Plasma welding tool for water pump impeller machining

Abstract

The invention provides a plasma welding tool for water pump impeller machining, and relates to the technical field of welding tools, and the plasma welding tool comprises a basic frame assembly, a rotating mechanism assembly, a lifting adjusting mechanism assembly, a positioning mechanism assembly and an auxiliary welding system. The auxiliary welding system comprises a main control unit, a sensor unit, a parameter storage unit, a welding parameter dynamic generation module and a man-machine interaction unit. According to the scheme, aiming at traditional tool pain points, parameters are dynamically generated through multi-sensor collection and machine learning, PID closed-loop correction is conducted, and the welding quality stability is improved; manual intervention is reduced and the period is shortened by means of automatic regulation and control of the main control unit and man-machine interaction; and different impellers are quickly adapted by virtue of the classification process database and the welding parameter dynamic generation module, so that the adjustment time consumption is reduced, the quality, efficiency and flexibility are comprehensively improved, and the production cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of welding tooling, in particular to a plasma welding tooling for processing a water pump impeller. Background Art

[0002] As the core component of the water pump, the structural integrity and welding quality of the water pump impeller directly determine the hydraulic efficiency, operational stability and service life of the water pump. During the impeller manufacturing process, the connection between the blades and the hub is mostly made by plasma welding. This process has the advantages of concentrated energy and a small heat-affected zone, but it places strict requirements on the positioning accuracy, clamping stability and adjustment flexibility of the welding tooling.

[0003] In the current water pump impeller welding process, traditional welding tooling presents numerous issues that require urgent resolution. For one thing, precision control of welding parameters is difficult to achieve, resulting in unstable weld quality and the weld being prone to defects such as porosity, cracks, and lack of fusion, seriously impacting the impeller's mechanical properties and service life.

[0004] On the other hand, traditional tooling has a low level of automation and intelligence and relies too much on manual operation. Not only is production efficiency low, but the welding quality is also significantly affected by human factors, making it difficult to meet large-scale, high-quality production needs.

[0005] At the same time, traditional tooling lacks sufficient adaptability for pump impellers of varying models and materials. Replacing tooling or adjusting parameters often requires significant time and labor costs, further restricting production efficiency and business profitability. As the manufacturing industry continues to demand higher performance from pumps, developing a welding tool that can precisely control the welding process, is highly adaptable to a variety of impellers, and possesses a high level of intelligence has become an urgent need for industry development. Therefore, it is necessary to provide a plasma welding tool for pump impeller machining to address these technical issues. Summary of the Invention

[0006] The plasma welding tooling for processing the water pump impeller of the present invention is mainly composed of a basic frame component, a rotating mechanism component, a lifting and adjusting mechanism component, a positioning mechanism component and an auxiliary welding system; each component works closely together to achieve high efficiency and high quality of plasma welding of the water pump impeller.

[0007] In order to solve the above technical problems, the present invention provides a plasma welding tool for processing a water pump impeller, comprising a basic frame assembly, a rotating mechanism assembly, a lifting and adjusting mechanism assembly, a positioning mechanism assembly and an auxiliary welding system; The auxiliary welding system is used to assist the welding process, and includes a main control unit, a sensor unit, a parameter storage unit and a human-computer interaction unit; The sensor unit includes a current sensor for monitoring welding current, a voltage sensor for monitoring welding voltage, a temperature sensor for monitoring molten pool temperature, a laser profile sensor for monitoring the microscopic topography of the impeller surface, and an acceleration sensor for monitoring the vibration state of the tooling; The parameter storage unit is used to store preset welding program parameters, a welding process database of impellers of different materials, and a welding parameter prediction model based on machine learning training; The main control unit is connected to the rotating mechanism assembly, the lifting and adjusting mechanism assembly, the positioning mechanism assembly, the sensor unit, the parameter storage unit and the human-computer interaction unit respectively, and can receive sensor data, call stored parameters and prediction models, and send control instructions to each mechanism to realize intelligent control of the welding process; The dynamic welding parameter generation module, based on the impeller surface micromorphology data collected by the laser profile sensor and the tooling vibration characteristics obtained by the acceleration sensor, uses the machine learning model in the parameter storage unit to generate personalized welding parameters adapted to the current welding conditions in real time, including dynamic adjustment of the welding current, voltage, welding gun motion trajectory and rotating mechanism speed; The human-machine interaction unit is used by the operator to set parameters, view welding status and receive system feedback information.

[0008] Preferably, the basic frame assembly includes a base plate, a positioning block and a base. The upper end of the base plate is rectangular and fixed with positioning blocks. Fixing holes are equidistantly provided on the positioning blocks. The upper end of the positioning blocks is fixed with a base. The rotating mechanism assembly is installed in the middle of the base. The positioning mechanism assembly is assembled on the rotating mechanism assembly. The lifting and adjusting mechanism assembly is installed on one side of the base.

[0009] Preferably, the parameter acquisition method of the sensor unit is as follows: The current sensor is connected in series in the welding circuit to collect the plasma arc welding current in real time; The voltage sensor is connected in parallel between the welding gun and the workpiece to collect the arc voltage; The temperature sensor uses a non-contact measurement method to aim at the molten pool area and collect the molten pool temperature; The laser profile sensor is installed on the side of the welding gun, and uses a line laser to scan the impeller surface to be welded and output surface profile point cloud data; The acceleration sensor is installed at the connection between the rotating mechanism and the base to collect the vibration acceleration of the tooling.

[0010] Preferably, in the parameter storage unit: The preset welding program parameters are stored by impeller model, including initial welding current, voltage, rotation speed and welding gun trajectory coordinates; The welding process database is stored by material classification, including the mapping relationship between materials and welding parameters; The welding parameter prediction model based on machine learning training includes a random forest model for predicting welding current and a time series prediction model for predicting molten pool temperature. It is stored in the form of a binary file and contains feature weights and training logs.

[0011] Preferably, the working process of the welding parameter dynamic generation module includes: Extract surface roughness, groove angle, and misalignment features from point cloud data collected by the laser profiler and perform standardization processing; Extract the vibration root mean square value and resonant frequency characteristics from the vibration data collected by the acceleration sensor, and calculate the normalized deviation from the preset threshold; The above features are input into a machine learning model in a parameter storage unit to output adjustment values ​​for welding current, voltage, rotation speed, and welding gun trajectory offset; The output adjustment value is checked for process constraints. If it exceeds the range, it is automatically cut off to the boundary value, and the operator is prompted to confirm if the deviation from the historical optimal parameter exceeds the preset ratio.

[0012] Preferably, the main control unit includes a device analysis unit, a device control unit, and a dynamic adjustment unit; The equipment analysis unit is used to analyze and process the corresponding parameters of each mechanism component; The dynamic adjustment unit is used to use the prediction model to predict the parameters at the next moment to obtain a predicted value; The equipment control unit adjusts the operation of the rotating mechanism component, the lifting and adjusting mechanism component, and the positioning mechanism component according to the analysis and processing results and the predicted value.

[0013] Preferably, the working process of the equipment analysis unit includes: Receive the raw data from the sensor unit, remove noise through sliding window filtering, and use interpolation to repair abnormal values; Compare the real-time parameters with the standard values ​​in the parameter storage unit and calculate the deviation rate; Analyze the parameter change rate, identify the parameter change trend based on the time series, and predict whether the parameter will exceed the safety range; The deviation rate and trend warning information are output to the dynamic adjustment unit and the human-computer interaction unit in a standardized format.

[0014] Preferably, the working process of the dynamic adjustment unit includes: Receive the deviation rate and trend warning information output by the equipment analysis unit, and construct input features based on the parameter sequence within the historical preset time period; Calling the time series prediction model in the parameter storage unit to predict the parameter value within a preset time period in the future; If the predicted value exceeds the safety range, a pre-adjustment amount is generated; Combined with the adjustment value and pre-adjustment amount output by the welding parameter dynamic generation module, the final adjustment strategy is formulated to ensure that the single parameter adjustment amount does not exceed the preset ratio.

[0015] Preferably, the working process of the device control unit includes: The final adjustment value output by the dynamic adjustment unit is converted into adjustment instructions for each mechanism, where the speed adjustment instruction of the rotating mechanism is sent through a pulse signal, the position adjustment instruction of the lifting adjustment mechanism is sent through an analog signal, and the clamping force adjustment instruction of the positioning mechanism is sent through a pneumatic valve control signal; Receive feedback signals from each mechanism, including the actual speed of the rotary encoder, the actual position of the grating ruler, and the actual clamping force of the pressure sensor; The PID algorithm is used to correct the deviation between the command value and the feedback value. The control cycle is a preset time length to ensure that the parameter adjustment lag time does not exceed the preset threshold.

[0016] Compared with related technologies, the plasma welding tool for processing water pump impellers provided by the present invention has the following beneficial effects: 1. This solution uses multiple sensors to collect real-time data on current, temperature, surface topography, and vibration. Combined with a machine learning model, it dynamically generates customized welding parameters, enabling precise control of key parameters such as welding current and voltage. A PID algorithm is also used for closed-loop deviation correction, effectively reducing defects such as porosity and cracks, significantly improving the impeller's weld quality, stability, and mechanical properties.

[0017] 2. This solution realizes automatic parameter analysis, prediction and mechanism regulation through the equipment analysis and dynamic adjustment of the main control unit and the coordinated work of the control unit, thereby improving the degree of tooling automation. The human-machine interaction unit supports parameter visualization and abnormal alarm, reducing manual intervention, shortening the welding cycle, and reducing the impact of human factors on quality, thus meeting the needs of large-scale high-quality production.

[0018] 3. The parameter storage unit of this solution stores welding procedures and process databases by model and material. Combined with the dynamic generation module of welding parameters, it can quickly adapt to impellers of different models and materials. Real-time parameter optimization is achieved through laser profile sensors and vibration monitoring, which reduces the time spent on tooling replacement and parameter adjustment, improves production flexibility and efficiency, and reduces enterprise production costs.

[0019] In summary, this solution addresses the pain points of traditional tooling. It uses multi-sensor acquisition and machine learning to dynamically generate parameters and PID closed-loop correction to improve welding quality stability. It relies on the main control unit's automated control and human-machine interaction to reduce manual intervention and shorten the cycle. It relies on a classified process database and a dynamic welding parameter generation module to quickly adapt to different impellers, reduce adjustment time, and comprehensively improve quality, efficiency, and flexibility, thereby reducing production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A three-dimensional diagram of a plasma welding tool for processing a water pump impeller provided by the present invention; Figure 2 for Figure 1 The partial perspective view shown; Figure 3 for Figure 1 Side view shown; Figure 4 for Figure 1 A perspective view of the other side is shown; Figure 5 for Figure 1 The structural diagram of the fixed anti-drop cap shown; Figure 6 for Figure 5 A side sectional view of the fixed anti-drop cap shown; Figure 7 This is a functional block diagram of the auxiliary welding system provided by the present invention.

[0021] Serial numbers in the figure: 1. Bottom plate; 2. Positioning block; 3. Base; 4. Rotating platform; 6. Fixed anti-drop cap; 7. Support rod; 8. Base; 9. Lifting motor; 10. Lifting seat; 11. Moving seat; 12. Linear motor; 13. Rotating motor; 14. Positioning cone; 15. Precision reducer; 16. Positioning ring; 17. Telescopic spring; 18. Fixed ring. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0023] The terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. As used in this disclosure and the appended claims, the singular forms "group," "class," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0024] It should be understood that although the terms first, second, third, etc. may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining."

[0025] Please refer to Figure 1-Figure 7 A plasma welding tool for processing a water pump impeller, comprising a basic frame assembly, a rotating mechanism assembly, a lifting and adjusting mechanism assembly, a positioning mechanism assembly and an auxiliary welding system; The auxiliary welding system is used to assist the welding process, including a main control unit, a sensor unit, a parameter storage unit and a human-computer interaction unit; The sensor unit includes a current sensor for monitoring welding current, a voltage sensor for monitoring welding voltage, a temperature sensor for monitoring molten pool temperature, a laser profile sensor for monitoring the microscopic topography of the impeller surface, and an acceleration sensor for monitoring the vibration state of the tooling; The parameter storage unit is used to store preset welding program parameters, a welding process database for impellers of different materials, and a welding parameter prediction model based on machine learning training; The main control unit is connected to the rotating mechanism assembly, lifting and adjusting mechanism assembly, positioning mechanism assembly, sensor unit, parameter storage unit and human-computer interaction unit respectively. It can receive sensor data, call stored parameters and prediction models, and send control instructions to each mechanism to realize intelligent control of the welding process. The dynamic welding parameter generation module uses the impeller surface micromorphology data collected by the laser profile sensor, combined with the tooling vibration characteristics obtained by the acceleration sensor, and the machine learning model in the parameter storage unit to generate personalized welding parameters adapted to the current welding conditions in real time. This includes dynamic adjustment of the welding current, voltage, welding gun motion trajectory, and rotating mechanism speed. The human-machine interaction unit is used for operators to set parameters, view welding status and receive system feedback information.

[0026] It should be noted that the tooling provides stable support for each actuator through the basic frame assembly, and the rotation, positioning, and lifting adjustment mechanisms work together to achieve precise positioning of the impeller and alignment of the welding gun. In the auxiliary welding system, multiple sensors collect welding current, voltage, molten pool temperature, impeller morphology, and tooling vibration data in real time. After analysis by the main control unit, combined with the process database and machine learning model of the parameter storage unit, the welding parameter dynamic generation module outputs a personalized adjustment strategy to achieve intelligent control of welding parameters and mechanism operation. This effectively solves the problems of low parameter control accuracy and unstable quality of traditional tooling, reduces defects such as pores and cracks, and reduces manual intervention and cycle time through automated control and human-machine interaction. It can also quickly adapt to impellers of different models and materials, improve production flexibility and efficiency, and reduce enterprise costs.

[0027] In the present application, the basic frame assembly includes a base plate 1, a positioning block 2 and a base 3. The upper end of the base plate 1 is rectangular and fixed with positioning blocks 2. Fixing holes are equidistantly provided on the positioning blocks 2. The upper end of the positioning block 2 is fixed with the base 3, which provides an installation reference for each functional mechanism and realizes the overall fixation of the tooling through the fixing holes; the rotating mechanism assembly is installed in the middle of the base 3 to drive the impeller to rotate; the positioning mechanism assembly is assembled on the rotating mechanism assembly to fix the impeller; the lifting and adjusting mechanism assembly is installed on one side of the base 3 to adjust the position of the welding gun.

[0028] It should be noted that the basic frame assembly forms a stable and rigid support through the hierarchical fixing structure of the base plate, positioning blocks and base: the base plate provides the overall installation basis for the tooling, and the rectangularly distributed positioning blocks realize the firm fixation of the tooling to the ground or workbench through the fixing holes, ensuring that there is no overall displacement during welding; the base serves as the installation carrier of the functional mechanism, and the fixing accuracy of the base and the positioning blocks ensures the relative position accuracy of the rotating mechanism (installed in the middle) and the lifting and adjusting mechanism (installed on one side), avoiding the influence of the impeller rotation concentricity and the welding gun alignment accuracy due to the shaking of the foundation.

[0029] The rotating mechanism includes a rotating platform 4 rotatably mounted in the middle of the upper end of the base 3, and a precision reducer 15 is installed between the lower end of the base 3 and the bottom plate 1 at the lower end of the rotating platform 4. The output end of the precision reducer 15 passes through the base 3 and is mutually engaged with the rotating platform 4. The positioning mechanism includes a positioning cone 14 fixed to the center of the upper end of the rotating platform 4. A threaded groove is formed on the upper end of the positioning cone 14, and movable grooves are formed on both sides of the positioning cone 14. The support rod 7 is slidably engaged in the movable groove, and the middle part of the support rod 7 is connected to the telescopic rod through a thread. The upper end of the positioning cone 14 is connected to the fixed anti-dropping cap 6 by a thread, and the lower end of the fixed anti-dropping cap 6 is slidably provided with a positioning ring 16. The bottom surface of the positioning ring 16 is fixedly connected to the limit block in an annular manner at equal distances, and a telescopic spring 17 is installed in an annular manner at equal distances between the positioning ring 16 and the inside of the fixed anti-dropping cap 6. A positioning hole is opened in the middle of the fixed anti-dropping cap 6 and the positioning ring 16; The lifting adjustment mechanism includes a base 8 fixed to a corner of the top of the base 3, a lifting slot is opened in the middle of the base 8, and a lifting motor 9 is installed on the upper end of the base 8. The output end of the lifting motor 9 is coaxially fixed to one end of the lifting screw, and the lifting screw is connected to the lifting seat 10 through a thread; A linear motor 12 is installed at the upper end of the lifting seat 10, and a moving seat 11 is mounted on the linear motor 12 and the lifting seat 10. A rotating motor 13 is installed on one side of the lower end of the moving seat 11. The output end of the rotating motor 13 passes through the moving seat 11 and is clamped on one side of the rotating disk. The rotating disk is rotatably installed on the other side of the lower end of the moving seat 11, and a fixing ring 18 is fixed to one side of the rotating disk.

[0030] In this application, the parameters of the sensor unit are obtained as follows: The current sensor is connected in series in the welding circuit to collect the plasma arc welding current in real time; The voltage sensor is connected in parallel between the welding gun and the workpiece to collect the arc voltage; The temperature sensor uses a non-contact measurement method to aim at the molten pool area and collect the molten pool temperature; The laser profile sensor is installed on the side of the welding gun, and uses a line laser to scan the impeller surface to be welded and output surface profile point cloud data; The acceleration sensor is installed at the connection between the rotating mechanism and the base to collect the vibration acceleration of the tooling.

[0031] It should be noted that the sensor unit achieves full-dimensional welding state perception through targeted installation methods and measurement logic: current and voltage sensors are connected to the welding circuit in series and parallel, respectively, accurately capturing the core parameters of the plasma arc energy, providing a direct basis for welding energy control. The non-contact temperature sensor focuses on the molten pool area to avoid contact interference with the welding process, while also monitoring the molten pool's thermal state in real time to prevent overheating or lack of fusion. The laser profile sensor is mounted on the side of the welding gun and synchronously scans the surface to be welded. The output point cloud data can be used to analyze micro-morphological features such as groove angle and misalignment, supporting dynamic correction of the welding gun trajectory. The acceleration sensor is installed at the vibration-sensitive part connecting the rotating mechanism to the base, accurately capturing the vibration state of the tooling during operation to suppress the influence of resonance. This coordinated layout and parameter acquisition design of multiple sensor types comprehensively covers key dimensions such as welding energy, thermal state, workpiece morphology, and tooling stability, providing accurate and comprehensive raw data support for intelligent control, effectively reducing the lag and error of traditional manual monitoring.

[0032] In this application, in the parameter storage unit: The preset welding program parameters are stored by impeller model, including initial welding current, voltage, rotation speed and welding gun trajectory coordinates; The welding process database is stored by material classification, including the mapping relationship between materials and welding parameters; The welding parameter prediction model based on machine learning training includes a random forest model for predicting welding current and a time series prediction model for predicting molten pool temperature. It is stored in the form of a binary file and contains feature weights and training logs. It should be further explained that when constructing the random forest model, the surface roughness, groove angle, misalignment amount extracted by the laser profile sensor and the vibration root mean square value and resonance frequency obtained by the acceleration sensor are used as input features, and the historical optimal welding current is used as the label. By integrating multiple decision trees for training, the hyperparameters such as the number of trees and the maximum depth are optimized, and the model containing the feature importance weights, the training error, and the feature selection log are saved; the time series prediction model adopts the LSTM network, and the welding current, voltage, speed and molten pool temperature of the historical time series are used as the input sequence. The sample set is constructed through the sliding window, and the parameters such as the number of network layers and the number of hidden units are optimized to realize temperature prediction. The model containing the network weights and the loss function curve and the iteration number log are saved. The above model construction methods are all based on existing machine learning technologies such as random forests and LSTM networks, and are conventional technical means in this field. Therefore, no further clarification is made in the application documents.

[0033] In this application, the working process of the welding parameter dynamic generation module includes: Surface roughness, groove angle, and misalignment features are extracted from the point cloud data collected by the laser profiler and standardized as follows: The laser profile sensor uses a line laser to scan the impeller surface to be welded and outputs high-density point cloud data. , covering the geometric morphology information of the impeller welding area; where N is the number of sampling points of the point cloud data, i represents the sampling index, is the three-dimensional coordinate of the point cloud; A sampling segment of length L is intercepted along the welding direction, the z-axis (height direction) data of the point cloud in the segment is extracted, and the arithmetic mean deviation is calculated to obtain the surface roughness. , the formula is: ; Where M is the number of valid points in the sampling segment, is the height value of point j, is the mean height of the sampling section; Perform edge detection on the point cloud of the welding groove area and fit the straight line equations of the groove surfaces on both sides , calculate the groove angle by the slope of the straight line, the formula is ; Extract the height data of the surfaces to be welded on both sides of the groove, and take the absolute value of the height difference at the corresponding position as the misalignment. The formula is: ;in is the height value of the same sampling point on both sides of the slope; Min-max normalization is used to map the eigenvalues ​​to the [0,1] interval. The formula is: ;in, is the original eigenvalue, 、 The minimum and maximum values ​​of the historical data of this feature are used to eliminate the dimensional differences and facilitate unified model processing; Extract the vibration root mean square value and resonant frequency characteristics from the vibration data collected by the acceleration sensor, and calculate the normalized deviation from the preset threshold, specifically: The acceleration sensor is installed at the connection between the rotating mechanism and the base to collect the vibration acceleration signal of the tooling Among them, the time domain continuous signal is obtained by discrete sampling. , K is the number of sampling points, k is the sampling index; Reflect vibration energy and calculate vibration root mean square value , calculation formula: ; Perform Fourier transform on the acceleration signal , extract the frequency component with the largest amplitude in the spectrum, which is the resonant frequency; Preset vibration signature thresholds (Based on process tests or standard database settings), calculate the normalized deviation between the current feature and the threshold, and obtain the normalized deviation of the vibration root mean square value and the resonant frequency. The formula is: ;in, They are the normalized deviation of the vibration root mean square value and the resonant frequency, respectively, which quantify the degree of deviation between the current vibration state and the ideal working condition; The above features are input into the machine learning model in the parameter storage unit, and the adjustment values ​​of welding current, voltage, rotation speed and welding gun trajectory offset are output, specifically: The normalized surface features Normalized deviation from vibration Splicing to form the model input vector ; Call the pre-trained machine learning model in the parameter storage unit (such as random forest, neural network, etc., based on historical welding data training, learning the mapping relationship between features and welding parameters), input Then, output the welding parameter adjustment value: ; is the welding current adjustment amount, is the voltage adjustment amount, is the rotation speed adjustment of the rotating mechanism, For the welding gun trajectory offset, dynamic adaptation of parameters is achieved; Unify the parameter adjustment values ​​into a real-time adjustment sequence , c represents the index of the above parameter adjustment value; The output adjustment value is checked for process constraints. If it exceeds the range, it is automatically cut to the boundary value. If the deviation from the historical optimal parameter exceeds the preset ratio, the operator is prompted to confirm: Preset welding parameter adjustment range , (based on material, impeller specifications and welding standard settings), the adjustment value output by the model is truncated: (Similar treatment ), ensure that the parameter adjustment is within the process safety range; Calculate the deviation rate between the adjusted value and the historical optimal parameter (like , is the historical optimal welding current), if , A preset deviation threshold, such as 20%, triggers a confirmation prompt from the human-computer interaction unit, and the operator decides whether to perform the adjustment, taking into account the flexibility of both intelligent decision-making and manual intervention.

[0034] It should be noted that the dynamic welding parameter generation module achieves precise adaptation of welding parameters through multi-dimensional feature extraction and intelligent optimization logic. First, surface roughness (reflecting the quality of the welded surface), groove angle, and misalignment (directly related to weld formation) are extracted from the laser profiler point cloud data, and then normalized to eliminate dimensional differences. Simultaneously, the root mean square value (RMS) (indicating vibration intensity) and resonant frequency (reflecting tooling stability) are extracted from the accelerometer vibration data, and the normalized deviation is calculated to quantify the degree of deviation from ideal working conditions. These features are then concatenated into a model input vector, which then uses a pre-trained machine learning model to generate personalized adjustment values ​​for current, voltage, speed, and torch trajectory. Adjustment safety is ensured through process constraint verification (automatic truncation for out-of-range conditions and manual confirmation for out-of-deviation conditions). This process dynamically maps workpiece topography and tooling status to welding parameters, addressing the poor adaptability associated with traditional fixed tooling parameters. This not only improves the accuracy and real-time nature of parameter adjustment, but also balances intelligent decision-making with process safety through dual verification.

[0035] In this application, the main control unit includes a device analysis unit, a device control unit, and a dynamic adjustment unit; The equipment analysis unit is used to analyze and process the corresponding parameters of each mechanism component: Real-time data is collected from sensors of the rotating, lifting, and positioning mechanisms (such as rotary encoders, grating scales, and pressure sensors) and the welding system (current and voltage sensors), covering the mechanism speed G1, position G2, clamping force G3, welding current G4, voltage G5, and weld pool temperature G6. The above parameters are unified into a real-time data sequence , b represents the index of the parameter in the real-time data; A sliding average filter with a window length of N=5 is used for each parameter sequence. The formula is: ,in is the filtered value at the kth moment, effectively suppressing high-frequency noise. represents the value of parameter b at the i-th moment; in this way, the random noise in the current data is effectively smoothed, eliminating the interference of data fluctuations on subsequent analysis; Use the 3σ criterion to identify outliers: calculate the mean value of the parameter per unit time , standard deviation ;like , it is determined to be an abnormal value; If it is a single point anomaly, linear interpolation is used to repair it, and the two adjacent normal data points before and after the anomaly are obtained and recorded as and , through the formula Calculate the parameter value after repair; If there are continuous abnormalities, such as three or more continuous abnormal values, the standby call mode is activated, the historical mean value of the same model impeller in the parameter storage unit is called to fill, and the human-computer interaction unit is triggered to alarm; Compare and analyze the pre-processed real-time parameters with the standard values ​​pre-stored in the parameter storage unit: Get any parameter of real-time data after sliding window filtering and outlier repair processing , and identify the standard values ​​corresponding to its parameters ; Through the formula Calculate the deviation rate of the parameter If the deviation rate If the deviation is greater than or equal to the preset allowable deviation threshold, it indicates a significant deviation; Further analyze the deviation rate change trend of each parameter and use the first-order difference method to calculate the change rate , combined with the change rate of the past 50 points to fit the slope of the straight line, if the absolute value of the slope is greater than the preset slope threshold, the prediction parameter will exceed the safety range; Generate a standardized data frame O whose elements are , Indicates the trend indicator (0 represents normal, 1 represents warning), which is sent synchronously to the dynamic adjustment unit and the human-machine interaction unit via industrial Ethernet; The dynamic adjustment unit is used to use the prediction model to predict the parameters at the next moment and obtain the predicted value. The specific steps are as follows: Based on the nearly 100 cycles (1 second) of data output by the device analysis unit, the input feature vector is constructed , including deviation rate, change rate and historical parameter series, covering short-term fluctuations and long-term trends; Call the LSTM prediction model in the parameter storage unit (input dimension 72, hidden layer 64 neurons, output dimension 6) to predict the parameter value for the next 200ms ,The model training data comes from the historical records of 1000 qualified welding processes; If the predicted value exceeds the safety range, calculate the pre-adjustment amount , combined with the real-time adjustment value output by the dynamic generation module of welding parameters , using weighted fusion algorithm to generate the final adjustment , the formula is ;in 、 2 respectively represent the weights corresponding to the pre-adjustment amount and the real-time adjustment value; the single adjustment amount is forced to be constrained , avoid parameter conflicts; The equipment control unit adjusts the operation of the rotating mechanism assembly, the lifting and adjusting mechanism assembly, and the positioning mechanism assembly according to the analysis and processing results and the predicted values: Obtain the final adjustment value of each parameter, compare the final adjustment value with its preset fluctuation threshold, and generate an adjustment instruction corresponding to the parameter if the final adjustment value is greater than its preset fluctuation threshold; The adjustment instruction is used to trigger the use of incremental PID algorithm to correct the deviation in real time. The formula is: ;in, , It is a PID parameter, pre-stored in the parameter storage unit, with a control period of 10ms, ensuring that the adjustment lag is less than 50ms.

[0036] It should be noted that the main control unit implements intelligent control of the welding process through a closed-loop logic of "equipment analysis - dynamic adjustment - equipment control". The equipment analysis unit first performs sliding window filtering (noise suppression) and 3σ anomaly correction (single-point interpolation, calling the historical mean for continuous anomalies and issuing alarms) on six parameters, such as speed and current. It then calculates the deviation rate and uses first-order difference and slope analysis to predict parameter out-of-limit risks, providing a reliable data foundation for control. The dynamic adjustment unit constructs a feature vector based on data from the past 100 cycles, uses an LSTM model to predict parameters for the next 200ms, and integrates pre-adjusted and real-time adjustments to generate a final strategy, enforcing a single adjustment limit of ≤5% of the standard value to prevent sudden changes. The equipment control unit converts the adjustment value into a control instruction (pulse, analog value, etc.) for the corresponding mechanism. It uses an incremental PID algorithm to correct deviations in real time, with a 10ms control cycle to ensure adjustment lag of less than 50ms. This layered collaborative mechanism solves the problems of delayed parameter response and blind adjustment in traditional tooling, enabling full-link intelligent control from data purification to precise execution, improving welding stability and parameter adaptability.

[0037] In this application, the implementation of the human-computer interaction unit includes: The operator inputs the impeller model, material, and thickness information through the touch screen, and supports manual fine-tuning of welding parameters. The input parameters must meet the preset process constraints. Real-time display of the time domain curves of welding current, voltage, temperature and the operating status of each mechanism, including rotation speed and lifting mechanism position information; When the parameters exceed the preset range, the system will provide abnormal information through sound and light alarm, and display the cause of the abnormality and recommended operation;

[0038] Supports manual / automatic mode switching and emergency stop operation, and saves current status data during emergency stop.

[0039] It should be noted that the human-machine interaction unit achieves human-machine collaboration through intuitive operation and intelligent feedback: the operator can input basic information such as the impeller model and material through the touch screen and manually fine-tune the parameters (subject to preset process constraints to avoid misoperation), and obtain real-time time domain curves of current, voltage, and temperature and mechanism operation status data, realizing visual monitoring of the welding process; when the parameters exceed the limit or are abnormal, the cause and recommended operation are displayed synchronously with sound and light alarms, quickly guiding the problem resolution; it supports flexible switching between manual and automatic modes, and automatically saves the current status data during emergency stop for easy traceability and review. This design not only lowers the operating threshold, allowing operators to accurately intervene in control, but also reduces human errors through status visualization and abnormal intelligent prompts. At the same time, it takes into account the data preservation needs in emergency situations, effectively connects the intelligent system with manual operation, and improves the usability of tooling and production safety.

[0040] The working principle of the plasma welding tool for processing water pump impellers proposed in this application is as follows: When the present invention is used, first align the center hole of the water pump impeller to be welded with the positioning cone 14, and put it on the positioning cone 14 so that the bottom surface of the impeller contacts the rotating platform 4. Slide the support rod 7 in the moving groove of the positioning cone 14 according to the size of the impeller, so that the upper end of the support rod 7 is against the impeller blade or edge to limit radial movement, and then rotate and fix the anti-drop cap 6 so that it is connected to the upper end of the positioning cone 14 through a thread. The positioning ring 16 presses the impeller downward under the elastic force of the telescopic spring 17, and cooperates with the support rod 7 to firmly fix the impeller to prevent displacement during welding rotation.

[0041] During the mechanical alignment stage, the precision reducer 15 is started to drive the rotating platform 4 to rotate precisely on the base 3, and the impeller to be welded is adjusted to the required angle by adjusting the rotation angle; the lifting motor 9 is started to drive the lifting screw to rotate, so that the lifting seat 10 moves up and down along the lifting slot of the base 8, and the vertical height of the welding gun (installed on the fixed ring 18) is adjusted; the linear motor 12 is started to drive the moving seat 11 to slide horizontally, and the horizontal position of the welding gun is calibrated; the rotating motor 13 is started to drive the rotating disk to rotate, and the angle of the fixed ring 18 is adjusted to ensure that the welding angle of the welding gun matches the impeller welding point, completing the mechanical structure alignment.

[0042] During the welding process, the auxiliary welding system simultaneously starts intelligent control: the sensor unit collects data in real time - the current sensor collects the welding current in series, the voltage sensor collects the arc voltage in parallel, the non-contact temperature sensor monitors the molten pool temperature, the laser profile sensor scans the impeller surface with a laser to output point cloud data, and the acceleration sensor collects the vibration acceleration of the connection between the rotating mechanism and the base.

[0043] The equipment analysis unit processes the raw data: removes noise through sliding window filtering, uses the 3σ criterion to identify outliers and repairs them with interpolation or historical mean; compares the preprocessed data with the standard value in the parameter storage unit, calculates the deviation rate, and uses the first-order difference method to analyze the parameter change trend, predicts whether it exceeds the safety range, and generates standardized deviation rate and trend warning information.

[0044] The dynamic generation module of welding parameters is based on the surface roughness, groove angle, misalignment (normalized) extracted by the laser profile sensor, and the vibration root mean square value and resonance frequency (calculated by normalized deviation) extracted by the acceleration sensor. It calls the random forest and other machine learning models in the parameter storage unit to output personalized adjustment values ​​of welding current, voltage, rotation speed and welding gun trajectory offset, and outputs them after process constraint verification (automatic truncation if out of range, manual confirmation if out of deviation).

[0045] The dynamic adjustment unit receives warning information from the equipment analysis unit, constructs a feature vector based on the historical parameter sequence, calls the LSTM time series prediction model to predict the parameter value in the next 200ms, and generates a pre-adjustment value if it exceeds the safety range; it integrates the pre-adjustment value with the adjustment value of the welding parameter dynamic generation module to formulate the final adjustment strategy, forcing the single adjustment amount to be ≤5% of the standard value to prevent parameter conflicts.

[0046] The equipment control unit converts the final adjustment value into a control instruction: the rotation speed instruction of the rotating mechanism is sent through a pulse signal, the position instruction of the lifting and adjusting mechanism is sent through an analog signal, and the clamping force instruction of the positioning mechanism is sent through a pneumatic valve control signal; at the same time, it receives feedback signals from the rotary encoder, grating scale, and pressure sensor, and uses an incremental PID algorithm to correct the deviation between the instruction value and the feedback value in real time (control cycle 10ms, adjustment lag <50ms), ensuring accurate and stable welding parameters.

[0047] The human-machine interaction unit displays the welding current, voltage, temperature time domain curves and mechanism operation status in real time. The operator can set parameters such as impeller model and material through the touch screen or manually fine-tune welding parameters (subject to process constraints). When the parameters exceed the preset range, the system will feedback the abnormal cause and suggest operations through sound and light alarms, support manual / automatic mode switching, and automatically save the current status data during emergency stop, realizing full-process closed-loop control of mechanical precise alignment, intelligent parameter adjustment, and human-machine collaborative monitoring, ensuring stable and efficient welding quality.

[0048] The above formula is calculated using numerical values ​​through dimension reduction methods such as normalization. The specific dimension reduction methods are not detailed here. The formula is generated based on extensive software simulations to closely resemble actual operating conditions. The preset parameters are set by those skilled in the art based on actual conditions.

[0049] The above embodiments may be implemented via software, hardware, firmware, or a combination thereof. If implemented via software, they may be embodied as a computer program product whose computer instructions, when loaded or executed, generate corresponding processes or functions. The computer may be a general-purpose computer, a network, or a programmable device. The instructions may be transmitted via wired or wireless means between computer-readable storage media, such as magnetic, optical, or semiconductor media.

[0050] It should be understood that the sequence numbers do not represent the order of execution, which must be determined by the functional logic. Professionals may use different methods to implement functions according to specific applications. The relevant units and algorithm steps can be implemented through electronic hardware or a combination of software and hardware.

[0051] It should be noted that the device embodiments are merely illustrative, and the unit divisions can be adjusted as needed; separate components can be physically separated or integrated, and functional units can be integrated into the processing unit or exist independently. If functions are implemented as software functional units and sold independently, they can be stored on a computer-readable storage medium containing instructions to drive a computer device to execute all or part of the steps of the methods of the various embodiments. Storage media include USB flash drives, ROM, RAM, magnetic disks, optical disks, etc.

[0052] Other embodiments of the present invention will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the invention being indicated by the following claims.

[0053] It should be understood that the present invention is not limited to the exact construction described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A plasma welding tool for processing a water pump impeller, characterized in that: It includes a basic frame assembly, a rotating mechanism assembly, a lifting and adjusting mechanism assembly, a positioning mechanism assembly and an auxiliary welding system; The auxiliary welding system is used to assist the welding process, and includes a main control unit, a sensor unit, a parameter storage unit and a human-computer interaction unit; The sensor unit includes a current sensor for monitoring welding current, a voltage sensor for monitoring welding voltage, a temperature sensor for monitoring molten pool temperature, a laser profile sensor for monitoring the microscopic topography of the impeller surface, and an acceleration sensor for monitoring the vibration state of the tooling; The parameter storage unit is used to store preset welding program parameters, a welding process database of impellers of different materials, and a welding parameter prediction model based on machine learning training; The main control unit is connected to the rotating mechanism assembly, the lifting and adjusting mechanism assembly, the positioning mechanism assembly, the sensor unit, the parameter storage unit and the human-computer interaction unit respectively, and can receive sensor data, call stored parameters and prediction models, and send control instructions to each mechanism to realize intelligent control of the welding process; The dynamic welding parameter generation module, based on the impeller surface micromorphology data collected by the laser profile sensor and the tooling vibration characteristics obtained by the acceleration sensor, uses the machine learning model in the parameter storage unit to generate personalized welding parameters adapted to the current welding conditions in real time, including dynamic adjustment of the welding current, voltage, welding gun motion trajectory and rotating mechanism speed; The human-machine interaction unit is used by the operator to set parameters, view welding status and receive system feedback information.

2. The plasma welding tool for processing a water pump impeller according to claim 1, characterized in that: The basic frame assembly comprises a bottom plate (1), a positioning block (2) and a base (3); the upper end of the bottom plate (1) is rectangular and fixedly connected to the positioning block (2); the positioning block (2) is provided with fixing holes at equal intervals; the upper end of the positioning block (2) is fixedly connected to the base (3); the rotating mechanism assembly is installed in the middle of the base (3); the positioning mechanism assembly is assembled on the rotating mechanism assembly; and the lifting and adjusting mechanism assembly is installed on one side of the base (3).

3. The plasma welding tool for processing a water pump impeller according to claim 2, characterized in that: The parameters of the sensor unit are obtained as follows: The current sensor is connected in series in the welding circuit to collect the plasma arc welding current in real time; The voltage sensor is connected in parallel between the welding gun and the workpiece to collect the arc voltage; The temperature sensor uses a non-contact measurement method to aim at the molten pool area and collect the molten pool temperature; The laser profile sensor is installed on the side of the welding gun, and uses a line laser to scan the impeller surface to be welded and output surface profile point cloud data; The acceleration sensor is installed at the connection between the rotating mechanism and the base to collect the vibration acceleration of the tooling.

4. The plasma welding tool for processing a water pump impeller according to claim 1, characterized in that: In the parameter storage unit: The preset welding program parameters are stored by impeller model, including initial welding current, voltage, rotation speed and welding gun trajectory coordinates; The welding process database is stored by material classification, including the mapping relationship between materials and welding parameters; The welding parameter prediction model based on machine learning training includes a random forest model for predicting welding current and a time series prediction model for predicting molten pool temperature. It is stored in the form of a binary file and contains feature weights and training logs.

5. The plasma welding tool for processing a water pump impeller according to claim 1, characterized in that: The working process of the welding parameter dynamic generation module includes: Extract surface roughness, groove angle, and misalignment features from point cloud data collected by the laser profiler and perform standardization processing; Extract the vibration root mean square value and resonant frequency characteristics from the vibration data collected by the acceleration sensor, and calculate the normalized deviation from the preset threshold; The above features are input into a machine learning model in a parameter storage unit to output adjustment values ​​for welding current, voltage, rotation speed, and welding gun trajectory offset; The output adjustment value is checked for process constraints. If it exceeds the range, it is automatically cut off to the boundary value, and the operator is prompted to confirm if the deviation from the historical optimal parameter exceeds the preset ratio.

6. The plasma welding tool for processing a water pump impeller according to claim 1, characterized in that: The main control unit includes a device analysis unit, a device control unit, and a dynamic adjustment unit; The equipment analysis unit is used to analyze and process the corresponding parameters of each mechanism component; The dynamic adjustment unit is used to use the prediction model to predict the parameters at the next moment to obtain a predicted value; The equipment control unit adjusts the operation of the rotating mechanism component, the lifting and adjusting mechanism component, and the positioning mechanism component according to the analysis and processing results and the predicted value.

7. The plasma welding tool for processing a water pump impeller according to claim 6, characterized in that: The working process of the equipment analysis unit includes: Receive the raw data from the sensor unit, remove noise through sliding window filtering, and use interpolation to repair abnormal values; Compare the real-time parameters with the standard values ​​in the parameter storage unit and calculate the deviation rate; Analyze the parameter change rate, identify the parameter change trend based on the time series, and predict whether the parameter will exceed the safety range; The deviation rate and trend warning information are output to the dynamic adjustment unit and the human-computer interaction unit in a standardized format.

8. The plasma welding tool for processing a water pump impeller according to claim 6, characterized in that: The working process of the dynamic adjustment unit includes: Receive the deviation rate and trend warning information output by the equipment analysis unit, and construct input features based on the parameter sequence within the historical preset time period; Calling the time series prediction model in the parameter storage unit to predict the parameter value within a preset time period in the future; If the predicted value exceeds the safety range, a pre-adjustment amount is generated; Combined with the adjustment value and pre-adjustment amount output by the welding parameter dynamic generation module, the final adjustment strategy is formulated to ensure that the single parameter adjustment amount does not exceed the preset ratio.

9. The plasma welding tool for processing a water pump impeller according to claim 6, characterized in that: The working process of the device control unit includes: The final adjustment value output by the dynamic adjustment unit is converted into adjustment instructions for each mechanism, where the speed adjustment instruction of the rotating mechanism is sent through a pulse signal, the position adjustment instruction of the lifting adjustment mechanism is sent through an analog signal, and the clamping force adjustment instruction of the positioning mechanism is sent through a pneumatic valve control signal; Receive feedback signals from each mechanism, including the actual speed of the rotary encoder, the actual position of the grating ruler, and the actual clamping force of the pressure sensor; The PID algorithm is used to correct the deviation between the command value and the feedback value. The control cycle is a preset time length to ensure that the parameter adjustment lag time does not exceed the preset threshold.

Images