A laser welding device for high-pressure outer cylinder machining

Abstract

This invention discloses a laser welding device for high-pressure outer cylinder processing, belonging to the field of high-pressure outer cylinder processing technology. It includes basic components such as a base, a robotic arm, control equipment, and a laser welder, and also features an adaptive curved surface positioning fixture, a real-time monitoring and intelligent parameter control module for the welding process, and a software communication technology module. The robotic arm and the laser welder form a multi-degree-of-freedom laser welding head, adaptable to complex curved surface welding; the adaptive curved surface positioning fixture enables precise clamping of the high-pressure outer cylinder; the software communication technology module constructs a closed-loop control architecture, combining dual algorithms to dynamically optimize welding parameters, along with edge computing and cloud-based collaborative optimization. This invention solves the problems of difficult positioning and large deformation in traditional welding, improving welding accuracy, weld quality, and processing efficiency, and meeting the strength and sealing requirements of high-pressure outer cylinders under high-pressure conditions.

Description

Technical Field

[0001] This invention relates to the field of high-pressure outer cylinder processing technology, and in particular to a laser welding device for high-pressure outer cylinder processing. Background Technology

[0002] As a core large component of critical equipment such as high-pressure vessels and steam turbines, the high-pressure outer cylinder is large in size, has an arc-shaped outer wall, and is heavy. Its processing quality directly determines the operational stability and safety of the equipment under high-pressure conditions. In the welding process of the high-pressure outer cylinder, traditional processing methods mostly rely on arc welding or manual welding, which have many technical bottlenecks and cannot meet the stringent requirements for cylinder strength and sealing under high-pressure conditions.

[0003] Traditional arc welding involves concentrated and uneven heat distribution during the welding process, which can easily lead to significant welding deformation of the high-pressure outer cylinder, resulting in a decrease in the dimensional accuracy of the cylinder body and directly affecting the subsequent assembly accuracy and equipment operation reliability. Manual welding is not only inefficient, but also affected by human factors such as the welder's skill level and operational stability. It results in poor weld density and insufficient quality consistency, making it impossible to form a uniform and stable welded structure. Under high pressure, stress concentration is likely to occur, reducing the overall strength and sealing performance of the cylinder body.

[0004] Regarding the positioning and clamping issues of large high-pressure outer cylinders, traditional support platforms lack a dedicated positioning structure adapted to curved outer walls. This makes the cylinder prone to displacement after placement, hindering rapid and accurate initial positioning. Furthermore, the traditional support platform and cylinder have a rigid sliding contact, resulting in high friction. When adjusting the workpiece welding position and calibrating the joints, operators must expend considerable manpower to overcome frictional resistance, increasing labor intensity and increasing the risk of insufficient alignment accuracy due to manual adjustments, further affecting weld quality. In addition, traditional welding methods lack real-time monitoring and dynamic parameter control mechanisms for the welding process. They cannot adjust process parameters based on real-time data such as welding temperature and deformation, making it difficult to effectively control welding stress. This further exacerbates the risk of welding deformation and weld defects, failing to meet the requirements for high-pressure outer cylinders under high-pressure conditions. Summary of the Invention

[0005] This invention aims to solve the problems of large welding deformation, poor weld density, low welding accuracy, difficulty in positioning and clamping large cylinders, and difficulty in controlling welding stress in the processing of high-pressure outer cylinders using traditional welding methods, so as to meet the strength and sealing requirements of high-pressure outer cylinders under high-pressure conditions.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A laser welding device for processing high-pressure outer cylinders includes a base, a robotic arm, a control device, a laser welder, a rotating tooling fixture, and a frame. Its distinguishing feature is that it further includes an adaptive curved surface positioning fixture, a real-time monitoring and intelligent parameter control module for the welding process, and a software communication technology module integrated into the control device. The robotic arm and the laser welder work together to form a multi-degree-of-freedom laser welding head, which is adapted to the large and complex curved surface of the high-pressure outer cylinder to complete automated welding operations. The adaptive curved surface positioning fixture adaptively fits the curved surface of the high-pressure outer cylinder, achieving precise clamping of the high-pressure outer cylinder; the real-time monitoring and intelligent parameter control module for welding process interacts bidirectionally with the soft communication technology module, used to collect physical quantity data of the welding process and transmit it to the soft communication technology module. The aforementioned soft-connect technology module constructs a closed-loop control architecture for acquisition, analysis, processing, and execution. After intelligently analyzing and processing the received physical quantity data, it outputs control commands to drive the coordinated action of each execution component, dynamically optimizes the laser welding process parameters, suppresses welding deformation, and improves the strength and sealing of the weld. In a preferred embodiment, the adaptive curved surface positioning fixture includes a support fixed to the top of the base, a movable block slidably disposed within a rectangular opening on the top surface of the support, and an electric actuator for driving the movable block to rise and fall. A support plate is fixedly connected to the top of the movable block, and a mounting seat is provided at the fixed end of the electric actuator. The mounting seat is fixed to the bottom surface of the support by fasteners, and the telescopic end of the electric actuator is connected to the movable block for adjusting the height of the support plate.

[0007] In a preferred embodiment, the top surface of the support plate is symmetrically and uniformly fixedly connected with a plurality of inclined blocks, the inclined surfaces of which are adapted to the arc-shaped outer wall of the high-pressure outer cylinder to achieve the initial positioning of the high-pressure outer cylinder.

[0008] In a preferred embodiment, each inclined block has multiple ball grooves on its inclined surface, and each ball groove is movably embedded with a ball, which converts the sliding friction between the high-pressure outer cylinder and the inclined block into rolling friction.

[0009] In a preferred embodiment, the iSoftStone technology module adopts a layered architecture design, including a data acquisition layer, an intelligent analysis and processing layer, and an execution control layer that interact sequentially. The data acquisition layer interfaces with the welding process real-time monitoring and intelligent parameter control module to acquire data. The intelligent analysis and processing layer is the core layer of the iSoftStone technology module, which completes the preprocessing, feature extraction, and algorithm calculation of the acquired data. The execution control layer converts the calculation results into standardized control commands and sends them to each execution component.

[0010] In a preferred embodiment, the data acquisition layer is configured with a data receiving unit, a data preprocessing unit, and a data storage unit. The data receiving unit receives data on welding area temperature, welding deformation, and contact pressure distribution between the tooling and the high-pressure outer cylinder collected by various sensors in the real-time monitoring and intelligent parameter control module of the welding process. The data preprocessing unit performs noise reduction, normalization, and outlier removal on the collected raw data. The data storage unit stores the preprocessed valid data to form a welding process database.

[0011] In a preferred embodiment, the intelligent analysis and processing layer has a built-in welding parameter optimization algorithm model based on a BP neural network. The algorithm model takes the temperature of the welding area, the deformation of the welding part, and the contact pressure distribution as input features, and the welding power, welding speed, and rotational angular velocity of the rotating tooling fixture as output features. The input layer vector of the BP neural network is ,in This is the normalized value of the temperature in the welding area. This is the normalized value of the deformation at the welded part. The normalized value for contact pressure distribution; the hidden layer output is ,in This is the weight matrix from the input layer to the hidden layer. Let be the hidden layer bias vector, and let be the ReLU activation function, expressed as follows: The output layer output is ,in This is the weight matrix from the hidden layer to the output layer. The output layer bias vector is given, and the output layer activation function is the Sigmoid function, expressed as follows: ; The loss function of the algorithm model is the mean squared error function, expressed as follows: ,in Predict the output value for the algorithm model. Here, n represents the optimal reference values ​​for the welding process parameters, and n is the number of samples. The BP neural network uses gradient descent to update and optimize the weights and biases. The weight update formula is as follows: The bias update formula is: , where η is the learning rate and t is the number of iterations, until the loss function value converges to the preset threshold, and the optimal laser welding process parameters are output.

[0012] In a preferred embodiment, the execution control layer is configured with an instruction generation unit, an instruction issuing unit, and an execution feedback unit. The instruction generation unit converts the optimal welding process parameters output by the intelligent analysis and processing layer into standardized control instructions for each execution component. The instruction issuing unit uses an industrial communication protocol to issue control instructions to the robot, laser welder, rotary tooling fixture, and electric actuator in real time. The execution feedback unit collects the action response data of each execution component in real time and sends it back to the intelligent analysis and processing layer to achieve closed-loop verification of the control instruction execution effect.

[0013] In a preferred embodiment, the iSoftStone technology module adopts an edge computing and cloud collaboration approach. The control device has a built-in edge computing unit, which performs local real-time analysis and processing of welding process data and outputs immediate control commands. The edge computing unit establishes a wireless communication connection with the cloud server, synchronizing the welding process database to the cloud server. The cloud server uses big data analysis algorithms to globally optimize the welding process parameters and sends the optimized parameter model to the edge computing unit, thereby realizing the iterative update of the iSoftStone technology module's algorithm model.

[0014] In a preferred embodiment, the intelligent analysis and processing layer also incorporates a fuzzy PID control algorithm as a backup welding parameter optimization algorithm, used to achieve rapid adjustment of welding parameters when the BP neural network algorithm model has not converged; the fuzzy PID control algorithm takes the deformation deviation e and the deviation change rate ec of the welding part as inputs, and the proportional coefficient of the PID controller as input. Integral coefficient Differential coefficients For the output; by sequentially performing fuzzification, fuzzy inference, and defuzzification on e and ec using a preset fuzzy rule base, the following is obtained: The real-time correction value; the output control quantity of the PID controller is in Let be the deformation deviation at time t. The rate of change of deviation at time t is corrected in real time. The welding power and welding speed of the laser welder are dynamically adjusted.

[0015] Due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This solution achieves precise clamping through an adaptive curved surface positioning fixture. The rectangular opening on the top surface of the support base provides stable guidance for the movable block, and the electric actuator drives the support plate to rise and fall smoothly, adapting to the height requirements of different specifications of high-pressure outer cylinders. The inclined block on the support plate conforms to the arc-shaped outer wall of the high-pressure outer cylinder, achieving initial positioning. The ball bearings convert sliding friction into rolling friction, reducing the resistance during workpiece adjustment and handling. This structure solves the problem of cylinder body misalignment caused by traditional support platforms, reduces positioning time, improves clamping accuracy, and lays the foundation for subsequent welding quality.

[0016] 2. This solution utilizes a multi-degree-of-freedom laser welding head and a dynamic parameter optimization module in synergy, with a robotic arm working in conjunction with the laser welder to adapt to welding complex curved surfaces. During the welding process, temperature, displacement, and pressure sensors collect data in real time. The software-enabled module dynamically adjusts the welding power, speed, and rotational angular velocity using a BP neural network and a fuzzy PID dual algorithm. This closed-loop control architecture solves the problem of uneven heat distribution in traditional welding, suppresses welding deformation, controls welding stress, and improves weld density and strength, meeting the sealing and strength requirements under high-pressure conditions.

[0017] 3. This solution achieves high-efficiency processing through an automated architecture and cloud-based collaborative optimization. Control equipment drives the coordinated operation of various components to complete automated material loading, positioning, welding, and unloading processes. Edge computing units enable local real-time control, while the cloud server uses K-means clustering and multiple linear regression algorithms to optimize global parameters and iteratively update the model. This approach solves the efficiency and consistency problems caused by manual operation, reduces labor intensity, and continuously optimizes the process through data accumulation, improving overall processing efficiency and quality stability.

[0018] In summary, this solution comprehensively addresses the core problems of traditional high-pressure outer cylinder welding—difficult positioning, large deformation, poor weld quality, and low efficiency—through the organic integration of adaptive curved surface positioning fixtures, multi-degree-of-freedom welding heads, real-time monitoring and intelligent control modules, and a cloud-based collaborative architecture. It achieves precise clamping, automated welding, and dynamic optimization of process parameters for large, complex curved surface high-pressure outer cylinders, effectively improving welding accuracy, weld quality, and processing efficiency, reducing labor intensity, and ensuring the performance of high-pressure outer cylinders under high-pressure conditions. This provides an efficient and reliable technical solution for high-pressure outer cylinder processing. Attached Figure Description

[0019] Figure 1 A schematic diagram of the overall structure provided according to an embodiment of the present invention is shown; Figure 2 A schematic diagram of the support base provided according to an embodiment of the present invention is shown; Figure 3 A structural schematic diagram of the support seat from an elevation perspective provided according to an embodiment of the present invention is shown; Figure 4 A schematic diagram of the structure of the tilting block provided according to an embodiment of the present invention is shown; Figure 5 A block diagram of the overall control logic connection of the device according to an embodiment of the present invention is shown; Figure 6 A block diagram showing the hierarchical architecture connection of the iSoftStone technology module according to an embodiment of the present invention is shown. Figure 7 A data acquisition layer unit connection diagram according to an embodiment of the present invention is shown; Figure 8 A block diagram showing the logical connection of the intelligent analysis and processing layer algorithm provided according to an embodiment of the present invention is shown. Figure 9 A block diagram of the execution control layer unit connection provided according to an embodiment of the present invention is shown.

[0020] Legend: 1. Base; 2. Support base; 3. Robot arm; 4. Control equipment; 5. Laser welder; 6. Rotary tooling fixture; 7. Frame; 8. Support plate; 9. Inclined block; 10. Movable block; 11. Mounting base; 12. Electric actuator; 13. Ball groove; 14. Ball. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figures 1-9 This invention provides a laser welding device for high-pressure outer cylinder processing. Its mechanical support and positioning execution structure includes a base 1, a frame 7, and an adaptive curved surface positioning fixture. It works in conjunction with a robot arm 3, a laser welder 5, a rotating fixture 6, and a control device 4 to complete the overall welding operation. The connection relationship, cooperation method, and function of each mechanical structure are as follows:

[0023] The frame 7 is fixedly connected to both ends of the base 1 by bolts. The robot arm 3 is fixedly installed on the top surface of the frame 7 by bolts. The laser welder 5 is fixedly installed on the robot arm 3. The two work together to form a multi-degree-of-freedom laser welding head, which can be adapted to the large and complex curved surface of the high-pressure outer cylinder to complete the automated welding operation. The rotating tooling fixture 6 is installed in the middle of the base 1 for clamping and fixing the high-pressure outer cylinder and driving its rotation. The control device 4 is installed on both sides of the base 1. The control device 4 adopts a PLC controller and is equipped with a touch operation panel. It has the characteristics of high control accuracy, stable operation and strong anti-interference ability. It can accurately control the operating parameters of each mechanical component and realize coordinated linkage.

[0024] The adaptive curved surface positioning fixture provides precise support and positioning for the high-pressure outer cylinder, and is fixed to the top of the base 1. It includes a support base 2, a movable block 10, an electric actuator 12, a support plate 8, a mounting base 11, an inclined block 9, and ball bearings 14. The support base 2 is fixed to the base 1 by bolts. A rectangular opening is provided on the top surface of the support base 2 to provide stable sliding guidance for the movable block 10. The two are fitted with a clearance to ensure smooth up-and-down sliding of the movable block 10 while preventing wobbling during sliding. The movable block 10 slides within this rectangular opening. Inside the opening, its top is fixedly connected to the support plate 8, and its bottom is connected to the telescopic end of the electric push rod 12. The fixed end of the electric push rod 12 is provided with a mounting seat 11. The mounting seat 11 is fixed to the bottom surface of the support seat 2 by bolt fasteners, which can increase the contact area between the electric push rod 12 and the support seat 2, making the electric push rod 12 more stable. As a lifting power component, the electric push rod 12 can drive the movable block 10 and the support plate 8 to rise and fall smoothly, realize the precise adjustment of the height of the high pressure outer cylinder, and adapt to the processing needs of high pressure outer cylinders of different specifications and welding positions.

[0025] Multiple inclined blocks 9 are symmetrically and evenly fixedly connected to the top surface of the support plate 8. The inclined surface of the inclined block 9 is adapted to the arc-shaped outer wall of the high-pressure outer cylinder, so that the high-pressure outer cylinder can naturally fit the inclined surface after placement, realizing the initial positioning of the high-pressure outer cylinder and avoiding displacement during placement. Multiple ball grooves 13 are opened on the inclined surface of each inclined block 9, and ball 14 is movably embedded in each ball groove 13. The ball 14 is made of bearing steel that has been quenched, with high hardness and strong wear resistance. It can convert the sliding friction between the high-pressure outer cylinder and the inclined block 9 into rolling friction, which greatly reduces the friction force during workpiece placement, position adjustment and picking, making it easy for operators to quickly calibrate the position of the high-pressure outer cylinder. At the same time, it avoids friction scratches on the outer wall of the high-pressure outer cylinder, protects the surface quality of the workpiece, and can bear the weight of the high-pressure outer cylinder, avoiding wear and deformation due to long-term use, and ensuring the long-term stable operation of the device.

[0026] The robotic arm 3, laser welder 5, rotating fixture 6, and electric actuator 12 are all electrically connected to the control device 4. The control device 4 can receive control commands output by the soft-connect technology module, driving each mechanical component to operate collaboratively according to preset logic to complete the laser welding operation of the high-pressure outer cylinder. It can precisely control parameters such as the movement trajectory of the robotic arm 3, the power of the laser welder 5, the rotation speed of the rotating fixture 6, and the lifting height of the electric actuator 12, realizing the coordinated linkage of each component and ensuring the orderly progress of the welding process. The matching touch operation panel makes it easy for operators to set welding parameters, start and stop welding operations, and can also display the operating status of each mechanical component in real time, making it easy to detect and handle faults in a timely manner, improving the ease of operation and reliability of the device.

[0027] The real-time monitoring and intelligent parameter control module for the welding process interacts bidirectionally with the Softcom technology module. This module collects physical quantity data from the welding process and transmits it to the Softcom technology module. Specifically, this module includes pressure sensors, temperature sensors, and displacement sensors. The pressure sensor is located at the contact point between the adaptive curved surface positioning fixture and the high-pressure outer cylinder to collect the contact pressure distribution data between the fixture and the high-pressure outer cylinder. The temperature sensor is located in the welding area of ​​the multi-degree-of-freedom laser welding head to collect real-time temperature data of the welding area. The displacement sensor is located at the welding point of the high-pressure outer cylinder to collect real-time deformation data of the welding point. All sensors are electrically connected to the control device 4 to achieve real-time transmission of monitoring data.

[0028] The iSoftStone technology module constructs a closed-loop control architecture for acquisition, analysis, processing, and execution. After intelligently analyzing and processing the received physical quantity data, it outputs control commands to drive the coordinated action of various execution components, dynamically optimizes laser welding process parameters, suppresses welding deformation, and improves the strength and sealing of the weld. The iSoftStone technology module adopts a layered architecture design, including a data acquisition layer, an intelligent analysis and processing layer, and an execution control layer that interact sequentially. Furthermore, the iSoftStone technology module adopts an edge computing and cloud collaboration implementation method. The control device 4 has a built-in edge computing unit, which completes local real-time analysis and processing of welding process data and outputs immediate control commands. At the same time, the edge computing unit establishes a wireless communication connection with the cloud server, which can synchronize the welding process database to the cloud server. The cloud server uses big data analysis algorithms to globally optimize the welding process parameters and sends the optimized parameter model down to the edge computing unit, realizing the iterative update of the iSoftStone technology module's algorithm model.

[0029] The data acquisition layer is equipped with a data receiving unit, a data preprocessing unit, and a data storage unit. The data receiving unit interfaces with various sensors of the real-time welding process monitoring and intelligent parameter control module to receive data on welding area temperature, welding deformation, and contact pressure distribution between the tooling and the high-pressure outer cylinder. The data preprocessing unit performs noise reduction, normalization, and outlier removal on the collected raw data to eliminate data interference and ensure data validity. The data storage unit classifies and stores the preprocessed valid data to form a traceable and callable welding process database, providing data support for welding parameter optimization.

[0030] The intelligent analysis and processing layer is the core layer of the Softcom technology module. It completes the preprocessing, feature extraction and algorithm operation of the collected data. It has a built-in welding parameter optimization algorithm model based on BP neural network, and also has a built-in fuzzy PID control algorithm as a backup welding parameter optimization algorithm, which is used to realize the rapid adjustment of welding parameters when the BP neural network algorithm model has not converged.

[0031] The welding parameter optimization algorithm model based on a BP neural network takes the welding zone temperature, welding deformation, and contact pressure distribution as input features, and the welding power, welding speed, and rotational angular velocity of the rotating fixture 6 of the laser welder 5 as output features; the input layer vector of the BP neural network is... ,in This is the normalized value of the temperature in the welding area. This is the normalized value of the deformation at the welded part. The normalized value for contact pressure distribution; the hidden layer output is ,in This is the weight matrix from the input layer to the hidden layer. Let be the hidden layer bias vector, and let be the ReLU activation function, expressed as follows: Where x is the input value of the hidden layer neuron, and the function's role is to retain non-negative inputs, suppress negative inputs, and prevent gradient vanishing; the output layer output is... ,in This is the weight matrix from the hidden layer to the output layer. The output layer bias vector is given, and the output layer activation function is the Sigmoid function, expressed as follows: ; The loss function of the algorithm model is the mean squared error function, expressed as follows: Where L represents the loss function value, used to measure the degree of deviation between the algorithm model's predicted output and the optimal reference value. This is the predicted output value of the algorithm model for the i-th sample (i.e., the predicted values ​​of laser welding power, welding speed, and rotational angular velocity). Let n be the optimal reference value of the welding process parameters corresponding to the i-th sample, and n be the number of samples. The BP neural network uses gradient descent to update and optimize the weights and biases. The weight update formula is as follows: The bias update formula is: η is the learning rate, which physically represents the step size for parameter updates. It controls the update magnitude of weights and biases and has a range of (0, 1]. The value needs to be determined based on actual operating conditions. This represents the weight matrix for the next iteration (t+1) after the t-th iteration, i.e., the updated weight matrix. Represents the weight matrix at the t-th iteration ( or ), where t is the number of iterations in the BP neural network. The partial derivative variable representing the loss function L represents a small change in the loss function. Denotes the weight matrix of the t-th iteration. The partial differential variables represent small changes in the weight matrix. The gradient of the loss function L with respect to the weight matrix Wt in the t-th iteration reflects the rate and direction of change of the loss function with respect to the weights. The gradient direction is the direction in which the loss function increases; therefore, the formula uses a "minus sign" to update the weights along the direction in which the loss function decreases. Let represent the bias vectors at the t-th or t+1-th iteration, respectively. Let represent the partial differential variable of the bias vector Bt in the t-th iteration, representing the small change in the bias vector. The gradient of the loss function L with respect to the bias vector Bt in the t-th iteration is represented, reflecting the rate and direction of change of the loss function with the bias. The iterative update process continues until the loss function value L converges to the preset threshold, at which point the iteration stops and the optimal laser welding process parameters are output.

[0032] The intelligent analysis and processing layer also incorporates a fuzzy PID control algorithm as a backup welding parameter optimization algorithm, used to achieve rapid adjustment of welding parameters when the BP neural network algorithm model has not converged. The fuzzy PID control algorithm takes the deformation deviation e and the deviation change rate ec of the welded part as inputs, and the proportional coefficient of the PID controller as input. Integral coefficient Differential coefficients For the output; by sequentially performing fuzzification, fuzzy inference, and defuzzification on e and ec using a preset fuzzy rule base, the following is obtained: The real-time correction value; the output control quantity of the PID controller is In the formula, Let be the deformation deviation at time t. Let be the rate of change of the deviation at time t. This term represents the integration operation, where "∫" is the integration operator, "0" is the start time of integration (the start time of welding operation), "t" is the end time of integration (the current time), "τ" is the integration variable (time element), "e(τ)" is the deformation deviation of the welded part at time τ, and "dτ" is the derivative of the integration variable τ. The overall meaning of this term is to accumulate all deformation deviations from the start of welding to the current time. This represents the differential operation term, where "de(t)" is the small change in the deformation deviation at time t, and "dt" is the small change in time. The overall meaning of this term is the rate of change of the deformation deviation at the welded part at time t, i.e., ec(t), which is used to reflect the trend of the deviation. Through real-time correction The welding power and welding speed of the laser welder 5 are dynamically adjusted to achieve rapid suppression of welding deformation.

[0033] It should be further explained that the fuzzy rule base of the fuzzy PID control algorithm is established based on the process characteristics and engineering test data of high-pressure outer cylinder laser welding. Multiple sets of fuzzy control rules are formulated by combining the influence law of welding deformation deviation and the sensitivity of welding parameter adjustment to achieve precise mapping between deviation and parameter correction. The fuzzy universes of discourse for both deviation e and deviation change rate ec are set to {-3, -2, -1, 0, 1, 2, 3}, and the corresponding fuzzy linguistic variables are {negative large, negative medium, negative small, zero, positive small, positive medium, positive large}. The proportional coefficient of the PID controller... Integral coefficient Differential coefficients The fuzzy universes are all set to {0, 1, 2, 3, 4, 5, 6}, and the fuzzy linguistic variables are {zero, small, medium-small, medium, medium-large, large, maximal}. Fuzzy inference employs the Mamdani inference method, and the centroid method is used for clarification, obtained by calculating the centroid of the fuzzy set. The precise correction value enables rapid and accurate control of welding parameters.

[0034] The execution control layer is configured with an instruction generation unit, an instruction issuance unit, and an execution feedback unit. The instruction generation unit converts the optimal welding process parameters output by the intelligent analysis and processing layer into standardized control instructions for each execution component, adapting to the motion control logic of the robot arm 3, laser welder 5, rotary tooling fixture 6, and electric actuator 12. The instruction issuance unit uses industrial communication protocols to issue control instructions to each execution component in real time, ensuring the real-time performance and stability of instruction transmission. Specific industrial communication protocols include Profinet, Modbus, and EtherNet / IP. The execution feedback unit collects the motion response data of each execution component in real time and sends it back to the intelligent analysis and processing layer, realizing closed-loop verification of the control instruction execution effect. If an execution deviation is detected, the intelligent analysis and processing layer can adjust the control instructions immediately to ensure precise control of the welding process.

[0035] The robotic arm 3, laser welder 5, rotating tooling fixture 6, and electric actuator 12 are all electrically connected to the control device 4. The control device 4 receives control commands output by the soft communication technology module and drives each component to operate collaboratively according to preset logic to complete the laser welding operation of the high-pressure outer cylinder.

[0036] It should be further explained that the real-time monitoring and intelligent parameter control module for the welding process includes a pressure sensor, a temperature sensor, and a displacement sensor. The pressure sensor is located at the contact point of the adaptive curved surface positioning fixture and is used to collect the contact pressure distribution data between the fixture and the high-pressure outer cylinder. The temperature sensor is located in the welding operation area of ​​the multi-degree-of-freedom laser welding head and is used to collect real-time temperature data of the welding area. The displacement sensor is located at the welding point of the high-pressure outer cylinder and is used to collect real-time deformation data of the welding point.

[0037] The working principle of this invention is as follows: Device debugging preparation: The staff completes the device initialization through the touch operation panel of the control device 4, starts the electric push rod 12, and the electric push rod 12 drives the movable block 10 and the support plate 8 to descend to the preset feeding height. At the same time, the various sensors of the welding process real-time monitoring and parameter intelligent control module complete self-test, and the soft communication technology module enters the data receiving ready state.

[0038] Workpiece loading and positioning: The high-pressure outer cylinder is placed on the inclined block 9 of the support plate 8. The inclined structure of the inclined block 9 is used to achieve the initial positioning of the high-pressure outer cylinder. The position of the high-pressure outer cylinder is quickly adjusted by the rolling action of the ball bearing 14 so that the welding part is initially aligned with the laser welder 5. At this time, the pressure sensor collects the contact pressure distribution data between the tooling and the high-pressure outer cylinder and transmits it to the soft communication technology module. After analyzing the data, the soft communication technology module outputs control commands to drive the electric push rod 12 to finely adjust the height of the support plate 8 so that the inclined block 9 and the curved surface of the high-pressure outer cylinder are closely fitted. Then, the control device 4 drives the rotating tooling fixture 6 to clamp and fix the high-pressure outer cylinder, completing the precise clamping.

[0039] Welding parameter preset: The operator presets parameters such as the initial welding power, welding speed, and initial rotational angular velocity of the rotating tooling fixture 6 of the laser welder 5 according to the material, thickness, and welding process requirements of the high-pressure outer cylinder through the control equipment 4. The softcom technology module retrieves the historical best parameters from the welding process database as a reference for the algorithm model.

[0040] Automated welding operation: The control device 4 starts the robot arm 3 and the laser welder 5. The robot arm 3 drives the laser welder 5 to move to the welding start position. The laser welder 5 starts and begins welding. At the same time, the rotating tooling fixture 6 drives the high-pressure outer cylinder to rotate at a preset angular velocity to realize continuous welding of the circumferential weld. During the welding process, the temperature sensor and displacement sensor collect the temperature of the welding area and the deformation of the welding part in real time and transmit them to the soft communication technology module.

[0041] Dynamic optimization of welding parameters: The data acquisition layer of the soft-connect technology module preprocesses the received monitoring data and then transmits it to the intelligent analysis and processing layer. The intelligent analysis and processing layer performs calculations and analysis on the data through a BP neural network algorithm model, and outputs the optimal welding process parameters in real time. The execution control layer converts these parameters into control commands and sends them to each execution component to dynamically adjust the welding power, welding speed of the laser welder 5, and rotational angular velocity of the rotating fixture 6. If the BP neural network algorithm model fails to converge, it immediately switches to a fuzzy PID control algorithm to achieve rapid adjustment of welding parameters, ensuring that the temperature and deformation of the welding area are always within the preset reasonable range, suppressing welding deformation, and controlling welding stress.

[0042] Welding closed-loop verification: The execution feedback unit collects the motion response data of the execution components such as the robot arm 3 and the laser welder 5 in real time, and sends it back to the intelligent analysis and processing layer for closed-loop verification. If a motion deviation is detected, the control command is adjusted immediately to ensure that the welding head is always accurately aligned with the welding part, thereby improving welding accuracy.

[0043] Finishing and unloading: After welding is completed, control equipment 4 controls laser welder 5 to stop emitting light, robot arm 3 drives laser welder 5 back to the initial position, rotating fixture 6 stops rotating and releases high-pressure outer cylinder; electric push rod 12 drives support plate 8 to descend to unloading height, and the worker quickly removes the welded high-pressure outer cylinder by the rolling action of ball bearing 14, completing one welding operation; at the same time, the softcom technology module stores the process parameters and monitoring data of this welding to the welding process database and synchronizes it to the cloud server, providing data support for subsequent global optimization of welding parameters.

[0044] It should be further explained that the big data analysis algorithm on the cloud server is a combination algorithm that integrates K-means clustering and multiple linear regression. First, the K-means clustering algorithm is used to perform cluster analysis on the massive process parameters, monitoring data and weld quality inspection data in the welding process database to divide the welding data clusters corresponding to high-pressure outer cylinders of different materials and specifications. Then, based on each data cluster, a mathematical model is constructed using the multiple linear regression algorithm to connect the welding process parameters with the welding deformation, weld density and weld strength. The optimization objective is multi-objective collaborative optimization, namely minimizing the welding deformation of the high-pressure outer cylinder and maximizing the weld density and weld strength, while constraining parameters such as laser welding power and welding speed within the allowable range of the process. The cloud server performs global optimization of the welding process parameters based on this model to obtain the optimal process parameter model under different working conditions and distributes it to the edge computing unit.

[0045] In this embodiment, the coordinated operation of various components enables precise clamping and automated welding of the large and complex curved surface of the high-pressure outer cylinder, solving problems such as difficult positioning and clamping, low welding accuracy, and large welding deformation in traditional welding methods. Through the closed-loop control architecture of the soft-connect technology module, real-time monitoring of the welding process and intelligent dynamic optimization of welding parameters are realized, effectively controlling welding stress and improving the density, strength, and sealing of the weld, meeting the usage requirements of the high-pressure outer cylinder under high-pressure conditions. At the same time, the setting of ball bearing 14 greatly reduces the labor intensity of workpiece loading, adjustment, and unloading, improving the overall efficiency of high-pressure outer cylinder processing.

[0046] The above description of the embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A laser welding device for processing high-pressure outer cylinders, comprising a base (1), a robotic arm (3), a control device (4), a laser welder (5), a rotating tooling fixture (6), and a frame (7), characterized in that, It also includes an adaptive curved surface positioning fixture, a real-time monitoring and intelligent parameter control module for the welding process, and a soft communication technology module integrated in the control device (4); the robot (3) and the laser welder (5) work together to form a multi-degree-of-freedom laser welding head, which is adapted to the large and complex curved surface of the high-pressure outer cylinder to complete the automated welding operation; The adaptive curved surface positioning fixture adaptively fits the curved surface of the high-pressure outer cylinder, achieving precise clamping of the high-pressure outer cylinder; the real-time monitoring and intelligent parameter control module for welding process interacts bidirectionally with the soft communication technology module, used to collect physical quantity data of the welding process and transmit it to the soft communication technology module. The aforementioned soft-connect technology module constructs a closed-loop control architecture for acquisition, analysis, processing, and execution. After intelligently analyzing and processing the received physical quantity data, it outputs control commands to drive the coordinated actions of various execution components, dynamically optimizes laser welding process parameters, suppresses welding deformation, and improves the strength and sealing of the weld.

2. The laser welding device for high-pressure outer cylinder processing according to claim 1, characterized in that, The adaptive curved surface positioning fixture includes a support base (2) fixed to the top of the base (1), a movable block (10) slidably disposed in a rectangular opening on the top surface of the support base (2), and an electric push rod (12) for driving the movable block (10) to rise and fall. A support plate (8) is fixedly connected to the top of the movable block (10). A mounting seat (11) is provided at the fixed end of the electric push rod (12). The mounting seat (11) is fixed to the bottom surface of the support base (2) by fasteners. The telescopic end of the electric push rod (12) is connected to the movable block (10) for transmission to adjust the height position of the support plate (8).

3. The laser welding device for high-pressure outer cylinder processing according to claim 2, characterized in that, The top surface of the support plate (8) is symmetrically and uniformly fixedly connected with a plurality of inclined blocks (9). The inclined surface of the inclined block (9) is adapted to the arc-shaped outer wall of the high-pressure outer cylinder to achieve the initial positioning of the high-pressure outer cylinder.

4. The laser welding device for high-pressure outer cylinder processing according to claim 3, characterized in that, Each inclined block (9) has multiple ball grooves (13) on its inclined surface, and each ball groove (13) is movably embedded with a ball (14). The ball (14) converts the sliding friction between the high-pressure outer cylinder and the inclined block (9) into rolling friction.

5. The laser welding device for high-pressure outer cylinder processing according to claim 1, characterized in that, The iSoftStone technology module adopts a layered architecture design, including a data acquisition layer, an intelligent analysis and processing layer, and an execution control layer that interact sequentially. The data acquisition layer interfaces with the welding process real-time monitoring and intelligent parameter control module to acquire data. The intelligent analysis and processing layer is the core layer of the iSoftStone technology module, which completes the preprocessing, feature extraction, and algorithm calculation of the acquired data. The execution control layer converts the calculation results into standardized control commands and sends them to each execution component.

6. The laser welding device for high-pressure outer cylinder processing according to claim 5, characterized in that, The data acquisition layer is configured with a data receiving unit, a data preprocessing unit, and a data storage unit. The data receiving unit receives data on welding area temperature, welding deformation, and contact pressure distribution between the tooling and the high-pressure outer cylinder collected by various sensors in the real-time monitoring and intelligent parameter control module of the welding process. The data preprocessing unit performs noise reduction, normalization, and outlier removal on the collected raw data. The data storage unit stores the preprocessed valid data to form a welding process database.

7. The laser welding device for high-pressure outer cylinder processing according to claim 5, characterized in that, The intelligent analysis and processing layer has a built-in welding parameter optimization algorithm model based on BP neural network. The algorithm model takes the temperature of the welding area, the deformation of the welding part, and the contact pressure distribution as input features, and the welding power, welding speed, and rotational angular velocity of the laser welder (5) and the rotating tooling fixture (6) as output features. The input layer vector of the BP neural network is ,in This is the normalized value of the temperature in the welding area. This is the normalized value of the deformation at the welded part. The normalized value for contact pressure distribution; the hidden layer output is ,in This is the weight matrix from the input layer to the hidden layer. Let be the hidden layer bias vector, and let be the ReLU activation function, expressed as follows: The output layer output is ,in This is the weight matrix from the hidden layer to the output layer. The output layer bias vector is given, and the output layer activation function is the Sigmoid function, expressed as follows: ; The loss function of the algorithm model is the mean squared error function, expressed as follows: ,in Predict the output value for the algorithm model. Here, n represents the optimal reference values ​​for the welding process parameters, and n is the number of samples. The BP neural network uses gradient descent to update and optimize the weights and biases. The weight update formula is as follows: The bias update formula is: , where η is the learning rate and t is the number of iterations, until the loss function value converges to the preset threshold, and the optimal laser welding process parameters are output.

8. The laser welding device for high-pressure outer cylinder processing according to claim 1, characterized in that, The execution control layer is configured with an instruction generation unit, an instruction issuance unit, and an execution feedback unit. The instruction generation unit converts the optimal welding process parameters output by the intelligent analysis and processing layer into standardized control instructions for each execution component. The instruction issuance unit uses an industrial communication protocol to issue control instructions to the robot (3), laser welder (5), rotary tooling fixture (6), and electric actuator (12) in real time. The execution feedback unit collects the action response data of each execution component in real time and sends it back to the intelligent analysis and processing layer to realize closed-loop verification of the execution effect of the control instructions.

9. The laser welding device for high-pressure outer cylinder machining according to claim 1, characterized in that, The iSoftStone technology module adopts an edge computing and cloud collaboration implementation method. The control device (4) has a built-in edge computing unit. The edge computing unit completes local real-time analysis and processing of welding process data and outputs instant control commands. The edge computing unit establishes a wireless communication connection with the cloud server and synchronizes the welding process database to the cloud server. The cloud server uses big data analysis algorithms to globally optimize the welding process parameters and sends the optimized parameter model to the edge computing unit to realize the iterative update of the iSoftStone technology module algorithm model.

10. The laser welding device for high-pressure outer cylinder processing according to claim 1, characterized in that, The intelligent analysis and processing layer also incorporates a fuzzy PID control algorithm as a backup welding parameter optimization algorithm, used to achieve rapid adjustment of welding parameters when the BP neural network algorithm model has not converged; the fuzzy PID control algorithm takes the deformation deviation e and the deviation change rate ec of the welded part as inputs, and the proportional coefficient of the PID controller as input. Integral coefficient Differential coefficients For the output; by sequentially performing fuzzification, fuzzy inference, and defuzzification on e and ec using a preset fuzzy rule base, the following is obtained: The real-time correction value; the output control quantity of the PID controller is in Let be the deformation deviation at time t. The rate of change of deviation at time t is corrected in real time. The welding power and welding speed of the laser welder (5) are dynamically adjusted.

Images