A multi-source information fusion intelligent welding robot system for high wind-sand environment

Abstract

The application discloses a multi-source information fusion intelligent welding robot system for high wind-sand environment. A bearing platform is used for realizing stable bearing and pose output of an execution module and an intelligent protection cabin in a high wind-sand operation area; the execution module drives a welding tool assembly through a multi-joint mechanical arm to complete welding operation; the intelligent protection cabin adopts a flexible air flow supporting structure with adjustable air pressure, dynamically adjusts the air flow state in the cabin according to a protection adjustment instruction, so as to reduce the influence of wind-sand disturbance; a collection module is used for acquiring environmental information; a fusion module performs time alignment and feature mapping on environmental disturbance characteristic data, welding process state data and pose information, and generates a welding parameter adjustment instruction, a welding pose correction instruction and a protection adjustment instruction; a cooperation module realizes remote transmission and decision cooperation of welding process data. The application can realize adaptive adjustment and stable control of the welding process in the high wind-sand environment, and improves the welding seam quality and system reliability.

Description

Technical Field

[0001] This invention relates to the field of welding control technology, and in particular to a multi-source information fusion intelligent welding robot system for high wind and sand environments. Background Technology

[0002] In existing technologies, welding robots have been widely used in steel structure manufacturing, pipeline construction, and engineering equipment assembly. These systems typically consist of a load-bearing or mobile platform, a robotic arm-type welding actuator, a welding power source, and a basic control unit. They achieve automatic or semi-automatic welding operations by pre-setting welding trajectories and process parameters. For complex working environments, some existing technologies have incorporated environmental monitoring devices or protective structures to detect environmental factors such as wind speed, temperature, or dust. Alternatively, simple protective covers or windbreaks can be used to reduce the impact of the external environment on the welding process, thereby improving the stability and applicability of welding operations to some extent.

[0003] However, the aforementioned existing technologies mostly focus on the detection of single environmental parameters or static protection measures. There is a lack of unified time correlation and collaborative processing mechanism between environmental perception data and welding process status, making it difficult to effectively predict and dynamically respond to rapidly changing environmental disturbances such as wind and sand. Furthermore, welding parameter adjustment, protection status adjustment and welding posture control are often independent of each other, resulting in significant fluctuations in welding quality and insufficient system adaptability and reliability under extreme working conditions such as high wind and sand.

[0004] Therefore, it is necessary to propose a new technical solution to meet the higher requirements for the stability and intelligent control of welding operations in high wind and sand environments. Summary of the Invention

[0005] This application provides a multi-source information fusion intelligent welding robot system for high wind and sand environments, which enables adaptive adjustment and stable control of the welding process in high wind and sand environments, thereby improving weld quality and system reliability.

[0006] This application provides a multi-source information fusion intelligent welding robot system for high-wind and sandy environments, including:

[0007] The platform is used to carry and position the execution module and intelligent protective cabin in high-wind and sandstorm operation areas, and output position and posture information;

[0008] The execution module includes a multi-joint robotic arm, a welding tool assembly, and an arc flame stabilization device. The multi-joint robotic arm is used to adjust the welding posture according to the welding trajectory instructions, the welding tool assembly is used to perform the actual welding operation, and the arc flame stabilization device is used to collect the current waveform information and arc flame state characteristic information generated during the welding process.

[0009] The intelligent protective cabin is installed on the outside of the execution module and adopts a flexible airflow support structure with adjustable air pressure. It is used to adjust the airflow direction, airflow speed and local wind pressure distribution inside the cabin after receiving the protection adjustment command.

[0010] The data acquisition module is used to collect wind speed, temperature and humidity, dust density, and environmental acoustic information in high-wind and sandy environments.

[0011] The fusion module is used to perform unified time alignment and feature mapping on environmental disturbance feature data, welding process status data and pose information, and generate welding parameter adjustment instructions, welding posture correction instructions and protection adjustment instructions based on time-series feature modeling and state prediction mechanism.

[0012] The collaboration module is used to receive video data of the welding process, environmental disturbance characteristic data, and welding parameter log data, and send remote decision information to the fusion module.

[0013] The beneficial effects of this application mainly include: (1) By unifying the time alignment and feature mapping of environmental disturbance feature data, welding process status data and pose information, and performing fusion analysis on this basis, the welding control no longer relies on a single sensor signal or static parameter setting, but can comprehensively reflect the real-time impact of the high wind and sand environment on the welding process, thereby significantly improving the integrity and accuracy of the welding process status perception. (2) By using the fusion module to generate welding parameter adjustment instructions and welding posture correction instructions based on the time-series feature modeling and state prediction mechanism, the system can adaptively adjust the welding power, welding rhythm and welding posture before or during environmental disturbance, effectively reducing the impact of wind speed fluctuations, dust disturbances and other factors on arc flame stability and weld formation quality. (3) The intelligent protective cabin adopts a flexible airflow support structure with adjustable air pressure and forms a linkage control relationship with the fusion module, so that the airflow direction, airflow speed and local wind pressure distribution of the protective cabin can be dynamically adjusted according to environmental changes, thereby achieving coordinated matching between the protective state and welding requirements without affecting the welding operation space, and improving the system's continuous operation capability in a high wind and sand environment. (4) The remote transmission and decision-making collaboration of welding process video data, environmental disturbance characteristic data and welding parameter log data are realized through the collaborative module, so that the welding robot system has the ability of remote monitoring, strategy optimization and safety joint control, which is conducive to improving the safety of operation, operation and maintenance efficiency and overall system reliability in complex or dangerous operating environments. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of a multi-source information fusion intelligent welding robot system for high wind and sand environments provided in the first embodiment of this application. Detailed Implementation

[0015] Many specific details are set forth in the following description to provide a full understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application; therefore, this application is not limited to the specific embodiments disclosed below.

[0016] The first embodiment of this application provides a multi-source information fusion intelligent welding robot system for high-wind and sandy environments. Please refer to... Figure 1 This figure is a schematic diagram of the first embodiment of this application. The following is in conjunction with... Figure 1 The first embodiment of this application provides a detailed description of a multi-source information fusion intelligent welding robot system for high-wind and sandy environments.

[0017] The multi-source information fusion intelligent welding robot system for high wind and sand environments includes a support platform 101, an execution module 102, an intelligent protective cabin 103, a data acquisition module 104, a fusion module 105, and a collaboration module 106.

[0018] The carrier platform 101 is used to carry and position the execution module and intelligent protective cabin in high wind and sand operation areas, and output position and posture information.

[0019] The support platform 101 is the fundamental unit for the overall stable operation and precise control of this system in high-wind and sandstorm working environments. Its main function is to provide reliable physical support, spatial positioning, and continuous and usable posture information output for the execution module 102 and the intelligent protective cabin 103 under complex ground conditions and strong wind and sand disturbances. High-wind and sandstorm working areas refer to outdoor or semi-open industrial environments with frequent wind speed changes, high concentrations of suspended particulate matter in the air, and unstable ground flatness and adhesion conditions, such as steel structure construction sites in desert areas, open-air pipeline welding areas, or equipment maintenance areas in mining areas. In such environments, traditional fixed or simply mobile welding platforms cannot guarantee the spatial stability and posture controllability of the welding equipment. Therefore, the support platform 101 in this invention is designed as a comprehensive support unit with active positioning and posture perception capabilities.

[0020] The support platform 101 structurally includes a chassis frame, a walking or support mechanism, an attitude detection component, and a positioning reference component. The chassis frame supports the overall weight of the execution module 102 and the intelligent protective cabin 103, and serves as the mounting base for each functional component. The walking or support mechanism can be a tracked walking mechanism, a wheeled walking mechanism, or a deployable multi-point support mechanism, depending on the actual application scenario, to adapt to soft sand, gravel ground, or uneven construction sites. During implementation, if a tracked structure is used, the pressure per unit area can be reduced by increasing the ground contact area, thereby reducing the risk of platform sinking in windy and sandy environments. If a multi-point support mechanism is used, the platform can automatically level itself on sloping ground by adjusting the height of each support point.

[0021] The platform 101 also integrates an attitude detection component to acquire the platform's attitude information in space in real time. This attitude information includes at least the platform's three-axis position parameters relative to the ground coordinate system and attitude angle parameters around each axis. Specifically, it can include the platform's displacement in the forward / backward, left / right, and vertical directions, as well as pitch, roll, and yaw angles. The attitude detection component can be composed of an inertial measurement unit, a tilt sensor, and an encoder. The inertial measurement unit senses the transient acceleration and angular velocity changes of the platform under wind or ground vibration conditions. The tilt sensor measures the platform's tilt relative to the direction of gravity, and the encoder acquires displacement feedback information from the walking mechanism or support mechanism. By fusing the above sensor signals, stable and continuous pose information can be output.

[0022] In this invention, "pose information" refers to a set of data that can completely describe the position and attitude state of the support platform 101 in space, including both position parameters and attitude parameters. This pose information is output to the fusion module 105 as basic reference data for time alignment and feature mapping with welding process state data and environmental disturbance feature data. For example, when the support platform 101 tilts slightly under the action of gusts of wind, its output attitude angle change will be identified by the fusion module 105 as a basic disturbance factor that may affect the accuracy of the welding trajectory, thereby providing the necessary input basis for the subsequent generation of welding attitude correction commands.

[0023] Furthermore, the support platform 101 is also used for spatial relative positioning of the execution module 102 and the intelligent protective cabin 103. Relative positioning refers to defining the installation position, installation height, and spatial relationship between the execution module 102 and the intelligent protective cabin 103 within the platform's own coordinate system, thereby ensuring that the geometric relationships between the modules remain consistent when the platform moves or changes its attitude. In actual implementation, this relative positioning relationship can be determined through calibration during the system initialization phase and dynamically updated during operation based on the platform's pose information.

[0024] Through the above structural and functional design, the support platform 101 not only plays a simple supporting role, but also constitutes the spatial reference unit and motion reference unit of the entire system. Its output pose information is directly used by the subsequent fusion module 105, thereby affecting the welding posture correction, protection adjustment and overall welding stability control process, thus ensuring the reliable deployment and stable operation of the intelligent welding robot system based on this support platform in high wind and sand environment.

[0025] The execution module 102 includes a multi-joint robotic arm, a welding tool assembly, and an arc flame stabilization device. The multi-joint robotic arm is used to adjust the welding posture according to the welding trajectory command, the welding tool assembly is used to perform the actual welding operation, and the arc flame stabilization device is used to collect the current waveform information and arc flame state characteristic information generated during the welding process.

[0026] The execution module 102 is the core operating unit in this invention that directly completes the welding operation and generates welding process status data. In high-wind and sandy environments, it not only performs welding actions but also undertakes the tasks of welding process status perception and stability assessment, thereby providing a reliable and quantifiable data foundation for the subsequent analysis and control of the fusion module 105. The execution module 102 is installed on the support platform 101 and located within the enclosed space of the intelligent protective cabin 103. Its structural and functional design must simultaneously meet the requirements of high-precision welding operation and adaptability to complex environments.

[0027] The execution module 102 includes a multi-joint robotic arm, a welding tool assembly, and an arc flame stabilization device. The multi-joint robotic arm is an actuator used to achieve welding trajectory following and welding posture adjustment. A multi-joint robotic arm refers to a mechanical motion structure composed of multiple sequentially connected rotary or swing joints. Each joint can change angle around a corresponding axis under control commands, thereby enabling continuous and controllable movement of the robotic arm's end effector in three-dimensional space. In this invention, the number of degrees of freedom of the multi-joint robotic arm can be set according to the shape of the workpiece and the spatial distribution of the weld seam, for example, using a six-degree-of-freedom or seven-degree-of-freedom structure to ensure that the welding tool assembly can maintain a suitable incident angle and welding posture even under complex weld seam paths.

[0028] During operation, the multi-joint robotic arm receives welding trajectory instructions and welding posture correction instructions generated by the fusion module 105. The welding trajectory instructions describe the desired motion path of the welding tool assembly's end effector in space, while the welding posture correction instructions fine-tune the original welding posture when environmental disturbances or platform posture changes are detected. Based on these instructions, the robotic arm controller converts the desired end effector pose into target angles or displacements for each joint. This conversion process can be achieved through inverse kinematics calculation. Inverse kinematics calculation refers to the process of solving for the motion parameters of each joint given the robotic arm's structural parameters and the target end effector pose. For example, in a six-DOF robotic arm, by establishing a joint coordinate system and link parameter model, the rotation angles of each joint are calculated, ensuring that the welding tool assembly's end effector reaches the specified position and maintains the specified orientation, thereby ensuring the repeatability and accuracy of the welding process.

[0029] The welding tool assembly, mounted at the end of a multi-joint robotic arm, is the actuator that directly interacts with the workpiece to form a weld. Depending on the application, the welding tool assembly can be selected from arc welding torches, gas shielded welding torches, or other welding tools suitable for high-wind, sandy environments. It includes at least a welding electrode, a shielding gas channel, and a welding current inlet. During welding, the welding tool assembly outputs welding energy according to preset welding parameters and moves along the weld path under the drive of the multi-joint robotic arm, thus achieving continuous welding.

[0030] An arc flame stabilization device is positioned near the welding tool assembly to monitor the arc flame's operating status in real time during welding. The device includes at least a current acquisition unit and an arc flame status sensing unit. The current acquisition unit collects current waveform information from the welding circuit. Current waveform information refers to a continuous data sequence of welding current changes over time, reflecting the stability of energy input during welding. For example, in a stable welding state, the current waveform exhibits periodic or relatively stable changes, while under wind and sand disturbances or when the welding torch's posture shifts, the current waveform may show abrupt changes, increased fluctuation amplitude, or abnormal spikes. The arc flame status sensing unit can employ optical sensors or plasma characteristic detection devices to acquire arc flame brightness, arc length changes, or arc flame vibration characteristics, thereby forming arc flame status characteristic information.

[0031] In this invention, "arc flame state characteristic information" refers to a set of parameters that characterize the stability of the welding arc flame and the continuity of the welding process. This can include the arc flame brightness change rate, arc flame flickering frequency, or arc flame duration, etc. Taking the arc flame brightness change rate as an example, it can be obtained by differential calculation of the arc flame brightness sampling values ​​per unit time. For example, if the brightness values ​​measured at two adjacent sampling times are L1 and L2 respectively, the brightness change rate can be expressed as |L2-L1|. The larger this change rate, the worse the arc flame stability. By continuously collecting the above characteristic information, a welding process state data sequence can be formed.

[0032] The arc flame stabilization device outputs the collected current waveform information and arc flame state characteristic information to the fusion module 105 in real time, serving as an important component of the welding process state data. In subsequent processing, the fusion module 105 performs unified time alignment and comprehensive analysis on the aforementioned welding process state data, environmental disturbance characteristic data, and pose information output by the support platform 101. This determines whether the current welding state is affected by the high wind and sand environment and generates welding parameter adjustment commands or welding posture correction commands accordingly.

[0033] Through the above structural and functional design, the execution module 102 can not only complete high-precision welding operations, but also continuously output process status information directly related to welding stability during the welding process. Thus, intelligent welding operations that are perceptible, adjustable, and controllable can be realized based on the execution module in high wind and sand environments.

[0034] The intelligent protective cabin 103 is installed on the outside of the execution module and adopts a flexible airflow support structure with adjustable air pressure. It is used to adjust the airflow direction, airflow speed and local wind pressure distribution inside the cabin after receiving the protection adjustment command.

[0035] The intelligent protective cabin 103 is a key protection and adjustment unit specifically designed for the stability and safety of welding operations in high-wind and sandy environments. Its core function is not merely to passively shield the execution module 102, but rather to actively participate in weakening, reconstructing, and guiding environmental interference during the welding process through a flexible airflow support structure with adjustable air pressure. This provides a relatively stable and controllable local microenvironment for the welding arc and welding posture. The intelligent protective cabin 103 is entirely enclosed on the outside of the execution module 102, forming a stable relative installation relationship with the support platform 101. This ensures that the protective cabin and the execution module maintain a consistent spatial coordination relationship even when the platform's position changes.

[0036] The term "enclosed structure" refers to the intelligent protective cabin 103 structurally covering at least the outer space of the welding tool assembly and the arc flame formation area, but not requiring complete sealing. This enclosed structure can be semi-enclosed or multi-opening, allowing the welding tool assembly to complete welding operations without physical obstruction, while simultaneously intervening in external wind and sand entering the space through airflow regulation. The main body of the intelligent protective cabin 103 can be constructed from a combination of flexible materials and rigid support components. The flexible materials form an airflow support cavity that deforms with changes in air pressure, while the rigid support components define the overall outline of the protective cabin and ensure structural stability.

[0037] The intelligent protective cabin 103 employs an adjustable air pressure flexible airflow support structure. This refers to the arrangement of multiple airflow channels and air pressure regulating units within the cabin or around its perimeter. By inputting gases of different pressures and flow rates into the airflow channels at different locations, a directional and gradient airflow field is formed within the cabin. Here, "air pressure" refers to the pressure value of the gas within the airflow channel relative to the external environment, and "adjustable" means that this pressure value can be continuously or progressively varied within a certain range according to control commands. For example, when the external wind speed is high and the wind direction is towards the welding arc flame area, the output air pressure of the windward airflow channel can be increased, creating a reverse or deflected airflow inside, thereby counteracting the direct impact of the external wind on the arc flame.

[0038] In this invention, the flexible airflow support structure specifically refers to an airflow cavity formed by flexible materials, which, when inflated, achieves a certain shape and rigidity. The structural form and airflow distribution are simultaneously adjusted through changes in internal air pressure. Compared to traditional rigid protective covers, this structure can change its protective form in real time according to environmental changes without increasing the load on the execution module. For example, when a significant increase in dust density is detected, the airflow velocity in the airflow channel below the protective chamber can be increased, guiding the dust outwards from the chamber rather than causing it to remain in the welding area.

[0039] During operation, the intelligent protective cabin 103 receives protective adjustment commands generated by the fusion module 105. These commands can include airflow direction adjustment parameters, airflow velocity adjustment parameters, and local wind pressure distribution adjustment parameters. Airflow direction refers to the dominant flow direction of airflow within the protective cabin, such as upward guidance, lateral deflection, or the formation of swirling airflow. Airflow velocity refers to the speed at which gas flows through the airflow channel per unit time. Local wind pressure distribution refers to the pressure differences formed in different areas of the protective cabin. Through the coordinated control of these parameters, an airflow barrier can be constructed within the protective cabin to address specific wind and sand disturbance conditions.

[0040] For example, when the acquisition module 104 detects an external wind speed of 8 meters per second, with the wind direction roughly aligned with the welding direction, and a significant increase in dust density, the fusion module 105 can generate a protective adjustment command. This command increases the output air pressure of the windward airflow channel to, for example, 1.5 times that under normal operating conditions, while maintaining a lower air pressure on the leeward side. This creates a directional airflow within the protective chamber, pointing from the windward side to the leeward side. This directional airflow counteracts the disturbance of the welding arc flame caused by the external wind speed and carries sand and dust particles entering the protective chamber away from the welding area, achieving a synergistic effect of protection and sand removal.

[0041] While adjusting the airflow state, the intelligent protective cabin 103 can also feed back its current protective status information to the fusion module 105. The protective status information may include the real-time air pressure value, airflow velocity value, and airflow stability evaluation parameters inside the protective cabin for each airflow channel. Through this feedback mechanism, the fusion module 105 can determine whether the current protective strategy has achieved the expected effect and correct the protective adjustment command when necessary, thereby forming a closed-loop adjustment process.

[0042] Through the above structure and working method, the intelligent protective cabin 103 can not only effectively reduce the direct impact of external wind speed and sand on the welding arc flame and weld formation in high wind and sand environments, but also form a synergistic effect with welding posture adjustment and welding parameter adjustment, so that the welding process is always in a controllable and stable local environment, thereby significantly improving the system's continuous operation capability and welding quality stability under extreme working conditions.

[0043] Furthermore, the intelligent protective cabin is specifically used for:

[0044] The system receives protection adjustment commands output by the fusion module, and parses the target airflow coordination parameter set based on the protection adjustment commands. The target airflow coordination parameter set includes at least the target airflow dominant direction, the target airflow velocity range, and the target local wind pressure distribution description, and outputs the target airflow coordination parameter set.

[0045] Based on the target airflow coordination parameter set, a partition mapping is performed on the flexible airflow support structure inside the intelligent protective cabin to generate and output the airflow action area division result. The airflow action area division result is used to clarify the airflow adjustment priority corresponding to different cabin areas.

[0046] Based on the division of airflow action areas, calculate the initial air pressure setting value for each airflow action area, generate an area initial air pressure configuration table, and output the area initial air pressure configuration table.

[0047] Based on the regional initial air pressure configuration table and the target local wind pressure distribution description in the target airflow coordination parameter set, air pressure difference redistribution is performed on each airflow action area to generate regional air pressure difference adjustment results and output the regional air pressure difference adjustment results. The regional air pressure difference adjustment results are used to drive the flexible airflow support structure to form a non-uniform air pressure field in the cabin.

[0048] Under the constraint of the regional air pressure difference adjustment result, the airflow guiding unit in the intelligent protective cabin is controlled to change the airflow release direction and release intensity, generate real-time airflow guiding state and output the real-time airflow guiding state, so that the airflow in the cabin forms a directional airflow field around the welding tool assembly that matches the welding posture adjustment direction.

[0049] Based on the real-time airflow guidance status, the stability of the cabin air pressure is continuously monitored and the adjustment result of the regional air pressure difference is closed-loop corrected to generate and output the airflow stability correction amount. The airflow stability correction amount is used to correct the air pressure output of the flexible airflow support structure in real time, so as to maintain the continuity and stability of the directional airflow field within the effective period of the protection adjustment command.

[0050] In this invention, the intelligent protective cabin is an active protective execution unit that is closely linked to the output of the fusion module. Inside, through an adjustable air pressure flexible airflow support structure and an airflow guiding unit, it dynamically constructs a local airflow environment that matches the changes in welding posture during the welding operation.

[0051] The intelligent protective cabin first receives a protective adjustment command output by the fusion module. This protective adjustment command is a control command generated by the fusion module after integrating environmental disturbance characteristic data, welding process status data, and posture information. Its content does not directly indicate specific air pressure values ​​or airflow switching states, but rather describes the protective target in a parameterized manner. Upon receiving this protective adjustment command, the intelligent protective cabin analyzes it, extracts, and forms a target airflow coordination parameter set. The so-called "target airflow coordination parameter set" is the core parameter set used in this invention to characterize the protective target. It includes at least the target airflow dominant direction, the target airflow velocity range, and a description of the target local wind pressure distribution. The target airflow dominant direction characterizes the spatial direction in which the main airflow within the protective cabin should be directed, typically related to the welding posture adjustment direction and the direction of external wind and sand intrusion. The target airflow velocity range limits the upper and lower limits of the airflow release intensity to avoid the airflow being too weak to form a protective effect or too strong to interfere with the welding arc flame. The target local wind pressure distribution description describes the pressure difference relationship that should be formed between different areas within the protective cabin. After completing the analysis, the intelligent protective cabin uses this target airflow coordination parameter set as input and output for subsequent control steps.

[0052] After obtaining the target airflow coordination parameter set, the intelligent protective cabin performs zonal mapping on its internal flexible airflow support structure based on this parameter set. The "flexible airflow support structure" refers to a structural system composed of multiple independently adjustable airflow support units, capable of creating different air pressure states in different areas. "Zonal mapping" refers to dividing the internal space of the protective cabin or the flexible airflow support structure into multiple airflow action zones based on the dominant direction of the target airflow and the target local wind pressure distribution. Each airflow action zone corresponds to a set of airflow support units and is assigned different adjustment priorities. After completing this mapping, the intelligent protective cabin generates and outputs the airflow action zone division results within the cabin. These results are used to determine which areas require priority airflow establishment and which areas only need auxiliary adjustment during subsequent adjustments, thereby avoiding energy waste and airflow interference caused by uniform adjustment throughout the cabin.

[0053] After clarifying the division of airflow action zones, the intelligent protective cabin calculates the initial air pressure setpoint for each airflow action zone based on the division results and generates a regional initial air pressure configuration table. The "initial air pressure setpoint" mentioned here refers to the base air pressure output value set for each airflow action zone before dynamic pressure differential adjustment. This value is typically determined based on the structural dimensions of the protective cabin, the performance parameters of the airflow support unit, and the target airflow velocity range. For example, for areas with higher priority that need to form the dominant airflow, their initial air pressure setpoint can be higher than other areas; for areas that only need to form auxiliary or buffer airflows, their initial air pressure setpoint can be relatively lower. The regional initial air pressure configuration table records the relationship between each airflow action zone and its corresponding initial air pressure setpoint in tabular or data structure form, and is output as the reference input for subsequent pressure differential redistribution.

[0054] Based on this, the intelligent protective cabin performs pressure differential redistribution on each airflow area according to the initial regional air pressure configuration table and the target local wind pressure distribution description in the target airflow coordination parameter set. "Pressure differential redistribution" refers to the differentiated adjustment of air pressure output in different airflow areas while maintaining overall air pressure output level control, thereby creating a non-uniform pressure field within the cabin. This non-uniform pressure field guides airflow along specific paths, enabling the airflow to form a stable protective structure around the welding tool assembly. For example, when the target local wind pressure distribution description indicates that a higher air pressure is needed on the windward side of the welding tool assembly and a lower air pressure on the leeward side, the intelligent protective cabin will correspondingly increase the air pressure output on the windward side and decrease the air pressure output on the leeward side, thereby driving the airflow along the welding posture adjustment direction. After completing the above adjustments, the intelligent protective cabin generates and outputs the regional pressure differential adjustment result, which is directly used to drive the flexible airflow support structure to form the aforementioned non-uniform pressure field.

[0055] Under the constraint of the regional air pressure difference adjustment, the intelligent protective chamber further controls its internal airflow guiding units to change the airflow release direction and intensity. The "airflow guiding unit" mentioned here refers to a mechanism installed inside the protective chamber to control the direction, angle, or opening of the airflow outlet; it can work in conjunction with the flexible airflow support structure. Based on the regional air pressure difference adjustment results, the intelligent protective chamber determines the opening status and guiding angle of each airflow guiding unit, ensuring that the airflow flows along the expected path during release, and generates and outputs real-time airflow guiding status. Through this control, the airflow inside the chamber forms a directional airflow field around the welding tool assembly that matches the welding posture adjustment direction, thereby continuously providing a stable local environment for the welding arc flame during posture changes.

[0056] After establishing a real-time airflow guidance state, the intelligent protective cabin does not cease adjustment. Instead, it continuously monitors the stability of the air pressure within the cabin based on the real-time airflow guidance state and performs closed-loop correction on the regional air pressure difference adjustment results. The "air pressure stability" referred to here refers to the degree of fluctuation in air pressure output over time in each airflow area within the cabin, which can be obtained through statistical analysis of air pressure sensor sampling values. When abnormal fluctuations or deviations from the expected distribution in a certain area's air pressure are detected, the intelligent protective cabin calculates the airflow stability correction amount based on the magnitude of the deviation and outputs this correction amount. The "airflow stability correction amount" refers to the adjustment amount used to correct the air pressure output of the flexible airflow support structure, which can be expressed as an increase or decrease in the air pressure output of a certain area. The intelligent protective cabin uses this airflow stability correction amount to correct the air pressure output of the flexible airflow support structure in real time, thereby maintaining the continuity and stability of the directional airflow field within the effective period of the protective adjustment command, avoiding attenuation of the protective effect or disruption of the welding process due to air pressure fluctuations.

[0057] Through the above execution process, the intelligent protective cabin realizes a complete closed-loop control logic from the parsing of protective adjustment commands to the stable maintenance of the directional airflow field, so that its protective behavior is highly consistent with the prediction results of the fusion module and the welding posture adjustment process.

[0058] The data acquisition module 104 is used to collect wind speed, temperature and humidity, dust density and environmental acoustic information in high wind and sand environments.

[0059] The acquisition module 104 is an environmental sensing unit in this invention used to acquire external disturbance information of high wind and sand operation environment. Its function is to transform natural environmental factors that are originally unquantifiable and cannot be directly used for control decisions into structured, calculable and time-alignable environmental disturbance feature data, thereby providing a reliable data source for the fusion module 105 to conduct comprehensive analysis and prediction of the welding process.

[0060] The data acquisition module 104 collects information including wind speed, temperature and humidity, dust density, and ambient acoustics. These are all key environmental parameters that directly or indirectly affect the stability of the welding process in high-wind, sandy environments. Wind speed refers to the speed of air relative to the ground per unit time, including both magnitude and direction. In this invention, wind speed information can be acquired using a wind speed sensor or an ultrasonic anemometer. The acquired wind speed values ​​can be quantified in meters per second and continuously sampled at fixed time intervals. For example, the sampling period can be set to 0.1 seconds, meaning the current wind speed value is output every 0.1 seconds, thus forming a sequence of wind speed data over time. This wind speed information reflects the potential disturbance intensity of external airflow on the welding arc and weld pool.

[0061] Temperature and humidity information refers to the temperature and relative humidity values ​​of the air within the welding operation area. Temperature information reflects the impact of ambient thermal conditions on heat conduction and cooling rate during the welding process, while humidity information is related to factors such as the effectiveness of gas shielding and the probability of weld porosity formation during welding. In this invention, temperature and humidity information can be collected using an integrated temperature and humidity sensor, where temperature is expressed in degrees Celsius and humidity as a percentage. The acquisition module 104 can sample temperature and humidity at time intervals that are the same as or different from the wind speed information, and uses timestamps to ensure alignment with other data later.

[0062] Dust density information is an important environmental parameter introduced in this invention for high-wind-sand environments, used to characterize the concentration level of suspended particulate matter in the air. Dust density refers to the mass or quantity of suspended particulate matter per unit volume of air, and can be quantified using milligrams per cubic meter (mg / m³) or particle counts per cubic meter (PDC). In this invention, dust density information can be collected using a light-scattering dust sensor or a laser particulate matter sensor. The basic principle is to utilize the scattering characteristics of particulate matter on light, converting the detected scattering signal into a particulate matter concentration value. For example, when the sensor detects a particulate matter concentration of 2.0 mg / m³ corresponding to the scattered light intensity within a certain sampling period, this value can be output as the dust density information at that moment. Through continuous sampling, a sequence of dust density changes over time can be obtained, which can be used to determine the intensity and duration of sandstorm disturbances.

[0063] Ambient acoustic information refers to the characteristic information of ambient sound within the welding operation area, which includes not only sound intensity but also sound spectrum characteristics. This ambient acoustic information can be collected through microphones or acoustic sensors, and parameters reflecting the characteristics of environmental disturbances can be obtained through signal processing. For example, in a windy and dusty environment, strong winds sweeping across the surface of protective structures or equipment will generate noise in specific frequency bands, while sand particles impacting the protective chamber or equipment surface will also introduce acoustic signals with impulse characteristics. By performing frequency domain analysis on the ambient sound, acoustic feature values ​​reflecting the intensity of wind and dust activity can be extracted. For instance, a fast Fourier transform can be performed on the collected sound signal to calculate the energy value within a specific frequency band as an ambient acoustic feature; the higher the energy value, the more severe the environmental disturbance.

[0064] During operation, the acquisition module 104 needs to synchronously or near-synchronously acquire the aforementioned multiple types of environmental information and attach a unified time stamp to each type of acquired data, thereby forming a set of environmental disturbance characteristic data that can be used for subsequent time alignment processing. Time alignment refers to the ability of sensor data from different sources and with different sampling frequencies to be mapped to the same time axis through timestamps, enabling the fusion module 105 to analyze the combined impact of wind speed, temperature, humidity, dust density, and acoustic characteristics on the welding process state under the same time reference. For example, when an abnormal fluctuation in the welding current waveform is detected at a certain time point, the fusion module 105 can use the time alignment relationship to find the corresponding changes in wind speed and dust density before and after that time point, thereby determining whether the anomaly was caused by a sudden sandstorm disturbance.

[0065] Through the above design, the acquisition module 104 transforms the complex, variable and random high wind and sand environmental factors into continuous, quantifiable and aligned environmental disturbance characteristic data, providing the basic conditions for the fusion module 105 to carry out time series modeling and state prediction.

[0066] The fusion module 105 is used to perform unified time alignment and feature mapping on environmental disturbance feature data, welding process status data and pose information, and generate welding parameter adjustment instructions, welding posture correction instructions and protection adjustment instructions based on time-series feature modeling and state prediction mechanism.

[0067] The fusion module 105 is the core unit in this invention that enables information flow and control linkage between various functional modules. Its function is to unify the understanding, correlation analysis and prediction of the heterogeneous data output by the carrier platform 101, execution module 102, intelligent protective cabin 103 and acquisition module 104, and on this basis form control instructions that can be directly executed by subsequent modules.

[0068] The environmental disturbance feature data received by the fusion module 105 originates from the acquisition module 104. This environmental disturbance feature data includes at least wind speed information, temperature and humidity information, dust density information, and environmental acoustic information. Simultaneously, the fusion module 105 also receives welding process status data from the execution module 102. This welding process status data specifically includes current waveform information and arc flame status feature information acquired by the arc flame stabilization device. Furthermore, the fusion module 105 also receives pose information from the support platform 101, used to characterize the real-time position and attitude state of the execution module 102 in space. These three types of data differ in physical meaning, sampling frequency, and variation characteristics; therefore, the fusion module 105 first needs to perform unified time alignment processing on them.

[0069] In practical implementation, the fusion module 105 can use a unified system clock as the time reference and add timestamps to all input data. Then, a fixed-duration analysis time window is used as the processing unit, for example, setting each analysis cycle to 0.2 seconds. Within this analysis cycle, the fusion module 105 aggregates environmental data such as wind speed and dust density output by the acquisition module 104 within this time range, current waveform information and arc flame state characteristic information output by the execution module 102 within the same time range, and pose information output by the carrying platform 101 within this time range, thereby forming a multi-source data set that is strictly corresponding in time. For high-frequency data such as current waveforms, their average value, maximum fluctuation amplitude, or rate of change can be calculated within this time window as representative features; for low-frequency data such as wind speed or pose, the latest sampled value or weighted average value within this time window can be used as a representative.

[0070] After time alignment, the fusion module 105 performs feature mapping processing on the aforementioned data, enabling it to be used for unified time-series modeling. Environmental disturbance feature data refers to a set of features that, after mapping, can quantify the intensity of external wind and sand disturbances. For example, wind speed information is mapped to a wind speed change rate feature, dust density information to a dust disturbance intensity feature, and environmental acoustic information to a wind and sand impact feature. Welding process state data is mapped to features reflecting welding stability, such as calculating the current fluctuation amplitude based on current waveform information and calculating the arc flame stability index based on arc flame state feature information. Pose information is mapped to attitude offset features, used to characterize the degree of deviation of the execution module 102 relative to the ideal welding posture.

[0071] Based on this, the fusion module 105 continuously analyzes the aforementioned features using a temporal feature modeling and state prediction mechanism. Temporal feature modeling refers to constructing a feature sequence from environmental disturbance features, welding process state features, and attitude deviation features over multiple consecutive analysis cycles, and analyzing their changing trends and interrelationships. For example, if, over three consecutive analysis cycles, a continuous increase in wind speed change rate, a synchronous increase in dust disturbance intensity, and a gradually amplifying trend in current fluctuation amplitude are detected, the fusion module 105 can establish a correlation between enhanced environmental disturbance and decreased welding stability. The state prediction mechanism, based on the aforementioned correlation, infers the development trend of the welding state in the next analysis cycle. For example, it predicts whether current fluctuations will exceed the allowable threshold or whether there is a risk of arc flame instability without control measures.

[0072] Based on the prediction results, the fusion module 105 further generates welding parameter adjustment commands, welding posture correction commands, and protection adjustment commands. Taking the welding parameter adjustment command as an example, when the prediction result indicates that the current fluctuation amplitude will significantly increase in the next cycle, the fusion module 105 can generate a welding parameter adjustment command for the execution module 102, such as instructing the welding tool assembly to appropriately reduce the welding current rise slope or adjust the welding rhythm, thereby reducing the sensitivity of the molten pool to wind and sand disturbances. This command directly corresponds to the welding tool assembly in the execution module 102 and can be directly executed by its control unit.

[0073] For example, in the process of generating welding posture correction commands, when the fusion module 105 determines, based on the pose information and environmental disturbance characteristics, that the supporting platform 101 has tilted slightly and the lateral wind force has increased, which may cause the welding torch incident angle to deviate, the fusion module 105 can generate a welding posture correction command. This command instructs the multi-joint robotic arm to make minor adjustments to the welding torch posture in the next time cycle, such as increasing the preset angle compensation around a specific joint, thereby offsetting the combined effects of platform posture changes and wind force on the welding posture. This posture correction command directly corresponds to the multi-joint robotic arm control logic in the execution module 102, ensuring that the posture adjustment has a clear execution target.

[0074] The protective adjustment command corresponds to the airflow adjustment function of the intelligent protective cabin 103. When the fusion module 105 predicts that the external wind speed and dust density will increase synchronously in a short period of time, it can generate a protective adjustment command, instructing the intelligent protective cabin 103 to increase the air pressure output of the windward airflow channel and adjust the airflow direction to form a directional airflow inside the protective cabin that is conducive to offsetting the external wind and sand. This protective adjustment command directly corresponds to the adjustable air pressure flexible airflow support structure of the intelligent protective cabin 103 by controlling the airflow direction, airflow speed, and local wind pressure distribution parameters.

[0075] It is important to emphasize that the above three types of instructions are not generated in isolation, but are generated collaboratively based on the same temporal feature modeling and state prediction results. For example, within the same prediction cycle, the fusion module 105 can simultaneously generate welding parameter adjustment instructions, welding posture correction instructions, and protection adjustment instructions, enabling the execution module 102 and the intelligent protection cabin 103 to respond collaboratively to environmental disturbance changes within the same time scale, thereby forming a complete closed-loop control process.

[0076] Furthermore, the fusion module is specifically used for:

[0077] The attitude offset vector is calculated and output based on the pose information. The attitude offset vector is used to characterize the influence trend of the attitude change of the bearing platform on the end direction of the welding tool assembly.

[0078] The attitude compensation requirement is calculated and output based on the attitude offset vector and welding process state data. Welding attitude correction command is generated and output based on the attitude compensation requirement.

[0079] The protection and countermeasure vector is calculated and output based on the environmental disturbance characteristic data. The protection and countermeasure vector is used to characterize the main intrusion direction and intensity of external wind and sand disturbance on the arc flame area.

[0080] Based on the attitude adjustment direction, attitude compensation requirement and protection and countermeasure vector represented in the welding attitude correction command, the airflow coordination target is constructed and output.

[0081] Based on the airflow coordination target, a protection adjustment command is generated. The protection adjustment command includes at least airflow direction parameters, airflow velocity parameters, and local wind pressure distribution parameters. The protection adjustment command is output and sent to the intelligent protection cabin to drive its adjustable air pressure flexible airflow support structure to form an airflow field that matches the attitude adjustment direction.

[0082] In this invention, the fusion module undertakes the core function of converting multi-source data generated by the carrier platform, execution module and acquisition module into executable control instructions. In this embodiment, the specific implementation logic of the fusion module in the coordinated control of welding posture correction and protection adjustment is further defined.

[0083] First, the fusion module calculates the attitude offset vector based on the pose information output by the carrier platform. The "pose information" referred to here refers to the position and attitude state of the carrier platform relative to the reference coordinate system at the current moment, which includes at least attitude angle information, such as pitch, roll, and yaw angles. The "attitude offset vector" specifically refers to a vector quantity used to describe the trend of the influence of the carrier platform's attitude changes on the spatial orientation of the welding tool assembly's end effector. This attitude offset vector is not directly equivalent to the platform's attitude angle change value, but rather, considering the installation relationship of the welding tool assembly relative to the carrier platform, it maps the platform's attitude changes to the potential directional offset trend of the welding tool assembly's end effector.

[0084] In the specific calculation process, the spatial direction vector of the welding tool assembly under ideal welding conditions can be used as the reference direction vector. When the bearing platform changes attitude, the reference direction vector is rotated and transformed according to the change in platform attitude angle and the installation direction of the welding tool assembly, thereby obtaining the current direction vector of the welding tool assembly's end. The difference vector between this current direction vector and the reference direction vector is output as the attitude offset vector. For example, when the bearing platform changes its roll angle to the right, the attitude offset vector will reflect the tendency of the welding tool assembly's end to shift to the left or right. This trend vector is used to subsequently determine which direction the welding attitude needs to be compensated in.

[0085] After obtaining the attitude offset vector, the fusion module further calculates the attitude compensation requirement based on the attitude offset vector and the welding process state data. The "welding process state data" mentioned here originates from the arc flame stabilization device in the execution module, and includes at least current waveform information and arc flame state characteristic information, reflecting the stability of the current welding process. The "attitude compensation requirement" refers to the magnitude of attitude compensation that the welding tool assembly needs to perform to offset the adverse effects of attitude offset on welding quality; it can be expressed as an angular compensation amount or a directional compensation ratio.

[0086] In practical implementation, the fusion module can determine the degree of attitude deviation based on the amplitude of the attitude deviation vector, and combine this with welding process status data to determine whether the attitude deviation has had a substantial impact on welding stability. For example, when the amplitude of the attitude deviation vector is small and the current waveform is stable, and the arc flame characteristics are in a stable range, the attitude compensation requirement can be determined to be a small value; while when the amplitude of the attitude deviation vector increases and is accompanied by increased current fluctuations or enhanced arc flame jitter, the attitude compensation requirement increases accordingly. The fusion module generates a welding attitude correction command based on this attitude compensation requirement. This welding attitude correction command explicitly instructs the multi-joint robotic arm to adjust the attitude of the welding tool assembly according to the attitude compensation requirement in subsequent control cycles, and outputs the welding attitude correction command.

[0087] Simultaneously with the generation of welding posture correction commands, the fusion module also calculates a protective countermeasure vector based on the environmental disturbance characteristic data output by the acquisition module. The "environmental disturbance characteristic data" mentioned here includes at least wind speed information, dust density information, and environmental acoustic information, used to comprehensively reflect the directionality and intensity of external wind and sand disturbances. The "protective countermeasure vector" specifically refers to a vector quantity used to characterize the intrusion tendency of external wind and sand disturbances on the welding arc flame area; this vector contains at least directional and intensity components.

[0088] In practical calculations, the fusion module can determine the main direction of wind and sand disturbance based on the wind direction detected by the wind speed sensor, and quantify the overall intensity of the disturbance by combining wind speed, dust density level, and ambient acoustic intensity, thereby constructing a protective countermeasure vector. For example, when a wind speed of 6 meters per second is detected and the wind direction is from the right front to the left rear of the welding area, while the dust density increases significantly, the protective countermeasure vector will point in the direction from the right front towards the welding arc flame area, and its vector amplitude is used to characterize the potential threat level of the disturbance to the stability of the arc flame.

[0089] After obtaining the welding posture correction command, posture compensation requirement, and protection countermeasure vector, the fusion module constructs an airflow coordination target based on the above three types of information. The so-called "airflow coordination target" refers to the overall airflow adjustment target that the intelligent protective cabin needs to form within the current control cycle. Its purpose is not simply to counteract external wind and sand disturbances, but to keep the airflow field formed inside the protective cabin coordinated with the adjustment direction of the welding posture, thereby avoiding the welding posture being subjected to new airflow disturbances just after it has been corrected.

[0090] In the specific construction process, the fusion module uses the attitude adjustment direction represented in the welding attitude correction command as the main reference direction, the attitude compensation requirement as the reference basis for the airflow adjustment intensity, and combines the protection and countermeasure vector to determine the direction of external disturbances that need to be focused on countering. For example, when the welding attitude correction command instructs the welding tool assembly to make a slight adjustment to the left, and the protection and countermeasure vector indicates that the external wind and sand mainly invade from the right, the airflow coordination target will be constructed to form an auxiliary airflow from left to right in the welding area, thereby providing support for the welding arc flame in the attitude adjustment direction and weakening the direct impact of external wind and sand in the countermeasure direction.

[0091] Finally, the fusion module generates a protection adjustment command based on the airflow coordination target. This protection adjustment command includes at least airflow direction parameters, airflow velocity parameters, and local wind pressure distribution parameters. The airflow direction parameter indicates the dominant flow direction of the airflow inside the intelligent protective cabin, the airflow velocity parameter indicates the airflow intensity, and the local wind pressure distribution parameter limits the air pressure differences in different areas of the protective cabin. The fusion module outputs this protection adjustment command and sends it to the intelligent protective cabin, which then drives its adjustable-pressure flexible airflow support structure to form an airflow field that matches the welding posture adjustment direction according to the protection adjustment command.

[0092] Through the above-mentioned step-by-step calculation, step-by-step output, and strong dependency process, the fusion module realizes deep collaborative control between welding posture correction and protective airflow adjustment, so that the intelligent protective cabin is no longer a passive protection unit independent of welding control, but becomes a dynamic adjustment unit that actively cooperates with changes in welding posture.

[0093] Furthermore, the fusion module is specifically used for:

[0094] The system acquires wind speed, temperature and humidity, dust density, and environmental acoustic information from the acquisition module, generates and outputs the original environmental disturbance sequence with a unified timestamp.

[0095] The current waveform information and arc flame state characteristic information output by the arc flame stabilization device in the execution module are obtained, and the original sequence of the welding process with a unified timestamp is generated and output. The timestamp of the original sequence of the welding process is aligned with the timestamp of the original sequence of environmental disturbance.

[0096] The pose information output by the carrier platform is obtained, and the pose information is interpolated or maintained based on the timestamp of the original environmental disturbance sequence. A pose alignment sequence corresponding to the original environmental disturbance sequence at each time step is generated and the pose alignment sequence is output.

[0097] Based on the original environmental disturbance sequence, the original welding process sequence, and the pose alignment sequence, a fusion sample within the same time window is constructed. The fusion sample is then mapped into a multi-source feature vector sequence that includes wind and sand disturbance intensity features, welding stability features, and attitude offset features. The multi-source feature vector sequence serves as the input for subsequent time-series feature modeling and state prediction mechanisms.

[0098] In this invention, the fusion module undertakes the key function of uniformly organizing and expressing data from different modules and different physical sources. This embodiment further clarifies and defines the specific implementation method of the fusion module in the stages of multi-source raw data construction, time alignment and feature mapping.

[0099] The fusion module first acquires wind speed, temperature and humidity, dust density, and ambient acoustic information output by the acquisition module, and then generates a raw environmental disturbance sequence based on this information. The "raw environmental disturbance sequence" referred to here is a data sequence set formed by organizing multiple environmental parameters reflecting the disturbance state of a high-wind-sand environment in chronological order. To ensure the comparability and consistency of different environmental parameters over time, the fusion module adds a unified timestamp to each piece of environmental data when generating the raw environmental disturbance sequence. This timestamp can originate from a unified system clock, such as a high-precision clock maintained internally by the fusion module or a time synchronization signal provided by the platform, thereby ensuring that wind speed, temperature and humidity, dust density, and ambient acoustic information are recorded within the same time reference frame.

[0100] In practical implementation, assuming the acquisition module outputs environmental data at fixed time intervals, such as current wind speed, temperature and humidity, dust density, and environmental acoustic characteristics every 0.1 seconds, the fusion module can combine these four types of data with the current timestamp to form an environmental disturbance data unit at the end of each acquisition cycle, and arrange them continuously in chronological order to form the original environmental disturbance sequence. This original environmental disturbance sequence fully reflects the process of high-wind and sandstorm environmental disturbance changing over time and serves as the basic input for subsequent analysis.

[0101] After generating the original environmental disturbance sequence, the fusion module further acquires the current waveform information and arc flame state characteristic information output by the arc flame stabilization device in the execution module, and generates the original welding process sequence. The "original welding process sequence" referred to here is a data sequence formed by organizing key parameters reflecting welding stability and arc flame state in chronological order during the welding process. Since there is a causal relationship between the welding process state and the environmental disturbance, to ensure a strict temporal correspondence between the two, the timestamp of the original welding process sequence is aligned with the timestamp of the original environmental disturbance sequence.

[0102] In practice, this means that when the fusion module receives current waveform information or arc flame state characteristic information, it will look for the closest or corresponding timestamp of the original environmental disturbance sequence and assign that timestamp to the corresponding welding process data. For example, if a significant fluctuation in the welding current waveform is detected at a certain time t, and there is an environmental data unit with timestamp t in the original environmental disturbance sequence, the fusion module will use that timestamp t as the time identifier for the welding process data, thereby ensuring a one-to-one correspondence between the original welding process sequence and the original environmental disturbance sequence on the time axis.

[0103] Subsequently, the fusion module acquires the pose information output by the carrier platform and performs interpolation or maintain-update processing on the pose information based on the timestamp of the original environmental disturbance sequence to generate a pose alignment sequence. The "pose information" referred to here refers to the attitude state information of the carrier platform in space, such as pitch angle, roll angle, and yaw angle, whose sampling frequency may be lower than that of the environmental disturbance data or welding process data. To solve the time inconsistency problem caused by different sampling frequencies, the fusion module uses the timestamp of the original environmental disturbance sequence as the target time point to align the pose information.

[0104] The term "interpolation or maintaining update" refers to the process where, when the sampling time of pose information does not completely coincide with the timestamp of a certain environmental disturbance, the most recent pose sample value can be kept unchanged, or interpolation can be performed between two adjacent pose sample values ​​to obtain the pose estimate corresponding to the timestamp of the environmental disturbance. For example, if pose information is sampled at times t1 and t2, and the timestamp of the environmental disturbance is t, and t1 < t < t2, then the pose information at time t can be calculated using linear interpolation, ensuring that the pose alignment sequence has a clear pose value at each environmental disturbance time point.

[0105] After obtaining the original environmental disturbance sequence, the original welding process sequence, and the pose alignment sequence, the fusion module constructs a fusion sample within the same time window based on these three types of sequences. The "time window" referred to here is a continuous time interval selected on the time axis, such as 0.5 seconds or 1 second, used to describe the comprehensive changes in environmental disturbance, welding process state, and platform attitude over a period of time. Within each time window, the fusion module combines the original environmental disturbance data, the original welding process data, and the pose alignment data falling within that time window range to form a fusion sample.

[0106] During the construction of the fused samples, the fusion module further performs feature mapping processing on the original data, mapping the fused samples into a multi-source feature vector sequence. A "multi-source feature vector sequence" refers to a set of feature vectors composed of multiple feature components arranged in chronological order, where each feature vector includes at least wind and sand disturbance intensity features, welding stability features, and attitude offset features. Wind and sand disturbance intensity features can be calculated based on wind speed, dust density level, and environmental acoustic characteristics, used to quantify the strength of environmental disturbances; welding stability features can be calculated based on the fluctuation amplitude of the current waveform or the arc flame state characteristics, used to reflect whether the welding process is stable; attitude offset features can be calculated based on the pose alignment sequence, used to characterize the degree of influence of the bearing platform's attitude change on the orientation of the welding tool components.

[0107] For example, within a certain time window, if the average wind speed is 5 meters per second, the dust density is 1.8 milligrams per cubic meter, the current waveform fluctuation amplitude increases significantly, and the bearing platform exhibits a slight roll angle change, then the corresponding multi-source feature vector will simultaneously contain feature values ​​reflecting strong wind and sand disturbance, feature values ​​reflecting decreased welding stability, and feature values ​​reflecting attitude deviation. The fusion module arranges the multi-source feature vectors formed within this time window in chronological order to form a multi-source feature vector sequence, and uses this multi-source feature vector sequence as input for subsequent time-series feature modeling and state prediction mechanisms.

[0108] Through the above continuous, explicit and strongly dependent processing flow, the fusion module achieves a complete transformation from multi-source raw data to a unified feature expression, enabling deep alignment of environmental disturbance information, welding process status information, and pose information at both the temporal and semantic levels.

[0109] Furthermore, the fusion module is specifically used for:

[0110] The environmental disturbance risk index is calculated based on the multi-source feature vector sequence and then output.

[0111] Based on the environmental disturbance risk index, a prediction window is selected from the multi-source feature vector sequence and the window length is determined. The prediction window description information is output, wherein the prediction window description information includes the window start and end time and the window length and is used to limit the range of subsequent prediction input.

[0112] Within the scope of the prediction window description information, time-series feature modeling is performed on the multi-source feature vector sequence to obtain and output the welding stability prediction results;

[0113] The control response type is determined based on the welding stability prediction results and environmental disturbance risk indicators, and the control response type determination result is output. The control response type determination result is used to select the instruction combination.

[0114] Based on the control response type determination result, an instruction set containing at least welding parameter adjustment instructions is generated. When the control response type meets the attitude compensation trigger condition, a welding attitude correction instruction is added to the instruction set. When the control response type meets the protection enhancement trigger condition, a protection adjustment instruction is added to the instruction set. The instruction set is output and sent to the execution module and the intelligent protection cabin respectively for coordinated control.

[0115] The fusion module first calculates the environmental disturbance risk index based on a multi-source feature vector sequence. The "environmental disturbance risk index" is a comprehensive evaluation metric used to quantify the likelihood of adverse effects of current and short-term environmental disturbances on the welding process. It is not a single environmental parameter, but rather the result of the combined effect of multiple feature components from the multi-source feature vector sequence. In its implementation, the fusion module can statistically analyze the intensity characteristics of wind and sand disturbances at multiple consecutive time points, and combine this with the changing trends of welding stability characteristics and the magnitude of attitude deviation characteristics to assess the degree of danger of the environmental disturbance. For example, when the intensity characteristics of wind and sand disturbances remain at a high level, and the welding stability characteristics show a significant downward trend while the attitude deviation characteristics gradually increase, the fusion module classifies this state as a high-risk environmental disturbance state and generates a high environmental disturbance risk index accordingly. Conversely, when both environmental disturbance characteristics and welding stability characteristics remain within a stable range, the generated environmental disturbance risk index is relatively low.

[0116] After obtaining the environmental disturbance risk index, the fusion module selects a prediction window and determines its length from the multi-source feature vector sequence based on this index. The "prediction window" here refers to a continuous interval selected on the time axis, used as the input range for time-series feature modeling and state prediction. The selection of the prediction window is not fixed but adaptively adjusted according to the environmental disturbance risk index. When the environmental disturbance risk index is low, the prediction window can be set to a longer time range to smooth short-term fluctuations; when the environmental disturbance risk index is high, the prediction window can be shortened to highlight rapid changes in the recent environment and welding state. When determining the prediction window, the fusion module generates prediction window description information, which includes at least the window's start time, end time, and length, and outputs this information to limit the subsequent prediction input range.

[0117] Within the scope defined by the information in the prediction window, the fusion module performs time-series feature modeling on the multi-source feature vector sequence to obtain the welding stability prediction result. "Time-series feature modeling" refers to using multi-source feature vectors arranged chronologically within the prediction window as input, analyzing the trends of each feature over time and their interrelationships to infer the stability state of the welding process in the near future. This modeling process focuses not on the state at a single point in time, but on the direction of evolution of continuous state changes. For example, when the wind and sand disturbance intensity characteristic shows an upward trend while the welding stability characteristic shows a downward trend within the prediction window, the time-series feature modeling result will indicate that the welding stability may further deteriorate in the next stage. The fusion module outputs this analysis result as a welding stability prediction result, which can be characterized as different state levels such as stable, critically unstable, or significantly unstable.

[0118] After obtaining the welding stability prediction results, the fusion module further determines the control response type based on the welding stability prediction results and environmental disturbance risk indicators. The "control response type" referred to here is the overall control strategy category that the system should adopt under the current predicted state, which guides the combination of subsequent control commands. The determination of the control response type comprehensively considers the severity of external disturbances reflected by the environmental disturbance risk indicators and the internal process change trend reflected by the welding stability prediction results. For example, when the environmental disturbance risk indicator is low and the welding stability prediction results show that the welding process is still in a stable state, the control response type can be determined as a mild response; when the environmental disturbance risk indicator is moderate and the welding stability prediction results show an unstable trend, the control response type can be determined as a moderate response; when the environmental disturbance risk indicator is high and the welding stability prediction results show that the welding process is about to enter or has already entered an unstable state, the control response type is determined as a strong response. The fusion module outputs this control response type determination result for subsequent command combination selection.

[0119] Based on the control response type determination result, the fusion module generates an instruction set. This instruction set includes at least welding parameter adjustment instructions, used to adjust the working parameters of the welding tool components to directly affect the energy input and rhythm of the welding process. When the control response type meets the attitude compensation trigger condition, the fusion module adds a welding attitude correction instruction to the instruction set. The so-called "attitude compensation trigger condition" refers to the condition that, based on the welding stability prediction result and attitude deviation characteristics, the current welding attitude deviation has already had or will soon have an adverse impact on the welding quality. For example, when the attitude deviation characteristics exceed a preset threshold and the welding stability prediction result shows that the welding process is becoming unstable, it can be determined that the attitude compensation trigger condition is met.

[0120] Similarly, when the control response type meets the protection enhancement trigger condition, the fusion module will add a protection adjustment command to the command set. The "protection enhancement trigger condition" refers to the condition where, based on environmental disturbance risk indicators and wind and sand disturbance intensity characteristics, the current or upcoming environmental disturbance intensity exceeds the ordinary protection capability. For example, when the environmental disturbance risk indicator is in the high-risk range and the wind and sand disturbance intensity characteristics continue to rise, the protection enhancement trigger condition can be determined to be met. The protection adjustment command is used to drive the intelligent protection cabin to enhance its airflow resistance capability, thereby providing a more stable local environment for the welding process.

[0121] Finally, the fusion module outputs the generated instruction set, sending the welding parameter adjustment instructions and welding posture correction instructions to the execution module, and the protection adjustment instructions to the intelligent protection cabin for coordinated control. Through the above-described process of step-by-step calculation, step-by-step judgment, and step-by-step output, the fusion module achieves complete closed-loop control from multi-source feature input to prediction-driven instruction combination output, enabling the system to adopt differentiated and coordinated control strategies based on the risk level and prediction results in high-wind and sandstorm environments.

[0122] Furthermore, the fusion module is specifically used for:

[0123] The disturbance mutation index is calculated and output based on environmental disturbance characteristic data;

[0124] The disturbance mutation index is compared with a preset change threshold to generate a participating switching quantity and output the participating switching quantity. The participating switching quantity is used to control whether the arc flame state feature information enters the subsequent prediction input.

[0125] When participation is permitted by the switch quantity indication, arc flame status feature information is extracted from the welding process status data and arc flame enhancement feature is output. The arc flame enhancement feature and the disturbance mutation index together constitute an anomaly confirmation input.

[0126] When participation in the switch quantity indication is not allowed, an alternative confirmation feature is generated and output. The alternative confirmation feature is calculated from the fluctuation amplitude or rate of change of the current waveform information and is used to replace the arc flame enhancement feature.

[0127] An anomaly determination result is generated and output based on the anomaly confirmation input or alternative confirmation feature. The anomaly determination result is used as the constraint input of the time-series feature modeling and state prediction mechanism to affect the generation of welding parameter adjustment command, welding posture correction command and protection adjustment command.

[0128] The fusion module first calculates and outputs the disturbance mutation index based on environmental disturbance characteristic data. The "disturbance mutation index" quantifies the degree of rapid change in environmental disturbances over a short period, focusing on whether wind speed, dust density, or environmental acoustic characteristics exhibit sudden jumps. Unlike environmental disturbance risk indicators, the disturbance mutation index emphasizes the rate of change and abrupt changes rather than long-term risk levels. In its implementation, the fusion module can perform differential or rate-of-change calculations on environmental disturbance characteristic data at several consecutive time points. For example, it can calculate changes in wind speed, dust density, or environmental acoustic energy within adjacent time windows, normalize these changes, and then combine them to obtain the disturbance mutation index. For instance, if the wind speed suddenly increases from 3 m / s to 7 m / s within a time window, and the dust density also increases significantly, the disturbance mutation index will increase significantly, characterizing the sudden nature of the current environmental disturbance.

[0129] After obtaining the disturbance mutation index, the fusion module compares the disturbance mutation index with a preset change threshold to generate and output the participating switching quantity. The "preset change threshold" mentioned here refers to a threshold determined based on experience with typical operating environments during system initialization or calibration, used to distinguish between normal environmental fluctuations and sudden environmental disturbances. When the disturbance mutation index is greater than or equal to the preset change threshold, the participating switching quantity generated by the fusion module indicates that participation is allowed; when the disturbance mutation index is less than the preset change threshold, the participating switching quantity indicates that participation is not allowed. The participating switching quantity can exist in the form of a logical quantity or a status flag, and its core function is to serve as a control signal for subsequent feature selection, used to determine whether arc flame state characteristic information is introduced into the prediction input.

[0130] When the switch indicator allows participation, the fusion module extracts arc flame state characteristic information from the welding process status data and outputs arc flame enhancement characteristics. The "arc flame state characteristic information" mentioned here originates from the arc flame stabilization device in the execution module, and may include characteristic quantities characterizing the degree of arc flame stability, such as arc flame brightness changes, arc flame jitter frequency, or arc flame duration. In this state, the fusion module outputs the arc flame state characteristic information as "arc flame enhancement characteristics," which, together with the disturbance mutation index, constitutes the anomaly confirmation input. The "anomaly confirmation input" refers to a set of combined characteristics used to determine whether the welding process has entered an abnormal state due to a sudden environmental disturbance. It simultaneously includes external environmental mutation information and internal welding process response information, thereby improving the accuracy of anomaly detection.

[0131] When participation in the switching signal indication is not permitted, the fusion module will not introduce arc flame state characteristic information. Instead, it will generate and output alternative confirmation characteristics. These "alternative confirmation characteristics" refer to features used to confirm the stability of the welding process without using arc flame state characteristic information. They are calculated from the fluctuation amplitude or rate of change of the current waveform information. In specific implementations, the fusion module can calculate the difference between the maximum and minimum values ​​of the welding current within a time window, or calculate the average rate of change of the current between adjacent sampling points, as the alternative confirmation characteristics. Since the current waveform is the most direct reflection of the energy input in the welding process, it can already characterize welding stability well under stable environmental conditions. Therefore, when no sudden changes occur during disturbances, using alternative confirmation characteristics can reduce system complexity and avoid unnecessary noise introduction.

[0132] After obtaining an anomaly confirmation input or alternative confirmation feature, the fusion module forms an anomaly determination result based on the aforementioned features and outputs the anomaly determination result. The "anomaly determination result" referred to here is the conclusion regarding whether the current welding process is in an abnormal state. It can be represented in binary or multi-level state form, such as normal state, potential abnormal state, or explicit abnormal state. When forming the anomaly determination result, the fusion module comprehensively considers the relationship between the disturbance mutation index, arc flame enhancement feature, or alternative confirmation feature. For example, when the disturbance mutation index is high and the arc flame enhancement feature shows a significant decrease in arc flame stability, it can be determined that the welding process has entered an abnormal state; while when the disturbance mutation index is low and the alternative confirmation feature shows a stable current waveform, it is determined that the welding process is in a normal state.

[0133] Ultimately, the fusion module uses the anomaly determination result as a constraint input to the time-series feature modeling and state prediction mechanism, influencing the generation of welding parameter adjustment commands, welding posture correction commands, and protection adjustment commands. The term "constraint input" means that the anomaly determination result does not directly generate control commands, but rather serves to limit the boundary conditions of subsequent prediction and decision-making processes. For example, when the anomaly determination result indicates that the welding process has entered an abnormal state, the fusion module can prioritize a more conservative or stronger control response type when generating subsequent control commands; when the anomaly determination result indicates that the welding process is normal, the prediction model is allowed to generate control commands according to a conventional strategy. In this way, the anomaly determination result is naturally embedded into the prediction and control closed loop without disrupting the consistency of the overall control logic.

[0134] Through the above-mentioned step-by-step calculation, step-by-step judgment, and step-by-step constraint implementation, the fusion module can quickly identify sudden environmental disturbances in high-wind and sandstorm environments, and dynamically adjust the feature participation mode according to the disturbance characteristics, thereby reducing unnecessary system complexity while ensuring prediction accuracy.

[0135] The collaboration module 106 is used to receive video data of the welding process, environmental disturbance characteristic data, and welding parameter log data, and send remote decision information to the fusion module.

[0136] The collaboration module 106 is a functional module in this invention used to realize information interaction and collaborative control between the welding robot system and external monitoring, decision-making, and management units. Its core function is to further extend the structured and computable operational data within the system to the outside, and to safely and reliably introduce externally generated decision results into the system. This enables the entire welding robot system to not only possess local adaptive control capabilities in complex environments such as high winds and sandstorms, but also to have the capabilities of remote monitoring, manual intervention, and collaborative optimization. The collaboration module 106 maintains a direct data communication relationship with the fusion module 105, serving as both the uplink information output interface and the external decision input interface for the fusion module 105.

[0137] The welding process video data received by the collaboration module 106 primarily originates from a visual acquisition device located near the execution module 102 or inside the intelligent protective cabin 103. This visual acquisition device can be an industrial camera or a specialized camera device resistant to high temperatures and dust, used to continuously capture images of the welding area, weld formation, and arc flame morphology. Welding process video data refers to image sequence data that records the dynamic changes of the welding process in the form of continuous image frames. This data can intuitively reflect whether the welding is continuous, whether there is spatter in the weld, and whether the arc flame has shifted. After receiving this video data, the collaboration module 106 can perform necessary encoding or compression processing to adapt to the remote transmission bandwidth requirements while maintaining sufficient resolution and frame rate, enabling the remote end to clearly observe the welding status.

[0138] The collaboration module 106 is also used to receive environmental disturbance feature data and welding parameter log data. The environmental disturbance feature data originates from the environmental feature results generated by the fusion module 105 after time alignment and feature mapping, rather than direct raw sensor data. This environmental disturbance feature data can reflect comprehensive information such as the intensity of wind speed changes, dust density levels, and the degree of environmental acoustic disturbances over a current or historical period. The welding parameter log data refers to the recorded data of the changes in parameters such as welding current, welding rhythm, and attitude correction amplitude actually executed by the welding tool components over time during the welding process. This log data is usually stored in time series form and can be used for post-process analysis or remote diagnostics.

[0139] In this invention, "welding parameter log data" specifically refers to the historical record set of various control parameters issued by the fusion module 105 and actually executed by the execution module 102 during the welding process. This log data includes at least the parameter type, parameter value, and corresponding time identifier. For example, at a certain point in time, if the welding current is adjusted from its original set value to a lower value to cope with a sudden increase in wind speed, this adjustment and its corresponding value will be recorded in the welding parameter log data, thus providing a basis for subsequent remote analysis.

[0140] After receiving the aforementioned video data, environmental disturbance characteristic data, and welding parameter log data, the collaboration module 106 transmits them to the remote monitoring or decision-making unit via wired or wireless communication. This communication method can be selected from industrial Ethernet, private wireless networks, or cellular communication networks, depending on the actual application environment, to ensure stable data transmission capabilities even in high-wind, sandstorm, or long-distance operation scenarios. The remote monitoring or decision-making unit can be a centralized control center, maintenance terminal, or expert system, which obtains real-time and historical information from the welding site through the collaboration module 106 to comprehensively evaluate the welding process.

[0141] While completing the uplink data transmission, the collaboration module 106 is also used to send remote decision information to the fusion module 105. Remote decision information refers to control suggestions or constraints formed in a remote monitoring or decision-making unit based on human experience, historical data analysis, or safety management strategies. For example, when remote personnel discover an abnormal spatter risk in a welding area through video data, or determine based on environmental disturbance characteristics that current wind and sand conditions are approaching the safe operating threshold, remote decision information can be generated to impose restrictive requirements on the welding process, such as reducing the maximum permissible welding power, suspending welding, or adjusting protection strategies.

[0142] When the collaboration module 106 sends remote decision information to the fusion module 105, it converts the information into a decision input format that the fusion module 105 can recognize, such as transmitting it in the form of parameter constraints, priority indicators, or control mode switching instructions. After receiving the remote decision information, the fusion module 105 comprehensively judges it with the control strategy generated locally based on time-series feature modeling and state prediction, thereby executing the remote decision or correcting the local control results while ensuring system safety and stability.

[0143] It should be noted that the coordination module 106 does not directly control the execution module 102 or the intelligent protective cabin 103. It does not generate specific welding parameter adjustment commands, protection adjustment commands, or attitude correction commands itself. Instead, through coordination with the fusion module 105, it enables remote decision-making to participate in the overall system control logic, thereby avoiding conflicts between multiple control commands. In this way, the coordination module 106 plays a role in information aggregation, remote interaction, and decision-making collaboration within the system.

[0144] Through the above-mentioned data receiving objects, data type definitions, transmission methods, and decision information introduction mechanisms, the collaboration module 106 enables the intelligent welding robot system of the present invention to have the capabilities of remote visibility, traceability, and collaborative control.

[0145] Furthermore, the collaboration module is specifically used for:

[0146] The system receives video data of the welding process, environmental disturbance feature data, and welding parameter log data, and outputs a remote situational awareness packet. The remote situational awareness packet includes a timestamp, video segment index, and corresponding environmental disturbance feature summary. It then transmits the remote situational awareness packet to a remote decision-making terminal and receives remote decision information returned by the terminal. A constraint description is generated and output, wherein the constraint description includes at least parameter range constraints, output priority constraints, and allowed control response type constraints. The constraint description is sent to a fusion module, which converts it into constraint inputs and outputs constraint inputs. These constraint inputs define the generation boundaries of welding parameter adjustment commands, welding posture correction commands, and protection adjustment commands.

[0147] The fusion module is specifically used for:

[0148] When generating welding parameter adjustment commands, welding posture correction commands, and protection adjustment commands, the consistency verification of the command set to be generated is first performed based on the constraint input, and the verification result is output. Then, the command set to be generated is trimmed or downgraded according to the verification result to output the final command set. The final command set is then sent to the execution module and the intelligent protection cabin for control.

[0149] In this invention, the collaboration module and the fusion module together constitute the system's remote collaboration and local autonomous control interface. Its core objective is to introduce remote decision-making capabilities and impose boundary and strategic constraints on local control behavior without disrupting the local multi-source information fusion and predictive control closed loop. This embodiment explicitly defines the specific implementation method of this remote collaboration mechanism, ensuring that remote decision-making does not intervene in the form of "direct control," but rather influences the instruction generation process through constraints, thereby guaranteeing the system's stability and security even in complex environments such as high-wind and sandstorm conditions.

[0150] The collaboration module first receives video data of the welding process, environmental disturbance feature data, and welding parameter log data, and generates and outputs a remote situational awareness package based on the above data. The "video data" refers to image or video stream data continuously or intermittently acquired from the welding area, used to visually reflect the welding process and the surrounding environment. The "environmental disturbance feature data" can originate from environmental disturbance features calculated or extracted by the fusion module or acquisition module, used to reflect the intensity and changing trend of the current wind and sand environment. The "welding parameter log data" refers to log data recording the changes in parameters such as welding current, voltage, and welding speed over time, used to reflect the historical control behavior of the welding process.

[0151] A "remote situational awareness packet" is used to compress and organize complex, multi-source local situational information into a data structure suitable for remote transmission and rapid understanding. A remote situational awareness packet includes at least a timestamp, a video clip index, and a corresponding environmental disturbance feature summary. The timestamp identifies the time reference of the situational awareness packet; the video clip index indicates the specific video clip location that the remote decision-maker can retrieve or replay when needed; and the environmental disturbance feature summary is a simplified summary of the environmental disturbance feature data, such as the current level of wind and sand disturbance and the trend of disturbance changes, used to help the remote decision-maker quickly assess the on-site environmental condition. Through this method, the remote situational awareness packet retains key decision-making information while avoiding the communication burden of full data transmission.

[0152] After generating the remote situational awareness packet, the collaboration module transmits it to the remote decision-making terminal and receives remote decision information returned by the terminal. The "remote decision-making terminal" can be a computing device or a human decision-making system located in the monitoring center. Based on the received remote situational awareness packet, it analyzes the welding process and environmental conditions and provides strategic suggestions for subsequent control actions. Upon receiving the remote decision information, the collaboration module generates and outputs a constraint description. The "constraint description" refers to the data expression that imposes restrictions on the process of generating control commands by the local fusion module. It includes at least parameter range constraints, output priority constraints, and allowed control response type constraints.

[0153] Among these, parameter range constraints limit the allowed value range of welding parameter adjustment commands and shielding adjustment commands, such as limiting the maximum value of welding current or the upper limit of shielding gas velocity; output priority constraints indicate which type of command should be reserved or executed first when multiple types of control commands exist simultaneously; and allowed control response type constraints limit the range of control response types that the local system can adopt within the current remote decision-making cycle, such as allowing only moderate responses and prohibiting strong responses. The above constraint descriptions do not include specific control command values, but only describe the boundaries and priority relationships of control behaviors.

[0154] The collaboration module sends the generated constraint descriptions to the fusion module, which then converts these descriptions into constraint inputs and outputs them. The "constraint inputs" referred to here are constraint data structures within the fusion module used in the instruction generation and verification process. These structures map external constraint descriptions to internal constraints that the fusion module can directly use. For example, parameter range constraints are mapped to specific upper and lower numerical limits, output priority constraints are mapped to instruction sorting rules, and allowed control response type constraints are mapped to a set of optional control strategies. These constraint inputs define the generation boundaries of welding parameter adjustment instructions, welding posture correction instructions, and protection adjustment instructions, but they do not directly replace the basic logic of the fusion module in generating instructions based on multi-source features and prediction results.

[0155] During the generation of welding parameter adjustment commands, welding posture correction commands, and protection adjustment commands by the fusion module, the fusion module first performs a consistency check on the set of commands to be generated based on the constraint inputs and outputs the check results. The "set of commands to be generated" refers to the set of commands initially generated by the fusion module based on the prediction results and control response types before the introduction of remote constraints. The purpose of the consistency check is to determine whether the set of commands to be generated meets the conditions limited by the constraint inputs. For example, it checks whether the welding parameter adjustment commands exceed the parameter range constraints, whether the execution order of multiple commands conforms to the output priority constraints, and whether the current control response type is included within the allowed control response type constraint range.

[0156] When the consistency check result shows that the instruction set to be generated contains input violations of constraints, the fusion module performs pruning or degradation processing on the instruction set to be generated based on the check result, and outputs the final instruction set. "Pruning" refers to removing instructions that do not meet the constraints from the instruction set to be generated, such as deleting protection adjustment instructions that exceed the allowable parameter range. "Degradation" refers to reducing the control strength or response level of the instructions while maintaining the integrity of the instruction set, such as adjusting the original strong response to a medium response. Through pruning or degradation processing, the fusion module ensures that the final instruction set meets remote constraints while still maintaining the basic control functions of the system.

[0157] Finally, the fusion module sends the final set of instructions to the execution module and the intelligent protective cabin for control. Welding parameter adjustment and welding posture correction instructions are sent to the execution module, executed by the multi-joint robotic arm and welding tool assembly; protective adjustment instructions are sent to the intelligent protective cabin to drive its adjustable-pressure flexible airflow support structure. Through this process, the system achieves a collaborative working mode between remote decision-making and local intelligent control, enabling remote decision-making to safely and controllably intervene in the local control closed loop in a constraint-based manner.

[0158] Through this clear data organization method, constraint generation mechanism, and consistency verification and pruning logic, the technical solution provided in this embodiment effectively avoids the uncertainties and risks brought about by remote direct control, while ensuring the autonomy and real-time performance of the local fusion module in a high-wind and sandy environment.

[0159] Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims of this application.

Claims

1. A multi-source information fusion intelligent welding robot system for high-wind and sandy environments, characterized in that, include: The platform is used to carry and position the execution module and intelligent protective cabin in high-wind and sandstorm operation areas, and output position and posture information; The execution module includes a multi-joint robotic arm, a welding tool assembly, and an arc flame stabilization device. The multi-joint robotic arm is used to adjust the welding posture according to the welding trajectory instructions, the welding tool assembly is used to perform the actual welding operation, and the arc flame stabilization device is used to collect the current waveform information and arc flame state characteristic information generated during the welding process. The intelligent protective cabin is installed on the outside of the execution module and adopts a flexible airflow support structure with adjustable air pressure. It is used to adjust the airflow direction, airflow speed and local wind pressure distribution inside the cabin after receiving the protection adjustment command. The data acquisition module is used to collect wind speed, temperature and humidity, dust density, and environmental acoustic information in high-wind and sandy environments. The fusion module is used to perform unified time alignment and feature mapping on environmental disturbance feature data, welding process status data and pose information, and generate welding parameter adjustment instructions, welding posture correction instructions and protection adjustment instructions based on time-series feature modeling and state prediction mechanism. The collaboration module is used to receive video data of the welding process, environmental disturbance characteristic data, and welding parameter log data, and send remote decision information to the fusion module; The fusion module is specifically used for: The attitude offset vector is calculated and output based on the pose information. The attitude offset vector is used to characterize the influence trend of the attitude change of the bearing platform on the end direction of the welding tool assembly. The attitude compensation requirement is calculated and output based on the attitude offset vector and welding process state data. Welding attitude correction command is generated and output based on the attitude compensation requirement. The protection and countermeasure vector is calculated and output based on the environmental disturbance characteristic data. The protection and countermeasure vector is used to characterize the main intrusion direction and intensity of external wind and sand disturbance on the arc flame area. Based on the attitude adjustment direction, attitude compensation requirement and protection and countermeasure vector represented in the welding attitude correction command, the airflow coordination target is constructed and output. Based on the airflow coordination target, a protection adjustment command is generated. The protection adjustment command includes at least airflow direction parameters, airflow velocity parameters, and local wind pressure distribution parameters. The protection adjustment command is output and sent to the intelligent protection cabin to drive its adjustable air pressure flexible airflow support structure to form an airflow field that matches the attitude adjustment direction. The fusion module is specifically used for: The system acquires wind speed, temperature and humidity, dust density, and environmental acoustic information from the acquisition module, generates and outputs the original environmental disturbance sequence with a unified timestamp. The current waveform information and arc flame state characteristic information output by the arc flame stabilization device in the execution module are obtained, and the original sequence of the welding process with a unified timestamp is generated and output. The timestamp of the original sequence of the welding process is aligned with the timestamp of the original sequence of environmental disturbance. The pose information output by the carrier platform is obtained, and the pose information is interpolated or maintained based on the timestamp of the original environmental disturbance sequence. A pose alignment sequence corresponding to the original environmental disturbance sequence at each time step is generated and the pose alignment sequence is output. Based on the original environmental disturbance sequence, the original welding process sequence, and the pose alignment sequence, a fusion sample within the same time window is constructed. The fusion sample is then mapped into a multi-source feature vector sequence that includes wind and sand disturbance intensity features, welding stability features, and attitude offset features. The multi-source feature vector sequence serves as the input for subsequent time-series feature modeling and state prediction mechanisms.

2. The multi-source information fusion intelligent welding robot system for high-wind and sandy environments according to claim 1, characterized in that, The fusion module is specifically used for: The environmental disturbance risk index is calculated based on the multi-source feature vector sequence and then output. Based on the environmental disturbance risk index, a prediction window is selected from the multi-source feature vector sequence and the window length is determined. The prediction window description information is output, wherein the prediction window description information includes the window start and end time and the window length and is used to limit the range of subsequent prediction input. Within the scope of the prediction window description information, time-series feature modeling is performed on the multi-source feature vector sequence to obtain and output the welding stability prediction results; The control response type is determined based on the welding stability prediction results and environmental disturbance risk indicators, and the control response type determination result is output. The control response type determination result is used to select the instruction combination. Based on the control response type determination result, an instruction set containing at least welding parameter adjustment instructions is generated. When the control response type meets the attitude compensation trigger condition, a welding attitude correction instruction is added to the instruction set. When the control response type meets the protection enhancement trigger condition, a protection adjustment instruction is added to the instruction set. The instruction set is output and sent to the execution module and the intelligent protection cabin respectively for coordinated control.

3. The multi-source information fusion intelligent welding robot system for high-wind and sandy environments according to claim 1, characterized in that, The fusion module is specifically used for: The disturbance mutation index is calculated and output based on environmental disturbance characteristic data; The disturbance mutation index is compared with a preset change threshold to generate a participating switching quantity and output the participating switching quantity. The participating switching quantity is used to control whether the arc flame state feature information enters the subsequent prediction input. When participation is permitted by the switch quantity indication, arc flame status feature information is extracted from the welding process status data and arc flame enhancement feature is output. The arc flame enhancement feature and the disturbance mutation index together constitute an anomaly confirmation input. When participation in the switch quantity indication is not allowed, an alternative confirmation feature is generated and output. The alternative confirmation feature is calculated from the fluctuation amplitude or rate of change of the current waveform information and is used to replace the arc flame enhancement feature. An anomaly determination result is generated and output based on the anomaly confirmation input or alternative confirmation feature. The anomaly determination result is used as the constraint input of the time-series feature modeling and state prediction mechanism to affect the generation of welding parameter adjustment command, welding posture correction command and protection adjustment command.

4. The multi-source information fusion intelligent welding robot system for high-wind and sandy environments according to claim 1, characterized in that, The collaboration module is specifically used for: The system receives video data of the welding process, environmental disturbance feature data, and welding parameter log data, and outputs a remote situational awareness packet. The remote situational awareness packet includes a timestamp, video segment index, and corresponding environmental disturbance feature summary. It then transmits the remote situational awareness packet to a remote decision-making terminal and receives remote decision information returned by the terminal. A constraint description is generated and output, wherein the constraint description includes at least parameter range constraints, output priority constraints, and allowed control response type constraints. The constraint description is sent to a fusion module, which converts it into constraint inputs and outputs constraint inputs. These constraint inputs define the generation boundaries of welding parameter adjustment commands, welding posture correction commands, and protection adjustment commands. The fusion module is specifically used for: When generating welding parameter adjustment commands, welding posture correction commands, and protection adjustment commands, the consistency verification of the command set to be generated is first performed based on the constraint input, and the verification result is output. Then, the command set to be generated is trimmed or downgraded according to the verification result to output the final command set. The final command set is then sent to the execution module and the intelligent protection cabin for control.

5. The multi-source information fusion intelligent welding robot system for high-wind and sandy environments according to claim 1, characterized in that, The intelligent protective cabin is specifically used for: The system receives protection adjustment commands output by the fusion module, and parses the target airflow coordination parameter set based on the protection adjustment commands. The target airflow coordination parameter set includes at least the target airflow dominant direction, the target airflow velocity range, and the target local wind pressure distribution description, and outputs the target airflow coordination parameter set. Based on the target airflow coordination parameter set, a partition mapping is performed on the flexible airflow support structure inside the intelligent protective cabin to generate and output the airflow action area division result. The airflow action area division result is used to clarify the airflow adjustment priority corresponding to different cabin areas. Based on the division of airflow action areas, calculate the initial air pressure setting value for each airflow action area, generate an area initial air pressure configuration table, and output the area initial air pressure configuration table. Based on the regional initial air pressure configuration table and the target local wind pressure distribution description in the target airflow coordination parameter set, air pressure difference redistribution is performed on each airflow action area to generate regional air pressure difference adjustment results and output the regional air pressure difference adjustment results. The regional air pressure difference adjustment results are used to drive the flexible airflow support structure to form a non-uniform air pressure field in the cabin. Under the constraint of the regional air pressure difference adjustment result, the airflow guiding unit in the intelligent protective cabin is controlled to change the airflow release direction and release intensity, generate real-time airflow guiding state and output the real-time airflow guiding state, so that the airflow in the cabin forms a directional airflow field around the welding tool assembly that matches the welding posture adjustment direction. Based on the real-time airflow guidance status, the stability of the cabin air pressure is continuously monitored and the adjustment result of the regional air pressure difference is closed-loop corrected to generate and output the airflow stability correction amount. The airflow stability correction amount is used to correct the air pressure output of the flexible airflow support structure in real time, so as to maintain the continuity and stability of the directional airflow field within the effective period of the protection adjustment command.

Images