A laser welding repair device for surface defects of a casting

Abstract

The application belongs to the technical field of laser welding, and particularly relates to a cast surface defect laser welding repair device, which comprises a welding device body and further comprises: a laser welding module comprising a welding gun, a gas nozzle, a wire feeding mechanism and a coaxial vision assembly; a multi-source sensing module comprising a dew point sensor arranged in a gas circuit, a flow rate probe arranged downstream of an outlet of the gas nozzle, a three-axis accelerometer arranged at a head of the welding gun, an oxygen sensor for measuring oxygen concentration at an edge of a molten pool and a laser range finder; and a control module in communication connection with the laser welding module and the multi-source sensing module, the control module being used for performing the following operations: acquiring dynamic oxygen partial pressure and a protective gas dew point temperature to construct an atmosphere cleanliness model and acquire an atmosphere cleanliness coefficient; welding quality out of control caused by extreme deviation or environmental fluctuation is effectively avoided, laser energy density is ensured to be always in an optimal state, and the precision and consistency of the repair process are significantly improved.

Description

Technical Field

[0001] This invention belongs to the field of laser welding technology, specifically a laser welding repair device for surface defects in castings. Background Technology

[0002] Luminescent materials are widely used in cutting-edge fields such as optoelectronic devices and biomedicine. As a core parameter for evaluating the luminescence performance of such materials, the accuracy of its testing is crucial. With the continuous emergence of new broadband excitation materials, traditional testing equipment has gradually revealed its limitations.

[0003] Existing absolute fluorescence quantum yield testing systems mostly adopt an integrating sphere combined with a high-sensitivity spectrometer as their basic architecture. In terms of light source and optical path design, they are usually configured with a single-wavelength laser or xenon lamp combined with a monochromator as the excitation source, and rely on a fixed focal length lens group to collimate and converge the beam. The excitation light is guided to the sample to be tested inside the integrating sphere through a fixed optical path. Then the system collects the scattered light and emitted fluorescence of the sample, and completes the calculation of quantum yield by spectral comparison.

[0004] However, traditional light source systems have limited excitation testing range and cumbersome wavelength switching, making it impossible to efficiently cover the wide excitation band requirements of new materials. On the other hand, the fixed focal length optical path of existing systems lacks dynamic focusing capability. When changing to samples of different forms (such as liquid cuvettes, solid films or powders) or different thicknesses, the focal position cannot be precisely adjusted, which will cause the excitation spot to diverge on the sample surface, the excitation energy density to be uneven, or even some beams to leak out of the sample, resulting in low fluorescence signal collection efficiency and seriously affecting the absolute accuracy of quantum yield measurement.

[0005] Therefore, the present invention provides a laser welding repair device for surface defects of castings. Summary of the Invention

[0006] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0007] The technical solution adopted by this invention to solve its technical problem is: a laser welding repair device for surface defects of castings, comprising a welding device body, and further comprising: The laser welding module includes a welding torch, a gas nozzle, a wire feeding mechanism, and a coaxial vision assembly. The multi-source sensing module includes a dew point sensor in the gas path, a flow rate probe downstream of the gas nozzle outlet, a triaxial accelerometer in the welding torch head, an oxygen sensor for measuring the oxygen concentration at the edge of the molten pool, and a laser rangefinder. The control module is communicatively connected to both the laser welding module and the multi-source sensing module. The control module is used to perform the following operations: Dynamic oxygen partial pressure and protective gas dew point temperature are obtained to construct an atmosphere cleanliness model and obtain the atmosphere cleanliness coefficient; The turbulence intensity and acceleration spectral kurtosis at the nozzle outlet are obtained to construct an airflow-mechanical stability model and obtain the airflow-mechanical stability coefficient. The two-dimensional offset between the welding wire and the spot and the radius of the fillet at the root of the groove are obtained to construct a repair posture compliance model and obtain the repair posture compliance coefficient. Based on the atmosphere cleanliness coefficient, airflow-mechanical stability coefficient, and real-time attitude fluctuation, a multi-source parameter collaborative stable attitude model is constructed to obtain the stable attitude coefficient. Based on the stable attitude coefficient, the repair posture compliance coefficient, and the current focus deviation, a relative position deviation optimization model is constructed to output the target focus deviation and drive the laser welding module to adjust the focus position.

[0008] The beneficial effects of this invention are as follows: 1. The laser welding repair device for surface defects of castings described in this invention dynamically calculates and outputs the target focus deviation based on the steady-state coefficient, the repair posture compliance coefficient, and the current focus deviation. Under complex and variable welding conditions, it can intelligently and smoothly transition between accurately tracking the real-time focus deviation and returning to the safe focus offset reference amount, effectively avoiding the loss of control over welding quality caused by extreme deviations or environmental fluctuations, ensuring that the laser energy density is always in the optimal state, and significantly improving the accuracy and consistency of the repair process.

[0009] 2. The laser welding repair device for casting surface defects described in this invention utilizes oxygen sensors and dew point sensors to obtain dynamic oxygen partial pressure and protective gas dew point temperature. These are then mapped to a single atmosphere cleanliness coefficient using a model to accurately determine the anti-oxidation effect of the protective gas. Simultaneously, by combining the turbulence intensity at the nozzle outlet obtained from the flow velocity probe with the vibration spectrum kurtosis extracted from the triaxial accelerometer, the airflow-mechanical stability coefficient is quantified. This eliminates the limitations of single-sensor monitoring or experience-based judgment, enabling sensitive detection of airflow turbulence and mechanical vibration risks, providing highly reliable environmental and physical condition guarantees for high-quality welding repair.

[0010] 3. The laser welding repair device for surface defects of castings described in this invention extracts the two-dimensional offset of the welding wire end and the actual error of the fillet radius at the root of the bevel, and introduces a high-order penalty factor to amplify the negative impact of severe deviations. This allows for a direct reflection of the degree to which the current repair posture deviates from the ideal state. On this basis, the system further integrates the posture state with environmental parameters to construct a multi-source parameter collaborative stable attitude model. This dynamically captures the real destructive effect of nozzle distance fluctuations and axis angles on the protective airflow coverage efficiency, overcoming the shortcomings of traditional solutions in sensing geometric features and dynamic disturbances, and ensuring a high degree of controllability of the repair operation under complex travel paths. Attached Figure Description

[0011] The invention will now be further described with reference to the accompanying drawings.

[0012] Figure 1 This is a perspective view of the present invention; Figure 2 In this invention Figure 1 Enlarged view of point A in the image; Figure 3 This is the control flowchart of the present invention; Figure 4 This is a flowchart of the atmosphere cleanliness model, airflow mechanical stability model, and repair posture compliance model in this invention; Figure 5 This is a flowchart of the multi-source parameter collaborative steady-state model and the relative position deviation optimization model in this invention.

[0013] In the diagram: 1. Welding device body; 2. Wire feeding mechanism; 3. Welding torch; 4. Gas nozzle. Detailed Implementation

[0014] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0015] The laser welding repair device for surface defects of castings according to an embodiment of the present invention includes a welding device body 1, and further includes: The laser welding module includes a welding torch 3, a gas nozzle 4, a wire feeding mechanism 2, and a coaxial vision assembly; The multi-source sensing module includes a dew point sensor in the gas path, a flow rate probe downstream of the outlet of the gas nozzle 4, a triaxial accelerometer at the head of the welding torch 3, an oxygen sensor for measuring the oxygen concentration at the edge of the molten pool, and a laser rangefinder. The control module is communicatively connected to both the laser welding module and the multi-source sensing module. The control module is used to perform the following operations: Dynamic oxygen partial pressure and protective gas dew point temperature are obtained to construct an atmosphere cleanliness model and obtain the atmosphere cleanliness coefficient; The turbulence intensity and acceleration spectral kurtosis at the outlet of nozzle 4 were obtained to construct an airflow-mechanical stability model and obtain the airflow-mechanical stability coefficient. The two-dimensional offset between the welding wire and the spot and the radius of the fillet at the root of the groove are obtained to construct a repair posture compliance model and obtain the repair posture compliance coefficient. A multi-source parameter collaborative attitude stability model is constructed based on the atmosphere cleanliness coefficient, airflow-mechanical stability coefficient, and real-time attitude fluctuation to obtain the attitude stability coefficient. Based on the stable attitude coefficient, the corrected pose compliance coefficient, and the current focus deviation, a relative position deviation optimization model is constructed to output the target focus deviation and drive the laser welding module to adjust the focus position.

[0016] The laser welding module can be configured in various forms. For example, the welding torch 3 can be a fiber optic laser transmission type, and its output port is designed as a replaceable gas nozzle 4 to adapt to different welding needs. The shielding gas nozzle 4 can adopt a simple cylindrical structure and be protected by a fixed flow rate of inert gas (such as argon). The wire feeding mechanism 2 can adopt a stepper motor driven roller wire feeder, and the wire feeding speed can be manually set. The coaxial vision component can adopt a simple industrial camera, and its image is transmitted to a display for the operator to observe. The operator may need to manually adjust the wire feeding speed and shielding gas flow rate based on experience and judge the welding quality based on visual feedback.

[0017] Multi-source sensing modules are used to acquire key data during the welding process in real time. For example, a dew point sensor can be a humidity sensor based on capacitance or resistance principles, directly installed in the gas path at the outlet of the protective gas cylinder to measure the humidity of the gas and convert it to dew point temperature through a lookup table or simple formula; a flow rate probe can be a hot-wire or Pitot tube sensor placed at a fixed position downstream of the gas nozzle 4 outlet to measure the airflow velocity, and its output signal is simply converted into a flow rate value; a triaxial accelerometer can be a MEMS (Micro-Electro-Mechanical Systems) sensor fixed on the head shell of the welding torch 3 to measure the vibration acceleration of the welding torch 3 in three orthogonal directions, and its raw data is directly recorded; an oxygen sensor can be a zirconia sensor with its probe placed near the edge of the molten pool to measure the local oxygen concentration; a laser rangefinder can be a sensor based on triangulation or time-of-flight principles to measure the distance between the gas nozzle 4 of the welding torch 3 and the workpiece surface. The data from these sensors can be independently acquired and displayed for operator reference.

[0018] In addition, the control module communicates with the aforementioned laser welding module and multi-source sensing module and performs a series of operations to achieve intelligent control. For example, when acquiring dynamic oxygen partial pressure and protective gas dew point temperature to construct an atmosphere cleanliness model and obtain the atmosphere cleanliness coefficient, the control module can simply compare the oxygen partial pressure and dew point temperature with preset thresholds. If any parameter exceeds the threshold, the atmosphere cleanliness coefficient is set to a lower value; otherwise, it is set to a higher value.

[0019] When acquiring the turbulence intensity and acceleration spectral kurtosis at the outlet of nozzle 4 to construct the airflow-mechanical stability model and obtain the airflow-mechanical stability coefficient, the control module can independently analyze the data from the velocity probe and the triaxial accelerometer. For example, when the velocity fluctuation exceeds a certain fixed percentage, the airflow is considered unstable; when the root mean square value of the acceleration exceeds a certain fixed threshold, the mechanical vibration is considered unstable. Then, the airflow-mechanical stability coefficient is determined through simple logical judgments (such as "AND" and "OR" relationships).

[0020] When acquiring the two-dimensional offset of the welding wire and the spot, as well as the radius of the bevel root fillet, to construct a repair posture compliance model and obtain the repair posture compliance coefficient, the control module can use the image acquired by the coaxial vision component to identify the center of the spot and the end of the welding wire through image processing algorithms, and calculate their pixel offset. Then, it can convert the offset into the actual offset through a preset scaling factor. The radius of the bevel root fillet can be obtained by manual measurement before welding or by acquiring contour data through a simple two-dimensional contour scanner and performing basic geometric fitting. The repair posture compliance coefficient can be compared with these measured values ​​and a preset fixed tolerance range. If it is within the range, it is considered compliant; otherwise, it is not compliant. This method may not accurately reflect the nonlinear relationship between the influence of different offsets and radii on welding quality.

[0021] When constructing a multi-source parameter collaborative stability model based on the obtained atmosphere cleanliness coefficient, airflow-mechanical stability coefficient, and real-time attitude fluctuation, the control module can use a simple weighted average method to obtain the stability coefficient. For example, the atmosphere cleanliness coefficient and airflow-mechanical stability coefficient can be multiplied by their respective fixed weights, and a simple penalty term calculated based on the distance fluctuation between the air nozzle 4 and the workpiece and the angle fluctuation of the air nozzle 4 measured by the laser rangefinder can be added to obtain the stability coefficient. However, this method may not fully reflect the nonlinear synergistic effect between the parameters and the complex influence on the overall stability.

[0022] Finally, based on the obtained stability coefficient, repair posture compliance coefficient, and current focus deviation, a relative position deviation optimization model is constructed to output the target focus deviation. When driving the laser welding module to adjust the focus position, the control module can adopt a linear adjustment strategy. For example, the product of the stability coefficient and the repair posture compliance coefficient is multiplied by a fixed scaling factor and then added to the current focus deviation to obtain the target focus deviation. Subsequently, the focus position is adjusted to the target position by controlling the focusing lens group drive mechanism inside the laser welding module. This linear adjustment method may not achieve optimal focus position control in complex and variable welding environments, especially under extremely unstable conditions.

[0023] Before welding begins, the multi-source sensing module starts working. The dew point sensor monitors the dew point temperature of the protective gas (e.g., argon) in the gas path entering the gas nozzle 4 in real time. At the same time, the oxygen sensor obtains the dynamic oxygen partial pressure at the edge of the molten pool in real time through its suction capillary extending to the protective gas-covered area at the edge of the molten pool. This data is transmitted to the control module.

[0024] After receiving the dynamic oxygen partial pressure and the protective gas dew point temperature, the control module inputs them into the atmosphere cleanliness model. Based on the preset mathematical relationship, it comprehensively evaluates the cleanliness of the current protective atmosphere and outputs an atmosphere cleanliness coefficient. For example, if the oxygen partial pressure is detected to be slightly higher than the normal value, or the dew point temperature is too high, the atmosphere cleanliness coefficient will be calculated as a relatively low value, indicating that there is a certain degree of pollution risk in the protective atmosphere.

[0025] Meanwhile, the flow velocity probe continuously monitors the airflow state at the outlet of the gas nozzle 4 and obtains the turbulence intensity at the outlet of the gas nozzle 4. The triaxial accelerometer at the head of the welding torch 3 collects the vibration data of the welding torch 3 in real time. The control module processes this data and obtains the kurtosis of the acceleration spectrum. These parameters are input into the airflow-mechanical stability model, which comprehensively evaluates the stability of the airflow and the mechanical vibration of the welding torch 3 and outputs an airflow-mechanical stability coefficient. For example, if an increase in the turbulence intensity of the airflow is detected, or if the vibration of the welding torch 3 intensifies, the airflow-mechanical stability coefficient will be calculated to a lower value, indicating that the mechanical stability of the welding process is affected.

[0026] During the welding process, the coaxial vision component captures the relative position of the welding wire tip and the laser spot in real time. The control module calculates the two-dimensional offset between the welding wire and the laser spot. Simultaneously, before welding, the line laser scanner scans along the bevel to obtain the cross-sectional contour point cloud. The control module extracts the radius of the fillet at the root of the bevel through an arc fitting algorithm. These geometric parameters are input into the repair posture compliance model, which evaluates the accuracy of the welding wire feed and the suitability of the bevel geometry, and outputs a repair posture compliance coefficient. For example, if there is a slight offset between the welding wire and the laser spot, or if the radius of the fillet at the root of the bevel deviates from the ideal value, the repair posture compliance coefficient will be calculated as a lower value.

[0027] Subsequently, the control module inputs the obtained atmosphere purity coefficient and airflow-mechanical stability coefficient, as well as the instantaneous fluctuation value of the distance between the gas nozzle 4 and the workpiece surface measured in real time by the laser rangefinder and the angle between the axis of the gas nozzle 4 and the tangent of the instantaneous welding direction (i.e., real-time attitude fluctuation) into the multi-source parameter collaborative stability model. This model performs collaborative analysis on these multi-source parameters, comprehensively evaluates the overall stability of the welding process, and outputs a stability coefficient. For example, if the atmosphere purity, airflow-mechanical stability, and attitude fluctuation are all in a poor state, the stability coefficient will be significantly reduced, indicating that the welding process faces high uncertainty.

[0028] Finally, the control module inputs the stability coefficient, the repair posture compliance coefficient, and the instantaneous relative position deviation between the laser beam focus and the workpiece surface measured by the current laser rangefinder into the relative position deviation optimization model. Based on these inputs, the model dynamically calculates a target focus deviation. For example, if both the stability coefficient and the repair posture compliance coefficient are low, it indicates that the welding environment and posture are not ideal. The model may output a more conservative target focus deviation to ensure the stability of the molten pool and the welding quality. The control module then drives the focusing lens group inside the laser welding module to precisely adjust the laser focus to the target focus deviation position. Through this dynamic and real-time focus adjustment, the stability of the molten pool and the welding quality can be effectively maintained even in the case of fluctuating ambient atmosphere, unstable airflow, or deviation in repair posture.

[0029] By introducing a multi-source sensing module, comprehensive real-time monitoring of the welding environment, mechanical condition, and repair posture is achieved. This contrasts with the single or limited sensor application in traditional solutions, enabling precise quantification of atmosphere cleanliness rather than simply relying on experience to determine whether the shielding gas is sufficient. The flow velocity probe and triaxial accelerometer provide quantitative indicators of airflow and mechanical vibration, overcoming the limitations of insufficient perception of these disturbances in traditional solutions.

[0030] Furthermore, models including atmosphere cleanliness model, airflow-mechanical stability model, repair posture compliance model, multi-source parameter collaborative stability model, and relative position deviation optimization model transform multi-source sensor data into quantifiable coefficients and perform collaborative analysis, thereby achieving a comprehensive assessment of the overall stability of the welding process. Compared with the control methods based on single parameter thresholds or simple logic judgments in traditional solutions, these models can more accurately reflect the requirements of the complex and ever-changing welding environment for the focal position.

[0031] The atmosphere cleanliness model is constructed in the following way: extract the dimensionless deviation square term of dynamic oxygen partial pressure relative to the reference safety threshold, and the dimensionless deviation square term of real-time gas dew point temperature relative to the process allowable boundary. The two dimensionless deviation square terms are weighted and summed by weight adjustment coefficients and used as the variable term in the denominator. This is used to construct a rational algebraic decay function, thereby converging the atmosphere cleanliness coefficient to the dimensionless preset interval.

[0032] The atmosphere cleanliness model is specifically represented as follows:

[0033] In the formula, The atmospheric cleanliness coefficient has dimensions and units of measurement. ; This refers to the dynamic oxygen partial pressure, with units of measurement of _____. ; For reference to the safe oxygen partial pressure threshold, the unit of measurement is: ; This refers to the real-time gas dew point temperature, with units of measurement of _____. ; and These are the minimum and maximum dew point temperature boundary values ​​allowed by the process, with units of measurement of [missing information]. ; and These are the oxygen partial pressure and water weighting adjustment coefficients, respectively, with units of _____. .

[0034] Among them, the atmosphere cleanliness model aims to quantify the purity of the shielding gas in the welding environment. By comprehensively considering two core indicators, oxygen partial pressure and dew point temperature, it transforms complex environmental factors into a single, operable cleanliness coefficient. Atmosphere cleanliness coefficient As a unit of measurement The dimensionless parameter of the shielding gas directly reflects the shielding effect of the shielding gas on the molten pool. The higher the value, the better the shielding effect.

[0035] Dynamic oxygen partial pressure This refers to the local pressure of oxygen in the shielding gas, measured in real time during the welding process. Its level directly affects the oxidation risk of the molten pool. This dynamic oxygen partial pressure... It can be acquired in real time by an oxygen sensor in a multi-source sensing module, for example, by an electrochemical oxygen sensor or a zirconia oxygen sensor.

[0036] Reference safe oxygen partial pressure threshold It is a preset upper limit of oxygen partial pressure, considered safe under specific welding processes, used to define the acceptable range of oxygen content in the shielding gas, and the real-time gas dew point temperature. This refers to the temperature at which water vapor in the shielding gas condenses into liquid water. It reflects the moisture content of the shielding gas, and moisture is another important factor leading to welding defects. This real-time gas dew point temperature... It can be acquired in real time by the dew point sensor in the multi-source sensing module, for example, by measuring with a cold mirror dew point meter or a capacitive dew point sensor.

[0037] The boundary value between the minimum and maximum dew point temperatures allowed by the process. and This defines the ideal operating range of the shielding gas dew point temperature under a specific welding process, and the oxygen partial pressure and moisture weighting adjustment coefficients. and These are parameters used to adjust the influence of oxygen partial pressure and dew point temperature on the atmosphere cleanliness coefficient. They allow for flexible configuration based on the differences in oxygen and moisture sensitivity of different materials or processes. For example, for easily oxidized materials, the oxygen partial pressure and dew point temperature can be appropriately increased. The weight.

[0038] By constructing an atmosphere cleanliness model, key chemical parameters in the welding environment are transformed into quantified atmosphere cleanliness coefficients, and dynamic oxygen partial pressure is introduced. Compared with the reference safe oxygen partial pressure threshold The ratio of oxygen to molten pool oxidation can effectively assess the impact of oxygen on the degree of oxidation. The introduction of a squared term enhances the sensitivity to deviations of oxygen partial pressure from the safe threshold. Furthermore, the model incorporates real-time gas dew point temperature. Boundary values ​​between the minimum and maximum dew point temperatures allowed by the process. , The negative impact of moisture content on the quality of protective gas was quantified through normalization.

[0039] By setting the oxygen partial pressure weight adjustment coefficient With moisture weighting adjustment coefficient This model can be flexibly configured according to the sensitivity of different materials or processes to oxygen and moisture, thus adjusting the atmosphere cleanliness coefficient. The calculations are more in line with actual welding requirements.

[0040] By mapping complex environmental parameters to a single cleanliness coefficient, real-time and dynamic evaluation of welding atmosphere quality is achieved, enabling the control module to more accurately determine whether the current atmosphere meets the cleanliness requirements for high-quality repair, thereby overcoming the limitation that single sensor data cannot accurately quantify the protective effect of shielding gas.

[0041] This model effectively solves the problem that traditional methods cannot accurately quantify the protective effect of shielding gas using data from a single sensor. It comprehensively considers the influence of oxygen partial pressure and dew point temperature on welding quality, enabling the control module to more accurately determine whether the current atmosphere meets the cleanliness requirements for high-quality repair. This provides a more reliable environmental quality guarantee for the laser welding repair process and significantly reduces the risk of welding defects caused by poor shielding gas quality.

[0042] The airflow-mechanical stability model is constructed as follows: the dimensionless deviation cubic term of the turbulence intensity at the outlet of the real-time nozzle 4 relative to the critical intensity of turbulence abrupt change, and the dimensionless absolute deviation cubic term of the acceleration spectrum kurtosis relative to the allowable maximum value are extracted. The two dimensionless deviation cubic terms are then weighted and summed by sensitivity adjustment gain and used as independent variables. In this way, an inverse triangular mapping function is constructed, thereby outputting the dimensionless airflow-mechanical stability coefficient.

[0043] The airflow-mechanical stability model is specifically represented as follows:

[0044] In the formula, The airflow-mechanical stability coefficient, with dimensions [missing information]. ; The real-time turbulence intensity at the outlet of nozzle 4 is given by units of . ; The reference turbulence intensity at the critical point where laminar flow abruptly transitions to turbulence, with dimensions in units of . ; The kurtosis of the triaxial vibration acceleration spectrum, with units of . ; The maximum permissible kurtosis deviation is expressed in units of . ; and These represent the sensitivity adjustment gain, with units of . .

[0045] This airflow-mechanical stability model aims to comprehensively evaluate the impact of the stability of the protective airflow and the mechanical vibration of the welding torch body on the welding quality during laser welding. It is used to quantify and weight the airflow state and mechanical vibration state to provide a unified airflow-mechanical stability coefficient to reflect the dynamic stability of the welding environment.

[0046] airflow-mechanical stability coefficient It is a dimensionless index used to quantify the airflow and mechanical stability during the welding process. Its value range is usually between 0 and 1, where 1 represents an ideal stable state, while values ​​close to 0 indicate poor stability. It can guide the control module to perform precise optimization of the focal position.

[0047] Real-time gas nozzle 4 outlet turbulence intensity It is a dimensionless parameter that measures the degree of turbulence of the protective airflow at the outlet of the air nozzle 4. It can be obtained in a variety of ways, such as by directly measuring the airflow velocity pulsation at the outlet of the air nozzle 4 using a hot-wire anemometer or a miniature Pitot tube array, and calculating the ratio of its root mean square value to the average velocity.

[0048] Reference turbulence intensity at the critical point of abrupt change from laminar to turbulent flow This is a preset threshold value, representing the critical point at which the protective airflow transitions from laminar to turbulent flow. This value can be determined experimentally, for example, by gradually increasing the airflow velocity in a controlled environment while monitoring the airflow state. The turbulence intensity corresponding to when the airflow begins to exhibit significant disturbance is this threshold value. .

[0049] Triaxial vibration acceleration spectrum kurtosis This is a statistical measure of the impact or non-Gaussianity of the vibration signal from the welding torch head. It can be obtained by processing the raw acceleration time series acquired by a triaxial accelerometer. First, a Fast Fourier Transform (FFT) is performed on the acceleration time series to obtain the spectrum. Then, the ratio of the fourth-order central moment to the square of the variance of the amplitude sequence of the spectrum is calculated, which is the kurtosis. The higher the kurtosis value, the more impact components are contained in the vibration signal, and the more unstable the vibration is. The maximum allowable kurtosis deviation is... It is a preset threshold that represents the maximum permissible deviation of the welding torch vibration kurtosis from the ideal value (e.g., the kurtosis of a Gaussian distribution is 3) within an acceptable range of welding quality. This value can be determined through process experiments, i.e., welding under different vibration conditions and evaluating the weld quality, thereby determining an upper limit of kurtosis deviation corresponding to good weld quality.

[0050] Sensitivity adjustment gain and These are dimensionless weighting coefficients used to adjust the effects of airflow turbulence intensity and mechanical vibration kurtosis on the airflow-mechanical stability coefficient. The relative sensitivity of the impact can be assessed, and these gain values ​​can be calibrated and adjusted through expert experience, historical data analysis, or optimization algorithms (such as genetic algorithms and particle swarm optimization) to ensure that the model accurately reflects the actual stability under different welding processes and material conditions. For example, in processes with higher requirements for airflow stability, the gain values ​​can be appropriately increased. The value of .

[0051] The airflow-mechanical stability model proposed in this application uses the real-time turbulence intensity at the outlet of the air nozzle 4. With triaxial vibration acceleration spectrum kurtosis By organically combining these two key parameters, a comprehensive quantitative assessment of the dynamic stability of the welding process is achieved. Specifically, the control module acquires the real-time turbulence intensity at the outlet of the gas nozzle 4 through the flow velocity probe in the multi-source sensing module. And compare it with the preset reference turbulence intensity at the critical point of transition from laminar to turbulent flow. A comparison is made to assess the degree of turbulence in the protective airflow. Simultaneously, the control module receives vibration data from the triaxial accelerometer at the welding torch head 3 and calculates the kurtosis of the triaxial vibration acceleration spectrum. This kurtosis value is related to the maximum permissible kurtosis deviation extreme value. By making comparisons, the impact of the mechanical vibration of the welding torch 3 body on welding stability can be quantified.

[0052] The model adjusts the gain of these two ratio terms through sensitivity. and Weighting is applied, and a cubic term is used to amplify the negative impact on stability, making the model more sensitive to airflow turbulence and mechanical vibration. Subsequently, the weighted results are normalized using the arctangent function, mapping the effects of airflow fluctuations and mechanical vibrations to a unified airflow-mechanical stability coefficient. Within the range, this design can not only smoothly handle extreme fluctuation data, but also differentiate the weight allocation of airflow and vibration effects under different working conditions by adjusting the gain, so as to ensure that the airflow-mechanical stability coefficient can truly reflect the comprehensive stability of the welding process.

[0053] The airflow-mechanical stability coefficient As an important input to the multi-source parameter collaborative stability model, it works together with the atmosphere cleanliness coefficient to provide a reliable decision basis for subsequent focus position optimization, thus effectively solving the problem that traditional single-dimensional monitoring methods cannot fully quantify the coupled influence of airflow fluctuations and mechanical vibrations on welding stability.

[0054] By introducing real-time turbulence intensity at the four outlets of the gas nozzle and kurtosis of the triaxial vibration acceleration spectrum and integrating them into a unified model, the problem that traditional single-dimensional monitoring methods cannot comprehensively assess the dynamic stability of the welding process is effectively solved. This model can sensitively capture the welding risks caused by airflow turbulence and mechanical vibration, providing more accurate and reliable input for subsequent focus position adjustment, thereby significantly improving the quality and stability of laser welding repair and reducing the defect rate.

[0055] The repair posture compliance model is constructed in the following way: extract the dimensionless deviation high-order term of the two-dimensional offset of the welding wire end relative to the process tolerance, and the dimensionless deviation high-order term of the actual error of the fillet radius at the root of the bevel relative to the allowable tolerance. The two terms are weighted and summed by a high-order penalty factor and superimposed with a basic constant as the proper term and the reciprocal is taken. In this way, a logarithmic decay mapping function is constructed to output the dimensionless repair posture compliance coefficient.

[0056] The specific representation of the repair pose compliance model is as follows:

[0057] In the formula, To correct the pose compliance coefficient, the unit of measurement is: ; This is the two-dimensional offset of the welding wire tip relative to the center of the laser spot, with units of . ; This represents the maximum permissible tolerance level of the two-dimensional offset, with units of . ; To correct the actual measured value of the fillet radius at the root of the bevel, the unit of measurement is: ; and These are the optimal theoretical root fillet radius and its allowable tolerance, respectively, with units of . ; and These are higher-order penalty factors, with units of measurement of [missing information]. .

[0058] This repair posture compliance model aims to quantify the degree of geometric matching between the welding wire, laser spot, and the bevel to be repaired during laser welding repair. Its core function is to integrate multiple key geometric deviation parameters into a single, quantifiable repair posture compliance coefficient. This coefficient can intuitively reflect the degree to which the current repair pose deviates from the ideal state. The model can be implemented in the form of mathematical formulas, lookup tables or regression models based on machine learning. Its output value is usually between 0 and 1. The larger the value, the higher the compliance, and vice versa.

[0059] Two-dimensional offset of the welding wire tip relative to the center of the laser spot This indicates the relative positional deviation between the end of the welding wire fed by the wire feeding mechanism 2 and the center of the laser spot emitted by the laser welding module on a two-dimensional plane. This offset is a key factor affecting the welding energy coupling efficiency and the stability of the molten pool. It can be obtained by capturing images of the end of the welding wire and the laser spot in real time through a coaxial vision component, calculating the pixel offset through an image processing algorithm, and then converting it into the actual physical size through calibration.

[0060] Maximum allowable tolerance level of two-dimensional offset This defines the maximum tolerable two-dimensional offset of the welding wire tip relative to the center of the laser spot under a specific welding process. Offsets exceeding this range may lead to welding defects, such as incomplete fusion, insufficient penetration, or poor weld formation. This value is usually determined by process experts through experiments or simulations based on factors such as material properties, welding power, and wire feed speed.

[0061] Actual measured value of the fillet radius at the root of the bevel. The radius of the bevel at the bottom of the casting defect to be repaired is the actual radius of the fillet. The geometry of the bevel root has a significant impact on the filling behavior of the molten pool and the final forming quality of the weld. It can be obtained by scanning the bevel with a line laser scanner, acquiring three-dimensional point cloud data, and then using a geometric fitting algorithm to extract the radius of the arc.

[0062] Optimal theoretical root fillet radius This represents the theoretical fillet radius that the root of the groove should have under ideal welding process conditions in order to obtain the best weld formation and mechanical properties. This value is usually set in advance according to welding specifications, material type and structural design requirements.

[0063] Allowable tolerance The maximum allowable deviation range of the actual measured value of the root fillet radius of the repair groove relative to the optimal theoretical root fillet radius is defined. Deviations exceeding this range may lead to poor flow of the molten pool, stress concentration, or a decrease in weld strength. This value is also determined by process experts based on experience or experimental data.

[0064] Higher-order penalty factors and Used to adjust the welding wire offset and bevel radius deviation to improve the compliance coefficient of the repair posture. The influence weights and sensitivity are determined using a high-order (e.g., fourth power) form, so that when the deviation is small, the impact on the repair pose compliance coefficient is insignificant, but when the deviation increases, its impact will be rapidly amplified, thereby achieving a strong penalty for serious deviations. These factors can be set by expert experience, or trained and adjusted by optimization algorithms (such as genetic algorithms and particle swarm optimization) combined with experimental data, so that the repair pose compliance coefficient output by the model is highly consistent with the actual welding quality assessment results.

[0065] The repair pose compliance model in this application comprehensively considers the two-dimensional offset of the welding wire tip relative to the center of the laser spot. And the actual measured value of the fillet radius at the root of the repair bevel. These two key geometric parameters enable a quantitative assessment of the geometric state of weld repairs. The model will... Two-dimensional offset from the maximum tolerance level allowed by the process Normalization is performed, and a higher-order penalty factor is introduced. To keenly capture the impact of welding wire alignment deviation on welding energy distribution, when the welding wire offset... When increased, its compliance coefficient with the repaired pose The negative impact will be rapidly amplified by the fourth term, thereby effectively avoiding problems such as insufficient penetration or incomplete fusion of the sidewalls caused by the welding wire deviating too far from the center of the spot.

[0066] Meanwhile, the model incorporates the actual measured value R_R of the fillet radius at the root of the repair bevel and the optimal theoretical fillet radius at the root. The deviation, and compare it with the allowable tolerance. Compare and combine with higher-order penalty factors When the radius of the fillet at the root of the bevel deviates from the ideal value, the deviation affects the compliance coefficient of the repair posture. The influence is also amplified through a fourth-order term, thus effectively assessing the constraint of groove geometry on the fusion behavior of filler metal and ensuring weld formation quality. Through a combination of logarithmic and higher-order power functions, this model can map the aforementioned multi-source geometric deviations into a single repair pose compliance coefficient. This coefficient can intuitively quantify the compliance level of the current repair pose, providing a basis for the control module to subsequently base its work on the stability coefficient. Repairing the posture compliance coefficient and current focus deviation Construct a relative position deviation optimization model and output the target focus deviation. It provides key compliance references, enabling the control module to make targeted and precise adjustments to the repair pose, thereby ensuring a high degree of controllability in the geometric dimension of the welding repair process and laying a solid mathematical foundation for achieving high-precision focus position adjustment.

[0067] By introducing a two-dimensional offset of the welding wire tip relative to the center of the laser spot Actual measured value of the fillet radius at the root of the repair bevel. Combined with a high-order penalty mechanism, this repair posture compliance model can sensitively reflect the impact of wire alignment deviation and groove geometry on welding quality. This allows the control module to obtain an intuitive and quantitative repair posture compliance coefficient. This overcomes the limitation of traditional methods in quantifying the specific impact of geometric deviations on welding quality. Furthermore, it increases the compliance coefficient of the repair posture. As an important input to the relative position deviation optimization model, and the stable attitude coefficient and current focus deviation The synergistic effect enables the control module to output the target focus deviation more accurately. It also drives the laser welding module to adjust the focus position, which significantly improves the forming accuracy and quality stability of laser welding repair, effectively avoiding defects such as insufficient penetration, sidewall incomplete fusion or poor weld formation caused by welding wire deviation or non-compliant bevel, thereby ensuring the reliability and consistency of casting repair.

[0068] The multi-source parameter collaborative steady-state model is constructed as follows: the atmosphere cleanliness coefficient and the airflow-mechanical stability coefficient are weighted and summed using prior weights to obtain the basic steady-state amplitude; the dimensionless deviation square terms of the instantaneous distance fluctuation value relative to the extreme height and the angle between the four axes of the air nozzle relative to the maximum deterioration angle are extracted respectively, and a rational fraction is constructed after being weighted by the sensitivity coefficient as the phase adjustment term; finally, the basic steady-state amplitude is multiplied by the triangular periodic decay function constructed by the phase adjustment term to output the dimensionless steady-state coefficient.

[0069] The multi-source parameter collaborative stable attitude model is specifically represented as follows:

[0070] In the formula, To stabilize the attitude coefficient, the unit of measurement is: ; and These are the atmosphere cleanliness coefficient and the airflow-mechanical stability coefficient, respectively, with units of _____. ; This represents the instantaneous fluctuation value of the real-time distance between the air nozzle 4 and the workpiece surface, with units of _____. ; The extreme vertical distance that triggers the critical failure of the protective gas, with dimensions in units of . ; The angle between the axis of the gas nozzle 4 and the tangent of the instantaneous welding direction, with dimensions in units of . ; The angle that causes the protective airflow to deteriorate significantly is measured in units of _____. ; and To satisfy and for State assignment weights, with units of measurement of . ; and This is the sensitivity coefficient for distance and angle fluctuations, with units of _____. .

[0071] Among them, the stable attitude coefficient This is the core output of the multi-source parameter synergistic stability model proposed in this application. It is used to quantify the comprehensive state of environmental cleanliness, mechanical stability, and protective gas flow coverage during laser welding. Its value range is usually between 0 and 1. The larger the value, the better the stability of the welding process, and vice versa. The calculation of this coefficient comprehensively considers a variety of key factors affecting welding quality, providing a quantitative basis for subsequent process parameter adjustments.

[0072] Atmosphere cleanliness coefficient The purity of the protective atmosphere in the welding area is characterized by dynamic oxygen partial pressure and shielding gas dew point temperature. It can be obtained through a pre-established mathematical model, such as calculation based on real-time measurement data from oxygen and dew point sensors. This coefficient is an important indicator for assessing the quality of the welding environment and is directly related to the degree of oxidation and mechanical properties of the weld.

[0073] airflow-mechanical stability coefficient It reflects the stability of the protective gas flow and the mechanical vibration level of the welding equipment during the welding process. It can be obtained by using a pre-established mathematical model, such as calculation based on the turbulence intensity at the outlet of the gas nozzle 4 and the kurtosis of the acceleration spectrum of the welding torch 3 head. This coefficient is a key indicator to ensure effective coverage of the protective gas flow and smooth welding process.

[0074] Instantaneous fluctuation value of the real-time distance between air nozzle 4 and workpiece surface This indicates the real-time change in the distance between the gas nozzle 4 and the workpiece surface during the welding process. It can be obtained by monitoring the relative distance between the gas nozzle 4 and the workpiece in real time using a laser rangefinder or a vision sensor, and calculating its instantaneous deviation relative to the set reference distance. Another method is to use a high-frequency displacement sensor to directly measure the small vibration of the gas nozzle 4 in the vertical direction. This fluctuation value directly affects the coverage and stability of the protective gas flow.

[0075] Extreme vertical distance that triggers the critical failure of the protective gas A critical distance is defined at which the protective airflow will be unable to effectively cover the molten pool when the distance between the gas nozzle 4 and the workpiece surface exceeds a certain threshold, thus leading to protection failure. This value can be obtained through experimental calibration, for example, by measuring the oxygen concentration in the molten pool area at different distances to determine the distance point where the protection effect significantly decreases.

[0076] The angle between the axis of the gas nozzle 4 and the tangent of the instantaneous welding direction This represents the angular deviation between the central axis of the gas nozzle 4 and the tangent to the edge of the molten pool in the welding travel direction. It can be obtained by using a coaxial vision component to identify the welding torch 3's posture and welding trajectory in real time, and calculating the angle between the axis of the gas nozzle 4 and the trajectory tangent. An excessively large angle may cause the shielding gas flow to deviate from the molten pool, resulting in a decrease in shielding effectiveness and the maximum deterioration angle leading to severe shielding gas flow leakage. A critical angle is defined where, when the angle between the axis of the gas nozzle 4 and the tangent of the welding direction exceeds a certain threshold, the protective gas flow will severely dissipate, failing to effectively protect the molten pool. This value can be obtained through experimental calibration, such as measuring the oxygen concentration in the molten pool region at different angles to determine the angle point where the protective effect severely deteriorates, or it can be determined by simulating the gas flow distribution at different gas nozzle 4 tilt angles, with state weighting. and Used to adjust the atmospheric cleanliness coefficient and airflow-mechanical stability coefficient The relative importance of these factors in the calculation of the stability coefficient, which satisfy the condition that their sum equals 1, can be set according to specific welding process requirements, material properties, or empirical knowledge. For example, for materials sensitive to oxidation, the stability coefficient can be appropriately increased. The value of the distance and angle fluctuation sensitivity coefficient. and The correction terms used to adjust the influence of the distance fluctuation between the gas nozzle 4 and the workpiece, and the included angle of the gas nozzle 4 axis on the steady-state coefficient, can be adjusted based on experimental data or expert experience to ensure the model has appropriate sensitivity to geometric position changes during actual welding. For example, if the distance fluctuation has a greater impact on the protection effect, the correction term can be appropriately increased. The value of .

[0077] By analyzing the atmosphere cleanliness coefficient and airflow-mechanical stability coefficient A weighted summation was performed to construct a basic stable attitude assessment benchmark, in which the atmosphere cleanliness coefficient was used. It comprehensively reflects the purity of the protective atmosphere in the welding area, while the gas flow-mechanical stability coefficient... This quantifies the stability of the protective airflow and the mechanical vibration level of the equipment. This weighted combination ensures that the cleanliness of the welding environment and the stability of equipment operation are balanced in the evaluation system. On this basis, the model introduces a correction term based on the cosine square function to dynamically capture the influence of the geometric position change between the gas nozzle 4 and the workpiece during the welding process on the protective airflow coverage effect.

[0078] The core of this correction lies in utilizing the instantaneous fluctuation value of the real-time distance between the air nozzle 4 and the workpiece surface. Extreme vertical distance from the critical point of triggering protective gas failure The ratio, and the angle between the axis of the gas nozzle 4 and the tangent of the instantaneous welding direction. The angle with the maximum deterioration that causes severe leakage of protective airflow By squaring and weighting these ratios, and then mapping them through a nonlinear function (cosine square function), the model can sensitively reflect the destructive effect of distance fluctuations and angle deviations on the integrity of the protective airflow.

[0079] When distance fluctuations or angular deviations increase, the value of the correction term decreases, thereby reducing the overall stability coefficient. Therefore, this model not only considers static environmental parameters and mechanical vibrations, but also incorporates dynamic geometric position changes, resulting in a more accurate and stable output coefficient. It can comprehensively and in real time reflect the overall robustness of the welding process, thus effectively solving the limitations of traditional models in evaluating the actual coverage effect of the protective gas flow.

[0080] By constructing a multi-source parameter collaborative stability model, the dynamic geometric positional changes of factors such as atmosphere cleanliness, mechanical stability, and the fluctuation of the distance between the gas nozzle 4 and the workpiece, and the included angle of the gas nozzle 4 axes are organically integrated, which makes the stability coefficient... It can more accurately quantify the overall robustness of the welding process, especially when facing complex repair paths or instantaneous posture adjustments. It can dynamically capture the actual coverage efficiency of the protective gas flow, thereby effectively improving the quality and stability of laser welding repair of surface defects in castings and reducing the risk of defects caused by protective gas flow failure.

[0081] The relative position deviation optimization model is constructed in the following way: the product of the stable attitude coefficient and the corrected pose compliance coefficient is used as the state driving factor, and a polynomial rational transition function with S-shaped curve characteristics is constructed as the dynamic interpolation weight; using the dynamic interpolation weight, a smooth difference transition is performed between the currently measured instantaneous relative position deviation of the focus and the system's preset safe focus offset reference quantity, and the output is used as the target focus deviation of the final execution command.

[0082] The relative position deviation optimization model is specifically expressed as follows:

[0083] In the formula, The instantaneous relative position deviation between the target laser beam focus and the workpiece surface, measured in units of . ; The instantaneous relative position deviation between the laser beam focus and the workpiece surface, measured in units of . ; and These are the stability attitude coefficient and the corrected posture compliance coefficient, respectively, with units of measurement of [missing information]. ; This is the system's default safety focus offset reference value, with units of [unit missing]. .

[0084] The instantaneous relative position deviation between the target laser beam focus and the workpiece surface in this model This represents the ideal vertical distance between the laser beam focus and the workpiece surface, calculated using the optimized model. It guides the laser welding module in precisely adjusting the focus position to ensure the laser energy density remains optimal throughout the welding process. The measured instantaneous relative position deviation between the laser beam focus and the workpiece surface is also included. This represents the actual vertical distance deviation between the laser beam focus and the workpiece surface, which is directly obtained through real-time sensing during the laser welding process. This can be achieved by continuously scanning the workpiece surface with a laser rangefinder and calculating the result in conjunction with the laser optical path parameters.

[0085] Stable Attitude Coefficient It is a comprehensive indicator measuring the overall stability of the current welding environment. Its value is calculated collaboratively from multiple factors such as atmosphere purity, airflow-mechanical stability, and real-time attitude fluctuations. The stability coefficient reflects the degree to which the welding process is affected by external interference and internal uncertainties. Its level is directly related to the potential risks to welding quality. (The text also mentions a repair posture compliance coefficient, but this seems unrelated to the previous sentence and is likely a separate topic.) It is used to evaluate the relative position of the welding wire and the spot, as well as the standardization of the repair groove geometry. It comprehensively considers key parameters such as the welding wire centering accuracy and the radius of the groove root fillet, and reflects the degree of geometric matching of the welding repair operation. The higher the repair posture compliance coefficient, the more the welding posture meets the process requirements, and the more conducive it is to forming a high-quality weld.

[0086] The system's default safety focus offset reference value It is a preset focus offset value used as a safety guarantee when the system is unstable or the pose is non-compliant. It represents a relatively conservative focus position that can ensure basic welding effect. When the welding conditions deteriorate, the optimization model increases the weight of this reference value to guide the focus to a relatively safe area, thereby avoiding welding defects caused by extreme deviations. The relative position deviation optimization model is a non-linear mathematical function. Its core function is to dynamically calculate the optimal target focus position based on the real-time measured focus deviation, the stability of the welding process, and the pose compliance. It realizes a smooth transition between accurately tracking the real-time deviation and returning to the safety reference value, ensuring the adaptive adjustment capability of the laser focus under complex and variable working conditions.

[0087] Using stable attitude coefficient With the compliance coefficient of the repaired pose The product of these factors serves as the core weighting factor, and is applied to the currently measured focal deviation through a nonlinear mapping function. Compared with the system's preset safety benchmark quantity Weighted fusion is performed. This is especially important when the welding environment is stable and the orientation compliance is high (i.e.,...). (If the value is large), the model will assign the currently measured real-time focus deviation. Higher weights result in a higher calculated target focus deviation. It more closely approximates real-time measurements, thus enabling precise tracking and rapid response of the welding process. Conversely, when environmental interference is significant, system stability decreases, or pose deviations exceed expectations (i.e., ... (If the value is small), the model will automatically increase the safety focus offset reference value. The weights cause the target focus to deviate. The shift towards a pre-defined safe zone effectively prevents welding quality from spiraling out of control due to environmental fluctuations. (Stability coefficient) This comprehensively reflects the impact of atmosphere purity, airflow-mechanical stability, and attitude fluctuations on the welding process, while the repair posture compliance coefficient... This quantifies the alignment between the welding wire and the laser spot, as well as the matching degree of the bevel geometry. By introducing these two key coefficients into the focus optimization model, the overall risk of the current welding state can be comprehensively assessed, and the focus position can be intelligently adjusted accordingly, significantly improving the robustness of the laser welding repair process and the consistency of welding quality.

[0088] The multi-source sensing module also includes a protective gas extraction capillary tube. One end of the capillary tube extends to the protective gas coverage area at the edge of the molten pool, and the other end is connected to an oxygen sensor for real-time gas extraction to obtain dynamic oxygen partial pressure. The dew point sensor directly measures the relative humidity and temperature of the gas in the gas path and converts them into an analog current signal characterizing the dew point temperature of the protective gas.

[0089] A protective gas extraction capillary is a long, thin tubular structure used to precisely extract gas samples from a specific area. The capillary can be made of high-temperature and corrosion-resistant quartz glass or stainless steel, and its inner diameter can be optimized according to the required gas flow rate and response speed, for example, from 0.1 mm to 0.5 mm.

[0090] One end of the capillary extends to the protective gas coverage area at the edge of the molten pool, ensuring that the collected gas sample can truly reflect the actual atmosphere in the molten pool area, effectively avoiding interference from the external ambient air, thereby improving the accuracy of oxygen partial pressure measurement.

[0091] The end of the capillary is fixed at a specific position on the welding torch 3, so that it always maintains a relative position with the edge of the molten pool during the welding process. Alternatively, the capillary end can be dynamically tracked and precisely aligned with the edge of the molten pool by a micro-robotic arm or guide mechanism integrated in the coaxial vision component, so as to adapt to different welding postures and changes in molten pool shape.

[0092] The other end of the capillary is connected to an oxygen sensor, which is used to deliver the protective gas sample extracted from the edge of the molten pool to the oxygen sensor so that the sensor can detect the oxygen content in real time and obtain dynamic oxygen partial pressure data. The capillary can be tightly connected to the air inlet of the oxygen sensor through a standard gas connector (such as a ferrule connector or a threaded connector) to ensure airtightness, and continuous pumping can be achieved through a micro air pump or negative pressure system.

[0093] Dew point sensors are used to directly measure the relative humidity and temperature of the gas in the gas path and convert them into an analog current signal that characterizes the dew point temperature of the protective gas. Their function is to monitor the moisture content in the protective gas in real time. By measuring the relative humidity and temperature and performing the conversion, the dew point temperature is obtained. The dew point sensor can be a combination of a capacitive or impedance humidity sensor and a thermistor temperature sensor. The built-in microprocessor calculates the dew point temperature in real time based on the humidity and temperature data using the standard dew point calculation formula and converts it into an analog current signal output of 4-20mA or 0-10V.

[0094] By precisely placing one end of the protective gas capillary tube in the protective gas coverage area at the edge of the molten pool, and by real-time extraction of the gas and delivery to the oxygen sensor, interference from ambient airflow on the oxygen sensor measurement results is effectively avoided, ensuring that the acquired dynamic oxygen partial pressure data can truly reflect the protective effect around the molten pool.

[0095] Meanwhile, the dew point sensor directly measures the relative humidity and temperature of the gas in the gas path and converts them into an analog current signal, enabling the control module to obtain the protective gas dew point temperature in real time and accurately. This provides high-precision data support for the atmosphere cleanliness model built in the control module, making the calculation of the atmosphere cleanliness coefficient more accurate. Ultimately, this allows the relative position deviation optimization model to output a more accurate target focus deviation and drive the laser welding module to adjust the focus position, thereby significantly improving the overall stability of laser welding repair of casting surface defects.

[0096] The process of the control module to obtain the kurtosis of the acceleration spectrum includes: performing a fast Fourier transform on the acceleration time series of the welding torch head 3 collected by the triaxial accelerometer to calculate the energy distribution of each frequency band, and further performing fourth-order central moment statistics on the acceleration amplitude series to obtain the kurtosis.

[0097] The Fast Fourier Transform (FFT) of the acceleration time series of the welding torch head 3 acquired by the triaxial accelerometer refers to the efficient conversion of the continuous or discrete acceleration signal of the welding torch head 3 in the time domain to the frequency domain using the Fast Fourier Transform (FFT) algorithm. The Fast Fourier Transform can decompose complex time-domain vibration signals into simple sine wave components of different frequencies, thereby enabling the analysis of the frequency characteristics of the signal and the separation of noise and effective vibration information. This transformation can be implemented by the FFT library functions built into the digital signal processor (DSP) or microcontroller, such as using the CMSIS-DSP library on the Cortex-M series microcontroller.

[0098] Calculating the energy distribution of each frequency band refers to determining the amount of energy contained in the signal in different frequency intervals after Fast Fourier Transform (FFT). By analyzing the energy distribution, key vibration frequencies related to the mechanical stability of the welding process can be identified, and background noise or high-frequency interference unrelated to welding can be filtered out. In the FFT results, the energy contribution of each frequency component is usually obtained by calculating the square of the amplitude. Then, the energy in a specific frequency range can be accumulated or averaged to obtain the energy distribution of that frequency band.

[0099] The kurtosis obtained by performing fourth-order central moment statistics on an acceleration amplitude sequence refers to using the kurtosis index in statistics to measure the shape of the probability distribution of the acceleration amplitude sequence. Kurtosis describes the thickness of the tail (i.e., the frequency of extreme values) and the sharpness of the peaks. It can sensitively capture the impact, transient anomalies, or non-Gaussian characteristics in vibration signals, which are often caused by mechanical instability or airflow disturbances. It reflects abnormal fluctuations in the welding process better than simple mean or variance. The kurtosis can be calculated by programming the mean and standard deviation of the acceleration amplitude sequence and then applying the definition formula of the fourth-order central moment, or it can be calculated using specialized statistical analysis software or libraries (such as R language). of These tools use functions to perform direct calculations and typically provide optimized algorithms for processing large amounts of data.

[0100] The control module first receives real-time acceleration time-series data from the triaxial accelerometer in the multi-source sensing module. To effectively separate noise and focus on vibration components meaningful for welding stability, this time-series data is processed by the Fast Fourier Transform (FFT) module. The FFT converts the time-domain signal into a frequency-domain signal. By analyzing the energy distribution of these frequency bands, the system can identify specific frequency ranges related to mechanical vibration and airflow disturbances, and effectively filter out irrelevant background noise. The control module further performs fourth-order central moment statistics on the acceleration amplitude sequence to obtain kurtosis values, which can accurately quantify the sharpness and tail thickness of the vibration signal distribution, thereby effectively reflecting abnormal fluctuations caused by airflow turbulence or mechanical vibration during the welding process. This combination of frequency domain analysis and high-order statistical feature extraction ensures the accuracy of the mechanical stability assessment of the welding process and provides a reliable basis for the accurate calculation of the subsequent stability coefficient.

[0101] The multi-source sensing module also includes a line laser scanner. The actual measured value of the fillet radius at the root of the bevel is obtained by scanning the bevel at the welding front edge with the line laser scanner to obtain the cross-sectional contour point cloud, and then extracting it through the arc fitting algorithm in the control module. The two-dimensional offset is obtained by the control module through real-time identification of the pixel offset between the center of the light spot and the contour of the end of the welding wire by the coaxial vision component and then calibration and conversion.

[0102] A Fast Fourier Transform (FFT) is performed on the acceleration time series of the welding torch head (3) acquired by a triaxial accelerometer to calculate the energy distribution in each frequency band. This step aims to convert the vibration signal generated by the welding torch head (3) during welding from the time domain to the frequency domain, thereby revealing the distribution of vibration energy at different frequencies. The FFT decomposes the complex, time-varying acceleration signal into a series of sinusoidal components of different frequencies and calculates the amplitude or energy of each frequency component, helping to identify specific frequency vibration modes that may be caused by mechanical resonance, airflow disturbance, or equipment wear. The implementation uses a digital signal processor (DSP) or the microcontroller's built-in FFT algorithm library to process the acquired discrete time series in real-time or near real-time.

[0103] Furthermore, the kurtosis is obtained by performing fourth-order central moment statistics on the acceleration amplitude sequence. Kurtosis is a statistical indicator that measures the steepness of a data distribution, reflecting the tail characteristics and the presence of outliers. For acceleration amplitude sequences, high kurtosis usually indicates the presence of more extreme vibration events or impacts, which may indicate unstable factors in the welding process, such as welding torch vibration, gas flow impact, or molten pool fluctuations. By calculating kurtosis, the non-Gaussianity or impact of the vibration signal can be quantified, thus more sensitively capturing potential mechanical instability states. Existing statistical analysis library functions, such as... In the library The function calculates the acceleration amplitude data.

[0104] Real-time vibration data of the welding torch head is continuously acquired using a triaxial accelerometer, forming an acceleration time series. These raw time-domain signals contain rich mechanical vibration information, but direct analysis of their stability is difficult. Therefore, the control module performs a Fast Fourier Transform (FFT) on these time series, converting them into frequency-domain signals to reveal the distribution of vibration energy at different frequencies. This conversion process effectively filters out background noise and highlights specific frequency components related to welding stability, such as periodic vibrations caused by equipment resonance or airflow turbulence. To more precisely capture the non-Gaussian or impact characteristics of the vibration signal, the control module further performs fourth-order central moment statistics on the acceleration amplitude series to obtain kurtosis values. Kurtosis, as a statistic measuring the sharpness of data distribution, is highly sensitive to abnormal vibration events and can promptly identify… During the welding process, non-stationary states caused by mechanical vibration or airflow impact can be addressed through a layered analysis from the time domain to the frequency domain and then to the statistical characteristic domain. This allows the system to extract the key indicator characterizing welding stability—acceleration spectrum kurtosis—from complex vibration data. This kurtosis value is then input into the airflow-mechanical stability model, where it interacts with the turbulence intensity at the outlet of nozzle 4 to generate an airflow-mechanical stability coefficient. This coefficient is further combined with the atmosphere cleanliness coefficient and real-time attitude fluctuations to construct a multi-source parameter collaborative stability model, ultimately obtaining the stability coefficient. This multi-level, multi-dimensional stability assessment mechanism enables the control module to perceive the mechanical and airflow stability during the welding process more comprehensively and accurately, effectively compensating for the limitations of single visual feedback in dynamic environments and ensuring the accuracy and quality of laser welding repair.

[0105] The terms "front," "back," "left," "right," "top," and "bottom" all refer to the figures in the accompanying drawings. Figure 1 Based on the perspective of the observer, the side of the device facing the observer is defined as the front, the left side of the observer is defined as the left, and so on.

[0106] In the description of this invention, it should be understood that the terms "center", "longitudinal", "lateral", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the scope of protection of this invention.

[0107] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A laser welding repair device for surface defects of castings, comprising a welding device body (1), characterized in that, Also includes: The laser welding module includes a welding torch (3), a gas nozzle (4), a wire feeding mechanism (2), and a coaxial vision assembly; The multi-source sensing module includes a dew point sensor in the gas path, a flow rate probe downstream of the gas nozzle (4) outlet, a triaxial accelerometer in the head of the welding torch (3), an oxygen sensor for measuring the oxygen concentration at the edge of the molten pool, and a laser rangefinder. The control module is communicatively connected to both the laser welding module and the multi-source sensing module. The control module is used to perform the following operations: Dynamic oxygen partial pressure and protective gas dew point temperature are obtained to construct an atmosphere cleanliness model and obtain the atmosphere cleanliness coefficient; To obtain the turbulence intensity and acceleration spectral kurtosis at the outlet of the air nozzle (4) in order to construct an airflow-mechanical stability model and obtain the airflow-mechanical stability coefficient; The two-dimensional offset between the welding wire and the spot and the radius of the fillet at the root of the groove are obtained to construct a repair posture compliance model and obtain the repair posture compliance coefficient. Based on the atmosphere cleanliness coefficient, airflow-mechanical stability coefficient, and real-time attitude fluctuation, a multi-source parameter collaborative stable attitude model is constructed to obtain the stable attitude coefficient. Based on the stable attitude coefficient, the repair posture compliance coefficient, and the current focus deviation, a relative position deviation optimization model is constructed to output the target focus deviation and drive the laser welding module to adjust the focus position.

2. The laser welding repair device for surface defects of castings according to claim 1, characterized in that: The atmosphere cleanliness model is constructed in the following way: extract the dimensionless deviation square term of the dynamic oxygen partial pressure relative to the reference safety threshold, and the dimensionless deviation square term of the real-time gas dew point temperature relative to the process allowable boundary. The two dimensionless deviation square terms are weighted and summed by weight adjustment coefficients and used as the variable terms in the denominator. A rational algebraic decay function is constructed in this way to converge the atmosphere cleanliness coefficient to the dimensionless preset interval.

3. The laser welding repair device for surface defects of castings according to claim 1, characterized in that: The airflow-mechanical stability model is constructed in the following way: extract the dimensionless deviation cubic term of the turbulence intensity at the outlet of the real-time air nozzle (4) relative to the critical intensity of turbulence change, and the dimensionless absolute deviation cubic term of the acceleration spectrum kurtosis relative to the allowable maximum value. The two dimensionless deviation cubic terms are weighted and summed by sensitivity adjustment gain and used as independent variables. In this way, an inverse triangular mapping function is constructed, thereby outputting the dimensionless airflow-mechanical stability coefficient.

4. The laser welding repair device for surface defects of castings according to claim 1, characterized in that: The repair posture compliance model is constructed in the following way: extract the dimensionless high-order term of the two-dimensional offset of the welding wire end relative to the process tolerance, and the dimensionless high-order term of the actual error of the fillet radius at the root of the bevel relative to the allowable tolerance. The two terms are weighted and summed by a high-order penalty factor and superimposed with a basic constant as the proper term and the reciprocal is taken. In this way, a logarithmic decay mapping function is constructed to output the dimensionless repair posture compliance coefficient.

5. The laser welding repair device for surface defects of castings according to claim 1, characterized in that: The multi-source parameter collaborative steady-state model is constructed in the following way: the atmosphere cleanliness coefficient and the airflow-mechanical stability coefficient are weighted and summed by prior weights to obtain the basic steady-state amplitude; the dimensionless deviation square terms of the instantaneous distance fluctuation value relative to the extreme height and the angle between the nozzle (4) axis and the maximum deterioration angle are extracted respectively, and a rational fraction is constructed after being weighted by the sensitivity coefficient as the phase adjustment term; finally, the basic steady-state amplitude is multiplied by the triangular periodic decay function constructed by the phase adjustment term to output the dimensionless steady-state coefficient.

6. The laser welding repair device for surface defects of castings according to claim 5, characterized in that: The relative position deviation optimization model is constructed in the following way: the product of the stable attitude coefficient and the repair pose compliance coefficient is used as the state driving factor, and a polynomial rational transition function with S-shaped curve characteristics is constructed as the dynamic interpolation weight; using the dynamic interpolation weight, a smooth difference transition is performed between the currently measured instantaneous relative position deviation of the focus and the system's preset safe focus offset reference amount, and the target focus deviation is output as the final execution instruction.

7. The laser welding repair device for surface defects of castings according to claim 1, characterized in that: The multi-source sensing module also includes a protective gas extraction capillary tube. One end of the capillary tube extends to the protective gas coverage area at the edge of the molten pool, and the other end is connected to the oxygen sensor for real-time gas extraction to obtain the dynamic oxygen partial pressure. The dew point sensor directly measures the relative humidity and temperature of the gas in the gas path and converts them into an analog current signal characterizing the dew point temperature of the protective gas.

8. The laser welding repair device for surface defects of castings according to claim 1, characterized in that: The process of the control module to obtain the kurtosis of the acceleration spectrum includes: performing a fast Fourier transform on the acceleration time series of the welding torch (3) head collected by the triaxial accelerometer to calculate the energy distribution of each frequency band, and further performing fourth-order central moment statistics on the acceleration amplitude series to obtain the kurtosis.

9. The laser welding repair device for surface defects of castings according to claim 1, characterized in that: The multi-source sensing module also includes a line laser scanner. The actual measured value of the radius of the corner at the root of the repair bevel is obtained by the line laser scanner scanning the bevel at the welding front to obtain the cross-sectional contour point cloud, and then extracted by the arc fitting algorithm in the control module. The two-dimensional offset is obtained by the control module through the coaxial vision component to identify the pixel offset between the center of the spot and the contour of the end of the welding wire in real time and then calibrating and converting it.

Images