Bridge steel member welding parameter self-adjusting control system and method

Abstract

The application relates to the technical field of bridge construction, and discloses a bridge steel member welding parameter self-adjusting control system and method, which comprises an acoustic sensing module, which is used for simultaneously acquiring real-time feedback data reflecting the current molten pool geometric shape and feedforward data reflecting disturbance in front of a welding path; a feedback calculation module, which calculates a feedback adjustment amount according to the deviation of the feedback data from a preset target; and a feedforward calculation module, which calculates a feedforward adjustment amount according to the feedforward data to suppress the disturbance in advance. By combining direct feedback control of the molten pool internal geometric shape, predictive feedforward control of the front disturbance and online self-adaptive updating of the control model, accurate closed-loop control of key indexes such as the penetration depth can be realized, sudden disturbances such as changes in the groove can be actively responded to, and changes in material properties in the welding process can be adapted, so that the welding quality and stability of the bridge steel member under complex dynamic working conditions are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of bridge construction technology, specifically to a self-adjusting control system and method for welding parameters of bridge steel components. Background Technology

[0002] As a key component of modern transportation infrastructure, the construction quality of bridge steel structures directly affects public safety. Welding is a core process in the manufacture of bridge steel structures, and the quality of the welds, especially their mechanical properties and internal density, is a crucial factor determining the overall load-bearing capacity and service life of the structure.

[0003] Currently, most bridge steel component welding employs automated welding technologies, such as submerged arc welding or gas shielded welding. These involve robots or automated specialized machines equipped with welding torches, following pre-set welding procedures. This open-loop or semi-closed-loop control method, based on fixed parameters, can ensure stable welding quality under ideal working conditions.

[0004] However, this control method reveals its inherent limitations when faced with the dynamic changes in the actual welding process. During the welding of long seams in large steel components, the continuous input of heat causes uneven thermal deformation of the component, thus altering the relative position of the welding torch and the workpiece. Furthermore, assembly errors, beveling tolerances, and temporary spot welds can all lead to unpredictable changes in the beveling morphology of the weld joint (such as gap width and misalignment) along the welding path. These disturbances directly disrupt the match between the preset process parameters and the actual welding conditions, potentially resulting in welding defects such as incomplete fusion, incomplete penetration, or burn-through.

[0005] To address these disturbances, existing technologies attempt to incorporate sensing techniques to achieve closed-loop control. However, commonly used sensing methods, such as arc signal-based sensing or machine vision-based sensing, have limitations. Arc signal sensing indirectly infers the molten pool state by analyzing fluctuations in welding current and voltage, but it is susceptible to electromagnetic interference and has limited information dimensions. Machine vision sensing mainly relies on image analysis of the arc and molten pool surface morphology, but due to interference from strong light and fumes, as well as its inherent limitation of not being able to penetrate metal surfaces, it is difficult to directly obtain key geometric information about the interior of the molten pool, especially the penetration depth that determines the strength of the weld joint.

[0006] Furthermore, these feedback control systems are inherently lagging, only able to compensate for welding deviations that have already occurred. By the time the system detects the deviation and makes adjustments, a substandard weld may have already formed, which is unacceptable for demanding bridge structures. Meanwhile, for more advanced model-based control strategies, their performance is highly dependent on the accuracy of the model. In long-duration welding tasks, the overall temperature rise of the workpiece alters the thermophysical parameters of the material, causing the initially established control model to gradually become inaccurate, thus affecting the stability of the control effect. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a self-adjusting control system and method for welding parameters of bridge steel components. This solves the problems that existing bridge steel component welding technologies are unable to adapt to dynamic changes in working conditions caused by thermal deformation, assembly errors, etc., and lack direct, real-time detection and closed-loop control of the internal geometry of the molten pool, especially the penetration depth.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] The first aspect of the present invention provides a self-adjusting control system for welding parameters of bridge steel components. The system includes: an acoustic sensing module, a feedback calculation module, a feedforward calculation module, a model update module, a fusion module, and a control execution module.

[0010] The acoustic sensing module is used to acquire real-time feedback data reflecting the current geometry of the molten pool, feedforward data reflecting disturbance information ahead of the welding path, and generate echo signals containing waveform morphology characteristics. In one specific embodiment, the acoustic sensing module includes an acoustic phased array system. This acoustic phased array system operates in two modes: a feedback mode, in which the system emits acoustic pulses towards the current welding area and calculates the penetration depth and weld width as real-time feedback data based on the echo signal generated at the interface between the solid metal and the liquid molten pool through time-of-flight calculation; and a feedforward mode, in which the system deflects the sound beam ahead of the welding path and generates feedforward data characterizing geometric discontinuities such as bevel gaps or misaligned edges by analyzing the differences between the detected echo characteristics and reference characteristics.

[0011] The feedback calculation module, connected to the acoustic sensing module, is used to calculate the feedback adjustment amount based on the deviation between the real-time feedback data and the preset target geometric state, using a Jacobian matrix. Specifically, this module uses the actual geometric vector formed by the real-time feedback data... With the preset target geometric vector By comparison, the geometric error vector is obtained. Feedback adjustment amount This is obtained by solving the following relation:

[0012] ;

[0013] in, The pseudo-inverse of the Jacobian matrix is ​​given. The geometric error vector is... This is a preset proportional gain matrix.

[0014] The feedforward calculation module, connected to the acoustic sensing module, is used to calculate the feedforward adjustment amount based on the feedforward data. Specifically, this module first calculates based on a disturbance influence model. The perturbation vector generated from the feedforward data This is converted into the expected geometric deviation. Subsequently, based on the Jacobian matrix, a feedforward adjustment amount that can offset this expected geometric deviation is calculated. Its calculation follows the following relationship:

[0015] ;

[0016] in, The pseudo-inverse of the Jacobian matrix is ​​given. The disturbance vector is... This is the disturbance impact model. It is a preset feedforward gain matrix.

[0017] The model update module, connected to the acoustic sensing module, is used to analyze the waveform morphology characteristics of the echo signal to estimate the acoustic characteristic parameters of the welding area in real time, and to update or calibrate the Jacobian matrix online based on the acoustic characteristic parameters. The waveform morphology characteristics include at least one of signal amplitude attenuation, frequency shift, and waveform dispersion.

[0018] The fusion module, connected to the feedback calculation module and the feedforward calculation module, is used to fuse the feedback adjustment amount and the feedforward adjustment amount to generate the total adjustment amount of the welding parameters. In one embodiment, the fusion method is algebraic summation.

[0019] The control execution module, connected to the fusion module, is used to adjust the welding parameters of the welding equipment according to the total adjustment amount. The welding parameters include at least one of welding current, arc voltage, and welding speed.

[0020] A second aspect of the present invention provides a method for self-adjusting control of welding parameters for bridge steel components, the method comprising the following steps:

[0021] S1. Through the acoustic sensing module, real-time feedback data reflecting the current geometry of the molten pool, feedforward data reflecting the disturbance information in front of the welding path, and echo signals containing waveform morphology characteristics are acquired.

[0022] S2. Through the feedback calculation module, the feedback adjustment amount is calculated based on the deviation between the real-time feedback data and the preset target geometric state and on a Jacobian matrix.

[0023] S3. Calculate the feedforward adjustment amount based on the feedforward data using the feedforward calculation module;

[0024] S4. Analyze the waveform characteristics of the echo signal through the model update module to update or calibrate the Jacobian matrix online.

[0025] S5. The feedback adjustment amount and the feedforward adjustment amount are fused together through the fusion module to generate the total adjustment amount of the welding parameters;

[0026] S6. Based on the total adjustment amount, adjust the welding parameters of the welding equipment through the control execution module.

[0027] This invention provides a self-adjusting control system and method for welding parameters of bridge steel components. It has the following beneficial effects:

[0028] 1. By setting up an acoustic sensing module, this invention can acquire the internal geometry of the weld pool, including the penetration depth and weld width, in real time as real-time feedback data. The feedback calculation module calculates the deviation between the direct measurement data and the preset target, thereby realizing closed-loop control of the weld pool geometry. This enables precise maintenance of the penetration state during the welding process and solves the technical problem of existing technologies that rely on indirect information and cannot accurately control the internal state of the weld pool.

[0029] 2. This invention uses an acoustic sensing module to detect disturbances such as the bevel gap ahead of the welding path in advance, and a feedforward calculation module to calculate a compensatory feedforward adjustment amount in advance. This feedforward control mechanism, combined with feedback control, enables the system to actively adjust parameters before the disturbance actually affects the molten pool, thereby effectively suppressing welding process fluctuations caused by sudden changes in working conditions and improving the stability of welding quality.

[0030] 3. This invention sets up a model update module to invert the real-time changes in the acoustic properties of the welding area material by utilizing the waveform characteristics of the echo signal, and updates the Jacobian matrix, which is the core of the control model, online accordingly. This mechanism enables the control system to adapt to changes in the physical properties of the material caused by factors such as heat accumulation, ensuring the accuracy of the control model throughout the welding process, and thus improving the control robustness of the system in complex and long-term welding tasks. Attached Figure Description

[0031] Figure 1 This is a system architecture diagram of the present invention;

[0032] Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0033] 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.

[0034] Example:

[0035] Please see the appendix Figure 1 This invention provides a self-adjusting control system for welding parameters of bridge steel components, comprising:

[0036] The acoustic sensing module is used to acquire real-time feedback data reflecting the current geometry of the molten pool, feedforward data reflecting the disturbance information in front of the welding path, and generate echo signals containing waveform morphology features.

[0037] In this embodiment, the acoustic sensing module is the core sensing unit of the control system of the present invention. It is responsible for providing all the necessary real-time status information for the subsequent feedback calculation module, feedforward calculation module and model update module.

[0038] Specifically, the acoustic sensing module preferably employs an acoustic phased array system, which is integrated at the front end of the welding actuator and maintains a preset, non-contact distance from the workpiece to be welded. The acoustic phased array system is used because it can quickly and flexibly control the focusing, emission angle, and scanning path of the sound beam electronically without any mechanical rotating parts, thus enabling the implementation of multiple detection functions within a compact structure.

[0039] To achieve the composite control strategy of this invention, the acoustic sensing module is configured to operate in two basic detection modes, namely feedback mode and feedforward mode, in a time-division or parallel manner within a control cycle.

[0040] In feedback mode, the acoustic sensing module acquires real-time feedback data reflecting the current geometry of the molten pool. To this end, the acoustic phased array system focuses its acoustic beam and emits it vertically or nearly vertically into the region directly below or slightly behind the current welding arc. The emitted high-frequency acoustic pulses propagate through the solid base material, and when they encounter the liquid molten pool formed by the high welding temperature, they are significantly reflected at the interface of abrupt change in acoustic impedance between the solid and liquid metals (i.e., the solid / liquid interface).

[0041] The acoustic sensing module's receiving array captures the echo signal and precisely measures the flight time of the signal received by each array element, assuming discrete time intervals. The complete set of echo time-of-flight information obtained is The processor integrated within or connected to the acoustic sensing module uses a pre-established acoustic reconstruction algorithm. The flight time information set is processed to reconstruct the two-dimensional cross-sectional profile of the current molten pool in the depth and width directions. This process can be represented as:

[0042] ;

[0043] in, Given the known sensor geometry parameters, That is, the output to the feedback calculation module, which is determined by the real-time penetration depth. and melt width The actual geometric vectors formed.

[0044] In feedforward mode, the acoustic sensing module acquires feedforward data reflecting disturbances ahead of the welding path. To this end, the acoustic phased array system utilizes its beam deflection capability to project the probe beam at a forward angle to a distance ahead of the welding path. By rapidly scanning the area to be welded ahead, the system continuously acquires forward-looking echo signals. When geometric discontinuities exist on the welding path, such as a sudden widening of the bevel gap, misalignment, or incomplete spot welds, the echo characteristics will exhibit identifiable changes compared to the reference echo characteristics of a smooth, uniform path.

[0045] The acoustic sensing module will acquire the look-ahead echo feature vector in real time. With a pre-calibrated reference feature vector Compare the differences It is fed into a perturbation mapping model The function of this model is to interpret abstract changes in echo characteristics into concrete, physically meaningful perturbations. This process can be expressed as:

[0046] ;

[0047] in, This is the disturbance vector that is output to the feedforward calculation module, representing the type and magnitude of the disturbance ahead.

[0048] Furthermore, to support the functionality of the model update module, the acoustic sensing module not only measures the time of flight of the echo, but is also configured to capture and analyze the waveform morphology characteristics of the complete returning acoustic pulse. These characteristics include, but are not limited to, the amplitude attenuation of the echo signal, the shift in the center frequency, and the degree of waveform dispersion.

[0049] It should be noted that these waveform characteristics are introduced because the propagation speed and attenuation coefficient of sound waves in high-temperature metals are closely related to the material's real-time temperature, grain structure, and other thermophysical states. Therefore, analyzing these characteristics can indirectly deduce the average acoustic properties of the material along the sound wave propagation path. The acoustic sensing module provides these echo signals containing waveform characteristics or extracted feature parameters to the model update module as the physical basis for its online update or calibration of the Jacobian matrix, thereby enabling the entire control system to adapt to changes in workpiece material properties caused by welding heat accumulation.

[0050] The feedback calculation module, connected to the acoustic sensing module, is used to calculate the feedback adjustment amount based on the deviation between real-time feedback data and the preset target geometric state, and based on a Jacobian matrix.

[0051] In this embodiment, the feedback calculation module is the core functional unit for implementing closed-loop control in the control system. Its main responsibility is to calculate the welding parameter adjustment amount with compensatory effect based on the geometric deviation of the molten pool that has occurred. It is a key link in the system to make real-time corrections to the current welding state.

[0052] Specifically, the input information for the feedback calculation module comes from two sources: first, real-time feedback data from the acoustic sensing module that characterizes the current geometry of the molten pool, i.e., the actual geometric vector. Secondly, it refers to the preset target geometric state, which serves as the welding process objective, as defined in the system; that is, the target geometric vector. .

[0053] The first step in the feedback calculation module is to determine the deviation between the current welding state and the target state. To do this, the module performs a vector subtraction operation between the actual geometric vector and the target geometric vector, thereby obtaining a geometric error vector that quantifies the current penetration depth and weld width errors. The calculation process is shown in the following formula:

[0054] ;

[0055] This geometric error vector is the direct basis for all subsequent feedback control calculations.

[0056] After obtaining the geometric error vector, the core technical challenge for the feedback calculation module is how to convert the error in one geometric dimension (usually in millimeters) into an adjustment amount (in units such as amperes, volts, or millimeters per second) in one or more welding process parameter dimensions. It should be noted that the effects of various welding parameters (such as current, voltage, and speed) on the weld pool geometry (weld depth and weld width) are interdependent; that is, a change in a single parameter often causes changes in multiple geometric dimensions simultaneously.

[0057] To solve this technical problem, the feedback calculation module introduces a Jacobian matrix. As a locally linearized dynamic model of the system, the Jacobian matrix describes the linear mapping between small changes in the welding parameter vector and the resulting changes in the molten pool geometry vector near the current welding point. It is worth noting that the Jacobian matrix used by this feedback calculation module... It continuously receives online updates or calibration results from the model update module, thereby ensuring the real-time accuracy of the model.

[0058] Based on this model, the task of the feedback calculation module is to solve an inverse control problem: to find a feedback adjustment amount. This adjustment amount allows the adjustment to produce a value that is consistent with the current geometric error vector. Geometric transformations of equal magnitude but opposite direction. To achieve stable solutions and decoupling control of the coupled system, this embodiment preferably employs the pseudo-inverse of the Jacobian matrix. To perform the calculations.

[0059] Finally, the feedback calculation module outputs the feedback adjustment amount to the fusion module. It is determined by the following relationship:

[0060] ;

[0061] In this relation It is the pseudo-inverse of the Jacobian matrix, which can provide a stable solution to the inverse problem; The geometric error vector obtained from the aforementioned calculation; and This is a proportional gain diagonal matrix that can be preset or adjusted online. The purpose of introducing this gain matrix is ​​to provide control system designers with a means to adjust the feedback control response characteristics, for example, to independently adjust the system's response speed and intensity to melt depth error and melt width error.

[0062] After completing the above calculations, the feedback calculation module will obtain the feedback adjustment amount. It is passed to the fusion module for further fusion processing with the feedforward adjustment.

[0063] The feedforward calculation module, connected to the acoustic sensing module, is used to calculate the feedforward adjustment amount based on the feedforward data.

[0064] In this embodiment, the feedforward calculation module is a functional unit in the control system that implements active disturbance suppression. Its unique feature is that it does not respond to process deviations that have already occurred, but rather performs forward-looking control actions based on future predictions, aiming to reduce or eliminate the impact of external disturbances on the stability of the welding process from the root.

[0065] Specifically, the input information to the feedforward calculation module comes solely from the feedforward data detected and generated by the acoustic sensing module in feedforward mode, which reflects the disturbance information ahead of the welding path. This feedforward data is processed to form a disturbance vector. This vector quantifies the type and extent of geometric discontinuities present in the upcoming welding path.

[0066] Upon receiving this disturbance vector, the primary task of the feedforward calculation module is to convert a physical disturbance (e.g., the number of millimeters by which the forward bevel gap will widen) into a desired impact on the geometry of the molten pool. Directly controlling the physical disturbance is difficult because the ultimate control objective of the system is the geometry of the molten pool.

[0067] To achieve this transformation, the feedforward computing module integrates or invokes a disturbance effect model. This model, based on welding physics principles or calibrated using extensive experimental data, functions to describe how specific external disturbances will affect the final molten pool geometry. Using this model, the module can calculate the impact of the disturbance vector on the final molten pool geometry without any control applied. The predicted geometric deviation will be caused by the disturbance. .

[0068] The core purpose of the feedforward calculation module is to calculate a feedforward adjustment amount. When this adjustment is applied to the welding parameters, the resulting geometric change precisely offsets the aforementioned expected geometric deviation. This is a typical control inverse problem.

[0069] To solve this problem, the feedforward computation module, similar to the feedback computation module, also utilizes the Jacobian matrix maintained in real time by the model update module. This Jacobian matrix establishes the relationship between changes in welding parameters and changes in the geometry of the molten pool, thus providing a mathematical basis for calculating the required parameter adjustments.

[0070] Based on this, the feedforward calculation module determines the feedforward adjustment amount to be finally output to the fusion module by solving the following relationship.

[0071] ;

[0072] In this relationship, the negative sign indicates that the adjustment amount is used to counteract the effects of the disturbance; It is the pseudo-inverse of the Jacobian matrix, used to stably solve this control inverse problem; The product of these is the expected geometric deviation obtained from the aforementioned calculation; and This is a feedforward gain matrix that can be preset or adjusted online. Its function is to adjust the strength of the feedforward control to adapt to different disturbance types and welding requirements.

[0073] After completing the above calculations, the feedforward calculation module will obtain a forward-looking feedforward adjustment amount. The data is passed to the fusion module, where it is finally fused with the feedback adjustment from the feedback calculation module to form a general adjustment command for the welding parameters.

[0074] The model update module, connected to the acoustic sensing module, is used to analyze the waveform morphology characteristics of the echo signal to estimate the acoustic characteristic parameters of the welding area in real time, and to update or calibrate the Jacobian matrix online based on the acoustic characteristic parameters.

[0075] In this embodiment, the model update module is a key component in the control system to achieve adaptive functionality. Its purpose is to address the technical problem that during the welding process, the continuous input and accumulation of heat causes dynamic changes in the thermophysical properties of the workpiece and its surrounding area, leading to the gradual inaccuracy of the initially calibrated system control model.

[0076] Specifically, the model update module does not directly participate in each control quantity calculation. Instead, it provides the feedback calculation module and the feedforward calculation module with a continuously accurate system model that matches the current physical state of the workpiece, namely the Jacobian matrix. .

[0077] The input information for the model update module is the unsimplified, complete echo signal captured by the acoustic sensing module. Unlike the feedback calculation module, which only utilizes time-of-flight information, this module focuses on analyzing the waveform characteristics of the echo signal. Preferably, these characteristics include, but are not limited to: the amplitude attenuation of the echo signal, which reflects the energy loss of the sound wave as it propagates in the medium; the center frequency shift of the echo signal, which is related to the nonlinear acoustic properties of the medium; and the waveform dispersion of the echo signal, which characterizes the differences in the propagation speed of sound waves of different frequency components.

[0078] The analysis of these waveform characteristics is based on the following physical principle: the propagation characteristics of sound waves in metallic media, especially the sound velocity and attenuation coefficient, are functions of material temperature, grain structure, and stress state. During welding, the drastic temperature gradient inevitably causes significant changes in these material properties in time and space. These changes are directly "encoded" in the morphology of the returned acoustic echo signal.

[0079] Therefore, the first step executed by the model update module is to extract the quantitative indicators of the waveform morphology features mentioned above from the continuous echo signals in real time.

[0080] Subsequently, the module uses a pre-established physical correlation model to reverse-engineer the equivalent acoustic properties of the material along the current sound wave propagation path, such as the real-time average sound velocity or sound attenuation coefficient, based on these extracted feature indicators.

[0081] The core function of the model update module is to update these real-time acoustic characteristic parameters with the Jacobian matrix, which is the core of the control model. The elements are associated. It should be noted that each element in the Jacobian matrix (e.g., ...) The physical nature of welding is determined by complex physical processes such as heat conduction and fluid flow in the welding area, and these processes are directly controlled by the thermophysical properties of the material (such as thermal conductivity and specific heat capacity).

[0082] The model update module internally establishes a mapping relationship between acoustic characteristic parameters and material thermophysical properties. Through this mapping, the module can indirectly estimate the changing trends of material thermophysical properties based on real-time changes in acoustic characteristics. Based on this, the module adjusts the Jacobian matrix... The elements in the matrix are updated or calibrated online to generate a Jacobian matrix that changes over time and better reflects the dynamic characteristics of the current system. .

[0083] Finally, the model update module will use this real-time calibrated Jacobian matrix. This updated model is simultaneously provided to both the feedback and feedforward calculation modules. These modules will use this updated model, rather than a fixed initial model, in their subsequent calculations. In this way, the control system of this invention achieves continuous self-calibration of its core model, ensuring the continued effectiveness of the control law throughout long-duration or high-heat-input welding tasks.

[0084] The fusion module, connected to the feedback calculation module and the feedforward calculation module, is used to fuse the feedback adjustment amount and the feedforward adjustment amount to generate the total adjustment amount of the welding parameters.

[0085] In this embodiment, the fusion module plays the role of the final convergence point for control decisions in the overall architecture of the control system. Its necessity lies in the fact that the control strategy of this invention generates two control components with different properties and complementary objectives in parallel, thus requiring a specific functional unit to integrate these components into a single, executable final control command.

[0086] Specifically, the input signal of the fusion module includes two independent control adjustment vectors: one is the feedback adjustment from the feedback calculation module. The adjustment amount is based on the response to the deviation between historical and current states, and has the nature of hysteresis compensation; secondly, it is the feedforward adjustment amount from the feedforward calculation module. This adjustment is based on predictions of future disturbances and has the property of anticipatory suppression.

[0087] The core function of the fusion module is to combine these two separate adjustment components, one for the past and one for the future, into a unified, immediately effective total adjustment for welding parameters. The total adjustment must simultaneously reflect the intention to correct current errors and to offset future disturbances.

[0088] In a preferred embodiment, the fusion process is achieved by algebraically summing the two adjustment vectors. This operation ensures that the effects of the two control components are linearly superimposed, working together on the final control output. The calculations performed by the fusion module follow the following relationship:

[0089] ;

[0090] In this way, the final total adjustment amount is generated. As a comprehensive control decision, its inherent logic includes both correcting existing deviations and proactively defending against impending disturbances. This fusion mechanism enables the entire control system to possess both the robustness of feedback control and the speed of feedforward control.

[0091] After completing the fusion calculation, the fusion module will finally generate the total adjustment amount. As a complete, multi-dimensional parameter adjustment command (each component of which corresponds to the adjustment value of welding current, arc voltage, welding speed, etc.), it is output to the control execution module so that it can make the final adjustment of the actual operating parameters of the welding equipment.

[0092] The control execution module, connected to the fusion module, is used to adjust the welding parameters of the welding equipment according to the total adjustment amount.

[0093] In this embodiment, the control execution module is the final interface between the control system and the physical welding equipment. Its function is to convert the abstract control commands generated by the upstream computing module and fused into hardware settings that can be directly executed by the physical equipment. This module is the final link in the physical implementation of the entire closed-loop control strategy.

[0094] Specifically, the input information to the control execution module comes from the fusion module, which is a total adjustment that integrates feedback and feedforward control intentions. It should be noted that this input vector represents an incremental "adjustment" to the current welding parameters, rather than an absolute "set value".

[0095] Therefore, the first step in the control execution module is to apply this incremental adjustment to the current welding parameter reference. To this end, the module maintains the current control cycle internally. Welding parameter vector This vector represents the actual operating parameters of the welding equipment at this moment. Upon receiving the total adjustment... Then, the module calculates the next control cycle through vector addition. Target welding parameter vector The update process follows the following relation:

[0096] ;

[0097] The welding parameter vector preferably includes welding current. Arc voltage and welding speed Core process parameters, etc.

[0098] Furthermore, to ensure physical feasibility and protect the welding equipment, preventing equipment damage or process failure due to control algorithm outputs exceeding the range, the control execution module also includes a saturation processing step. The function of this step is to convert the target welding parameter vector calculated in the previous step... Each component corresponds to a preset upper limit threshold for welding parameters, defined by the physical limits of the equipment or the welding process specifications. and lower threshold Compare them.

[0099] If any calculated parameter component exceeds its corresponding upper or lower limit, the module will perform a limiting process, replacing the exceeding calculated value with the corresponding upper or lower threshold. This ensures that the final instructions output to the hardware are always within a safe and permissible operating range.

[0100] After completing the above update calculations and saturation processing, the control execution module will finally determine the absolute and physically feasible target welding parameter vector. It is converted into a communication protocol and data format that can be recognized by specific hardware.

[0101] Finally, the module sends these instructions to the welding power controller to set new current and voltage values ​​and to the robot motion controller to set new welding speed through the corresponding communication interface, thereby completing the dynamic adjustment of the welding process within a control cycle.

[0102] Please see the appendix Figure 2 The self-adjustment control method for welding parameters of bridge steel components includes the following steps:

[0103] S1. Through the acoustic sensing module, real-time feedback data reflecting the current geometry of the molten pool, feedforward data reflecting the disturbance information in front of the welding path, and echo signals containing waveform morphology characteristics are acquired.

[0104] S2. Through the feedback calculation module, the feedback adjustment amount is calculated based on the deviation between the real-time feedback data and the preset target geometric state and a Jacobian matrix.

[0105] S3. Calculate the feedforward adjustment amount based on the feedforward data using the feedforward calculation module.

[0106] S4. Through the model update module, analyze the waveform characteristics of the echo signal to update or calibrate the Jacobian matrix online.

[0107] S5. The feedback adjustment amount and the feedforward adjustment amount are fused through the fusion module to generate the total adjustment amount of the welding parameters;

[0108] S6. Based on the total adjustment amount, adjust the welding parameters of the welding equipment through the control execution module.

Claims

1. A self-adjusting control system for welding parameters of bridge steel components, characterized in that, include: The acoustic sensing module is used to acquire real-time feedback data reflecting the current geometry of the molten pool, feedforward data reflecting the disturbance information in front of the welding path, and generate echo signals containing waveform morphology features. The feedback calculation module, connected to the acoustic sensing module, is used to calculate the feedback adjustment amount based on the deviation between the real-time feedback data and the preset target geometric state, and based on a Jacobian matrix. A feedforward calculation module, connected to the acoustic sensing module, is used to calculate the feedforward adjustment amount based on the feedforward data; The model update module, connected to the acoustic sensing module, is used to analyze the waveform morphology characteristics of the echo signal to estimate the acoustic characteristic parameters of the welding area in real time, and to update or calibrate the Jacobian matrix online based on the acoustic characteristic parameters. A fusion module, connected to the feedback calculation module and the feedforward calculation module, is used to fuse the feedback adjustment amount and the feedforward adjustment amount to generate a total adjustment amount for the welding parameters; A control execution module, connected to the fusion module, is used to adjust the welding parameters of the welding equipment according to the total adjustment amount; The acoustic sensing module includes an acoustic phased array system, which is configured as follows: In feedback mode, an acoustic pulse is emitted to the current welding area, and the penetration depth and weld width, which are the real-time feedback data, are calculated based on the echo signal of the solid-liquid interface. In feedforward mode, the acoustic beam is deflected in front of the welding path, and feedforward data characterizing the bevel gap or misalignment is generated based on the detected changes in echo characteristics. The feedback calculation module compares the actual geometric vector formed by the real-time feedback data with the target geometric vector formed by the preset target geometric state to obtain a geometric error vector, and performs inverse operation on the geometric error vector to obtain the feedback adjustment amount. The feedback adjustment amount The calculation follows the following relationship: ； in, The pseudo-inverse of the Jacobian matrix is ​​given. Let be the geometric error vector. This is the proportional gain matrix; The feedforward calculation module operates in the following ways: Based on a perturbation effect model, the feedforward data is converted into the expected geometric deviation; Based on the Jacobian matrix, a parameter adjustment amount that can offset the expected geometric deviation is calculated as the feedforward adjustment amount; The feedforward adjustment amount The calculation follows the following relationship: ； in, The pseudo-inverse of the Jacobian matrix is ​​given. This is the perturbation vector generated based on the feedforward data. This is the disturbance impact model. This is the feedforward gain matrix; The waveform morphology characteristics of the echo signal analyzed by the model update module include at least one of signal amplitude attenuation, frequency shift, and waveform dispersion. The fusion module generates the total adjustment by algebraically summing the feedback adjustment and the feedforward adjustment.

2. The bridge steel component welding parameter self-adjustment control system according to claim 1, characterized in that, The welding parameters adjusted by the control execution module include at least one of welding current, arc voltage, and welding speed.

3. A method for self-adjusting control of welding parameters for bridge steel components, as described in any one of claims 1-2, characterized in that, Includes the following steps: S1. Through the acoustic sensing module, real-time feedback data reflecting the current geometry of the molten pool, feedforward data reflecting the disturbance information in front of the welding path, and echo signals containing waveform morphology characteristics are acquired. S2. Through the feedback calculation module, the feedback adjustment amount is calculated based on the deviation between the real-time feedback data and the preset target geometric state and on a Jacobian matrix. S3. Calculate the feedforward adjustment amount based on the feedforward data using the feedforward calculation module; S4. Analyze the waveform characteristics of the echo signal through the model update module to update or calibrate the Jacobian matrix online. S5. The feedback adjustment amount and the feedforward adjustment amount are fused together through the fusion module to generate the total adjustment amount of the welding parameters; S6. Based on the total adjustment amount, adjust the welding parameters of the welding equipment through the control execution module.

Images