing lens was cracked, causing the laser to be unable to focus. After replacing the focusing lens, the system returned to normal.

[0061] On the other hand, in one specific embodiment, laser shock peening of aero-engine blades is monitored for quality, and water film flow rate mismatch is handled. The processing apparatus is the same as in the above specific embodiment.

[0062] Process parameters: Impact strengthening of TC17 titanium alloy blades, using laser energy of 20J, spot size of 4mm, water film flow rate of 1.2L / min, and collecting corresponding flow signals and laser energy.

[0063] After the enhanced processing is started, the control system acquires acoustic signals and current laser energy in real time. After passing through the signal preprocessing unit and the decision control unit, the flow sensor detects that the actual water film flow rate is 0.9 L / min during the processing. The decision control unit calculates that the flow deviation is -25%, which exceeds the 10% threshold.

[0064] The output unit implements PID control response, using a PID control loop u(t)=Kp×e(t)+Ki×∫e(t)dt+Kd×de(t) / dt to automatically adjust the water supply system. The proportional term is Kp×(-0.3), the integral term is Ki×∫(-0.3)dt, and the derivative term is Kd×d(-0.3) / dt. After PID adjustment, the valve opening of the water supply system increases by 15%, the flow rate recovers to 1.18 L / min, the deviation is controlled within 2%, and the system continues normal operation.

[0065] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0066] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-sensor fusion method for monitoring the quality of laser shock enhancement, characterized in that, Includes the following steps: Step 1: Collect laser energy spectrum characteristics, flow matching curves, acoustic signal feature templates and standard surface quality images under different materials and process conditions; set the set values ​​for laser energy, water film flow rate and acoustic spectrum; and construct a process parameter benchmark database in combination with the standard surface quality images. Step 2: Acquire multi-source signals during the processing; Step 3: Preprocess the multi-source signals to generate a multi-dimensional feature matrix; Step 4: Conduct a comprehensive analysis based on the multidimensional feature matrix and the process parameter benchmark database to obtain quality monitoring results.

2. The multi-sensor fusion laser shock enhancement quality monitoring method according to claim 1, characterized in that, The multi-source signals include laser energy signals, flow signals, surface images, and acoustic signals.

3. The multi-sensor fusion laser shock enhancement quality monitoring method according to claim 1, characterized in that, The preprocessing first employs a hierarchical denoising and multidimensional normalization strategy to achieve precise time synchronization and spatial registration, generating a multidimensional feature matrix. The hierarchical denoising and multidimensional normalization strategy includes Kalman filtering, moving average filtering, median filtering, Gaussian filtering, bandpass filtering, and normalization.

4. The laser shock enhancement quality monitoring method based on multi-sensor fusion according to claim 2, characterized in that, Based on the laser energy signal, flow rate signal, and acoustic signal in the multidimensional feature matrix, and combined with the set values ​​of laser energy, water film flow rate, and acoustic spectrum in the process parameter benchmark database, the processing quality is calculated, including laser quality, flow rate quality, and noise level. Surface images and standard surface quality images are used to detect edge detection using the Canny operator and region growing algorithm to identify defect regions in the surface images and calculate the geometric feature parameters of these defect regions. The workpiece surface quality score is then calculated based on these geometric feature parameters. The processing quality and the workpiece surface quality score together constitute the quality monitoring results.

5. The laser shock enhancement quality monitoring method based on multi-sensor fusion according to claim 4, characterized in that, Based on the collected laser energy signal and the laser energy setpoint, the energy deviation is calculated. When the energy deviation is less than the lower energy threshold, the laser quality is slightly deviated. When the energy deviation is within the range of the lower energy threshold and the upper energy threshold, the laser quality is moderately deviated. When the energy deviation is greater than the upper energy threshold, the laser quality is severely deviated. The flow deviation is calculated based on the collected flow signal and the water film flow setpoint. When the flow deviation is less than the lower limit threshold, the flow quality is slightly deviated. When the flow deviation is within the range of the lower limit threshold and the upper limit threshold, the flow quality is moderately deviated. When the flow deviation is greater than the upper limit threshold, the flow quality is severely deviated. Multi-domain feature extraction technology is used to extract the spectral features of acoustic signals. The system status is determined based on the extracted spectral features and the spectral grading criteria constructed based on the acoustic spectral setpoints. The spectral grading criteria are as follows: if the similarity between the spectral features and the acoustic spectral setpoints is greater than the upper limit threshold, the noise quality is normal; if the similarity between the spectral features and the acoustic spectral setpoints is within the range of the lower limit threshold and the upper limit threshold, the noise quality is in warning mode; and if the similarity between the spectral features and the acoustic spectral setpoints is less than the lower limit threshold, the noise quality is abnormal.

6. The laser shock enhancement quality monitoring method based on multi-sensor fusion according to claim 5, characterized in that, The process also includes step 5, which generates a response action based on the quality monitoring results and a graded response strategy. The graded response strategy includes: when the laser quality and flow rate quality have slight deviations and the noise level is at the warning level, a level 1 response is executed to adjust the signal; when the laser quality and flow rate quality have moderate deviations and the noise level is at the warning level, a level 2 response is executed to suspend the alarm and initiate manual intervention; when the laser quality and flow rate quality have severe deviations, the noise level is abnormal, and the workpiece surface quality score is lower than the set surface threshold, a level 3 response is executed to shut down the machine immediately and simultaneously feed the adjustment signal back to the process parameter benchmark database for recording.

7. The laser shock enhancement quality monitoring method based on multi-sensor fusion according to claim 6, characterized in that, An improved fuzzy PID control is used to adjust the laser energy signal, expressed as P_newtarget=P_current×(1+K1×ΔE+K2×ΔQ+K3×ΔS), where ΔE is the energy deviation, ΔQ is the flow deviation, ΔS is the acoustic signal deviation, K1=0.5, K2=0.3, K3=0.2, P_newtarget represents the target laser power, and P_current represents the current laser power. The flow signal is adjusted using a PID control loop, expressed as u(t)=Kp×e(t)+Ki×∫e(t)dt+Kd×de(t) / dt, where u(t) represents the target flow rate, e(t) represents the flow deviation, Kp represents the proportional gain, Ki represents the integral gain, and Kd represents the derivative gain.

8. A multi-sensor fusion laser shock-enhanced quality monitoring device, characterized in that, The laser shock enhancement quality monitoring method according to any one of claims 1-7 includes: a multi-sensor fusion module and an intelligent control system; Multi-sensor array module, acquiring signals from multiple sources; The intelligent control system includes a database, a signal preprocessing unit, and a decision control unit. The database stores a baseline database of process parameters. The signal preprocessing unit preprocesses multi-source signals to generate a multi-dimensional feature matrix. The decision control unit performs comprehensive analysis based on the multi-dimensional feature matrix and the baseline database of process parameters to obtain quality monitoring results.

9. A multi-sensor fusion laser shock enhancement quality monitoring device according to claim 8, characterized in that, The multi-sensor array module includes a laser energy sensor, a flow rate sensor, a surface quality sensor, and an acoustic wave sensor. The laser energy sensor is installed inside the laser to collect the laser energy signal during the laser shock strengthening process, including laser power and pulse frequency. The flow rate sensor collects the flow rate signal of the water film during the laser shock strengthening process. The surface quality sensor collects images of the workpiece surface. The acoustic wave sensor collects the acoustic signal of the laser shock strengthening process.

10. A multi-sensor fusion laser shock enhancement quality monitoring device according to claim 8, characterized in that, The intelligent control system also includes an execution output unit, which generates response actions based on the quality monitoring results and a graded response strategy, and uses a PID control algorithm to adjust the signal execution response.

Images