Resistance spot welding machine capable of automatically monitoring welding quality

Abstract

The invention provides a resistance spot welding machine capable of automatically monitoring welding quality. The resistance spot welding machine comprises a cabinet body, a positioning mechanism, a welding executing mechanism and a welding quality automatic monitoring system. A console and a visual terminal are arranged at the top end of the cabinet body, a first linear motor, a transmission case and a transmission case belt positioning mechanism are arranged in the cabinet body, and second linear motors and welding executing mechanisms are arranged on the two sides; the welding quality automatic monitoring system comprises a multi-dimensional sensing layer, an intelligent analysis layer and a result interaction layer. The multi-dimensional sensing layer is used for collecting process parameters, positioning data and visual data, the intelligent analysis layer is used for extracting welding seam features, fusing data and judging quality, and the result interaction layer is used for achieving visual feedback and graded early warning. The positioning mechanism clamps the welding piece in a self-adaptive mode through a spring to prevent the welding piece from shaking, and the welding head and the visual monitoring instrument are in linkage through a rack and a gear to achieve instrument protection. The problems that a traditional welding machine is poor in positioning and single in monitoring are solved, real-time control over the welding quality is achieved, and stability and accuracy are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of double-sided electric welding tool, and particularly to a resistance spot welding machine capable of automatically monitoring welding quality. BACKGROUND

[0002] In the vast landscape of modern industrial production, welding technology is like a solid bridge, bearing the heavy responsibility of connecting various materials, becoming an indispensable key link in manufacturing, from the precise body assembly of automobile manufacturing to the high-end component splicing in the aerospace field, to the large-scale structure construction in the construction industry, welding technology is ubiquitous, and its quality directly affects the performance, safety of the product and the efficiency and cost of the entire production process. Without exaggeration, welding technology is the "steel tailor" of manufacturing, and is the core force of material fusion and structure forming, playing a crucial role in the industrial system.

[0003] However, the traditional welding device gradually exposes many serious problems in long-term practical application. On the one hand, due to the lack of effective positioning and limiting structure of the welding part, it is easy to shake during the welding process due to stress and vibration, resulting in welding deviation and insufficient joint strength, which poses a safety hazard. On the other hand, the welding quality automatic monitoring lacks multi-source cooperation. The existing welding quality monitoring only collects single electric parameters such as welding current and voltage, without establishing a correlation between such parameters and welding part clamping force feedback data and welding seam area visual image data, so it is impossible to determine whether the welding part clamping is stable and whether the welding seam has defects through multi-dimensional data combination, and thus it is difficult to fully and truly reflect the actual state of the entire welding process.

[0004] Therefore, it is necessary to provide a resistance spot welding machine capable of automatically monitoring welding quality to solve the above technical problems. SUMMARY

[0005] The present application provides a resistance spot welding machine capable of automatically monitoring welding quality to solve the problems raised in the background art.

[0006] To solve the above technical problems, the present application provides a resistance spot welding machine capable of automatically monitoring welding quality, which comprises a cabinet body, a control console and a visualization terminal are installed at the top end of the cabinet body, a first linear motor and a transmission box are installed inside the cabinet body, the transmission box is located at the rear end of the first linear motor, a positioning mechanism is installed on the transmission box, and a second linear motor is provided on both sides of the transmission box, and a welding execution mechanism is installed on the second linear motor. A welding quality automatic monitoring system is integrated on the cabinet body, and the welding quality automatic monitoring system is respectively in communication connection with the visualization terminal, the welding execution mechanism and the positioning mechanism, and the welding quality automatic monitoring system comprises a multi-dimensional perception layer, an intelligent analysis layer and a result interaction layer. The multi-dimensional perception layer is used for synchronously collecting process parameters, positioning state data of a welding piece, and visual data of a welding area in a welding process. The intelligent analysis layer is used for fusion processing of the multi-source data collected by the perception layer, realizing welding seam feature recognition, defect determination, and process parameter compliance verification. The result interaction layer is used for feeding back the analysis result to a visual terminal and triggering a pre-warning mechanism when quality is abnormal.

[0007] Preferably, the multi-dimensional perception layer comprises a parameter perception unit, a visual perception unit, and a mechanical perception unit, specifically: The parameter perception unit is used for collecting welding current, welding voltage, and welding timing parameters output by a welding execution mechanism, including pre-pressing time, welding time, and pressure maintaining time. The visual perception unit is internally provided with an image preprocessing subunit, which is used for denoising, enhancing, and extracting a region of interest of the collected original image of the welding area, and outputs standardized visual data. The mechanical perception unit is used for collecting clamping force feedback data in a clamping process of the welding piece, and the mechanical perception unit adopts a pressure sensing component and is linked with an elastic buffer structure of a welding piece positioning mechanism to realize real-time acquisition of clamping stability data.

[0008] Preferably, the intelligent analysis layer comprises a feature extraction module, a multi-source data fusion module, and a quality determination module, specifically: The feature extraction module adopts a deep learning algorithm to extract welding seam features from the standardized visual data output by the visual perception unit, wherein the welding seam features include a nugget diameter, a weld mark flatness, an edge continuity, and geometric parameters of surface defects including pores, cracks, and spatter residues. The multi-source data fusion module is used for correlatively analyzing process parameters of the parameter perception unit, clamping force data of the mechanical perception unit, and welding seam feature data of the feature extraction module, and establishing a mapping relationship between process parameters, positioning states, and welding seam quality. The quality determination module is internally provided with a preset welding quality standard library including process parameter thresholds and welding seam feature thresholds of welding pieces of different materials and thicknesses, compares the fusion data with the threshold values of the standard library, outputs a determination result of whether the welding quality is qualified, and classifies defect types of unqualified items.

[0009] Preferably, the result interaction layer comprises a visual feedback module and a hierarchical pre-warning module. The visual feedback module outputs the determination result of the intelligent analysis layer, welding seam feature images, and key process parameters to the visual terminal in a chart form, supports switching display of single welding point quality details and multi-welding point statistical data. The graded early warning module triggers different levels of early warnings based on the quality assessment results: Level 1 warning alerts operators with audible and visual cues; Level 2 warning triggers suggestions for fine-tuning process parameters and suspends welding at the next weld point; Level 3 warning directly cuts off the working signal of the welding actuator, forcibly stops the welding process, and outputs troubleshooting guidance.

[0010] Preferably, the positioning mechanism includes a DC motor installed inside the transmission box, a lead screw mounted on the output end of the DC motor via a coupling, a first guide rod mounted on the top surface of the transmission box, a third guide rail threaded onto the lead screw, the first guide rod being inserted into the third guide rail, a third slider slidably connected to the third guide rail, a first slide rail mounted on the first linear motor, a first slider slidably connected to the first slide rail, and limit components mounted on both the first slider and the third slider.

[0011] Preferably, the limiting assembly includes a base plate mounted on the first slider and the third slider. Limiting plates are provided on both sides of the base plate. Multiple insertion rods are fixed to the inner side of the limiting plates. Multiple slots for cooperating with the insertion rods are equally spaced on both sides of the base plate. A compression spring is installed in the slot, and the other end of the compression spring is connected to one end of the insertion rod.

[0012] Preferably, the welding actuator includes a second guide rail mounted on one side of the second linear motor, a mounting plate slidably connected to the second guide rail, a first electric actuator and two second guide rods mounted on the mounting plate, the two second guide rods being located on both sides of the first electric actuator, and a translation block being mounted on the output end of the first electric actuator.

[0013] Preferably, the translation block is slidably connected to the mounting plate, the second guide rod is inserted into the translation block, two mounting rods are fixedly connected to the inner side of the translation block, and a second electric actuator is installed inside the translation block, with a welding head installed at the output end of the second electric actuator.

[0014] Preferably, racks are fixed to both sides of the welding head, and visual monitoring instruments are provided on both sides of the welding head. A gear is fixed to one end of the visual monitoring instrument. The gear meshes with the rack for transmission, and the gear is rotatably connected to the mounting rod through a central shaft. The two visual monitoring instruments and the welding head have the same focal point.

[0015] Compared with related technologies, the resistance spot welding machine for automatically monitoring welding quality provided by the present invention has the following beneficial effects: 1. To address the issue of welded parts lacking effective positioning and prone to wobbling, this solution uses a DC motor-driven lead screw in the positioning mechanism to adjust the height of the third guide rail. This, combined with the compression spring of the limiting component, enables the welded parts to self-adaptively clamp. Simultaneously, the first linear motor adjusts the position of the welded parts, effectively improving the stability of the welded parts during the welding process and avoiding defects such as weld deviation and insufficient joint strength caused by wobbling.

[0016] 2. To address the problem of insufficient multi-source collaborative monitoring and the inability of single-parameter acquisition to fully reflect the welding status, the automatic welding quality monitoring system of this solution simultaneously collects welding current, voltage, clamping force and weld visual data through a multi-dimensional perception layer. After fusion processing by the intelligent analysis layer, it realizes weld feature recognition and quality judgment, which can comprehensively reflect the status of the entire welding process and effectively ensure the timeliness and accuracy of welding quality judgment. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the overall appearance of the device proposed in this invention; Figure 2 This is a schematic diagram of the internal structure of the device proposed in this invention; Figure 3 This is a schematic diagram of the cross-sectional structure of the limiting component proposed in this invention; Figure 4 This is a schematic diagram of the limiting mechanism structure proposed in this invention; Figure 5 This is a schematic diagram of the moving mechanism structure proposed in this invention; Figure 6 This is a schematic diagram of the welding actuator structure proposed in this invention; Figure 7 This is a schematic diagram of the internal structure of the welding actuator proposed in this invention; Figure 8 The present invention proposes Figure 3 Enlarged schematic diagram of the structure at part A in the middle; Figure 9 The present invention proposes Figure 7 Enlarged schematic diagram of the structure of part B in the middle; Figure 10 This is a block diagram showing the connection relationships of each module in the automatic welding quality monitoring system proposed in this invention.

[0018] The components in the diagram are numbered as follows: 1. Cabinet; 2. Visual terminal; 3. First linear motor; 4. Transmission box; 5. Second linear motor; 6. DC motor; 7. First guide rod; 8. Lead screw; 9. Third guide rail; 10. Third slider; 11. First slider; 12. Base plate; 13. Limiting plate; 14. Compression spring; 15. Mounting plate; 16. First electric actuator; 17. Second guide rod; 18. Translation block; 19. Mounting rod; 20. Second electric actuator; 21. Welding head; 22. Rack; 23. Visual monitoring instrument; 24. Gear. Detailed Implementation

[0019] 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.

[0020] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “group,” “class,” and “the” as used in this disclosure and the appended claims are also 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 includes any or all possible combinations of one or more of the associated listed items.

[0021] 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 used only to distinguish information of the same type from one another. 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 "when," "when," or "in response to determination."

[0022] Please refer to the following: Figures 1-10 An automatic resistance spot welding machine for monitoring welding quality includes a cabinet 1, a control panel and a visualization terminal 2 installed on the top of the cabinet 1, a first linear motor 3 and a transmission box 4 installed inside the cabinet 1, the transmission box 4 is located at the rear end of the first linear motor 3, a positioning mechanism is installed on the transmission box 4, and a second linear motor 5 is provided on both sides of the transmission box 4, with a welding execution mechanism installed on the second linear motor 5; The cabinet 1 integrates an automatic welding quality monitoring system, which establishes communication connections with the visualization terminal 2, the welding actuator, and the positioning mechanism. The automatic welding quality monitoring system includes a multi-dimensional perception layer, an intelligent analysis layer, and a result interaction layer, wherein: The multi-dimensional perception layer is used to simultaneously collect process parameters, weldment positioning status data, and visual data of the welding area during the welding process; The intelligent analysis layer is used to fuse and process multi-source data collected by the perception layer to achieve weld feature recognition, defect judgment, and compliance verification of process parameters; The results interaction layer is used to feed back the analysis results to the visualization terminal 2 and trigger an early warning mechanism when there are quality anomalies.

[0023] Specifically, the multi-dimensional perception layer includes parameter perception units, visual perception units, and mechanical perception units, as follows: The parameter sensing unit uses a high-precision data acquisition card to be electrically connected to the control circuit of the welding actuator. It is used to collect the welding current I_raw, welding voltage U_raw and welding timing parameters output by the welding actuator, including pre-pressure time T_1, welding time T_2 and holding pressure time T_3. Due to electromagnetic interference during the welding process, a moving average filtering algorithm is used to reduce noise in the collected data. The filtering formula is: I_cal (t)=1 / W ∑_(i=t-W+1)^t▒I_raw (i), U_cal (t)=1 / W ∑_(i=t-W+1)^t▒U_raw (i); where W is the length of the moving window, used to balance the noise reduction effect and the real-time performance of the data, I_raw (i) and U_raw (i) are the original current and voltage sample values ​​at time i, respectively, and I_cal (t) and U_cal (t) are the calibrated welding current and welding voltage at time t, respectively. The calibrated welding current and welding voltage are used to compare with the standard current threshold to determine whether there is an abnormal weld nugget caused by overcurrent / undercurrent and to verify whether the voltage stability affects the weld formation. The logic for acquiring welding timing parameters is as follows: T_1 is the time interval from when the welding head 21 contacts the workpiece (triggered by the displacement sensor configured on the second electric push rod 20) to when the welding circuit is energized (triggered by the current signal of the power supply circuit), used to determine whether the preload is sufficient to avoid the weld nugget shift caused by the contact gap; T_2 is the time interval between the energization and de-energization of the welding circuit, which is used to determine whether the heating time meets the requirements for weld nugget formation in conjunction with the calibrated welding current I_cal(t). T_3 is the time interval from the power outage of the welding circuit to the welding head 21 being separated from the weldment, used to determine whether the pressure holding and cooling is sufficient to prevent weld nugget crystallization defects; all three are obtained through the timing module of the data acquisition card; The visual perception unit establishes data interaction with the visual monitoring instrument 23 on the welding actuator through industrial Ethernet. The visual perception unit is built-in with an image preprocessing subunit, which is used to perform noise reduction, enhancement, and region of interest extraction on the original welding area image Img_raw collected by the visual monitoring instrument 23, and outputs standardized visual data Img_std. The specific operations are as follows: Noise suppression operation: For the salt-and-pepper noise generated by welding arc light and metal spatter, the Gaussian filtering algorithm is used for processing. The formula is: G(x,y)=1 / √2πδ e^(-(x^2+y^2) / (2δ^2 )); where G(x,y) is the weight of the filter kernel at the pixel coordinates (x,y), (x,y) is the two-dimensional coordinate of the image pixel in the original welding area image Img_raw, and δ is the Gaussian standard deviation, which is used to control the filtering intensity; perform convolution processing on the original welding area image Img_raw according to G(x,y) to obtain the filtered image Img_blur. The formula is Img_blur=Img_raw*G(x,y), where * represents the convolution operation; Contrast enhancement operation: To highlight the gray-scale difference between the weld seam and the base material, the adaptive histogram equalization algorithm is used for processing. The formula is s=T(r)=1 / S ∑_(k=0)^r▒H(k); s is the pixel gray-scale value after enhancement, r is the original pixel gray-scale value in Img_blur, S is the total number of image pixels, H(k) is the number of pixels with gray-scale value k, and T(r) is the gray-scale conversion function; obtain the contrast-enhanced image Img_enh according to the formula Img_enh=T(Img_blur); Region of interest (ROI) extraction operation: Use the Canny edge detection algorithm to identify the weld seam contour. The judgment method is: When Img_grad (x,y)≥H_th, it is determined as a strong edge; When L_th<Img_grad (x,y)<H_th, if it is connected to a strong edge, it is determined as an edge, otherwise it is a non-edge; When Img_grad (x,y)≤L_th, it is determined as a non-edge; Among them, Img_grad (x,y) is the gradient value of Img_enh at (x,y), H_th is the preset high edge threshold, L_th is the preset low edge threshold, and H_th=2L_th; generate the minimum bounding rectangle through contour fitting as the region of interest, and normalize the ROI size to obtain Img_std; The mechanical perception unit is assembled on the inner side of the limiting plate 13 of the limiting component of the positioning mechanism, and联动采集夹持力原始电压信号V_raw, converts it into the clamping force data F and the stability parameter σ_F. The specific operations are as follows: It should be noted that there is an unclear expression "联动采集夹持力原始电压信号V_raw" in the original text. It might need further clarification to ensure an accurate translation. Here, I translated it as "联动采集夹持力原始电压信号V_raw" as best I could. The force-to-voltage conversion operation establishes the conversion relationship through pre-calibration with standard weights, using the formula F=K⋅(V_cal-V_zero). Here, F is the actual clamping force of the weldment, used to determine whether the clamping force is within the standard range, avoiding the weldment shifting due to excessive looseness and the workpiece deformation due to excessive tightness. K is the sensor calibration coefficient, V_cal is the calibration voltage, used to eliminate temperature drift, using the formula V_cal=V_raw-V_noise. V_noise is the sensor temperature drift compensation voltage, and V_zero is the sensor zero-point voltage. The clamping stability determination operation is performed by calculating the standard deviation of the clamping force per unit time. The formula is σ_F=√(1 / M ∑_(j=1)^M▒( F_j-F ̄)^2 ), where σ_F represents the standard deviation of the clamping force, used to determine whether the clamping is stable. A preset stability threshold ε is set. When σ_F<ε, the clamping is considered stable; M is the number of samples per unit time, F_j is the clamping force value of the j-th sample, and F ̄ is the average clamping force of the M samples.

[0024] Specifically, the intelligent analysis layer includes a feature extraction module, a multi-source data fusion module, and a quality assessment module, as follows: The feature extraction module uses a deep learning algorithm to extract weld features from the standardized visual data output by the visual perception unit. Specifically, an improved ResNet-18 convolutional neural network is used for weld feature extraction. The network input is the specified Img_std, and the outputs are the weld nugget diameter D, weld smoothness R, edge continuity C, and surface defect area S. The calculation logic for each feature is as follows: The diameter D of the molten core is extracted by semantically segmenting the standardized visual data Img_std output by the visual perception unit using an improved ResNet-18 convolutional neural network to output a mask image Mask_nugget for the molten core region. The diameter D is calculated using the minimum circumcircle method, with the formula D=2√(S_nugget / π). This formula is used to determine whether the molten core meets the strength requirements. Here, S_nugget represents the actual area of ​​the molten core, with the formula S_nugget=∑_(x,y)▒M ask_nugget (x,y)×K_size^2, and Mask_nugget (x,y) is the pixel value of the mask image at (x,y). Weld smoothness R is extracted by performing semantic segmentation on standardized visual data using the above method, dividing the weld area, and counting the total number of pixels N_scam in the weld area. Based on this, the weld smoothness R is calculated using the formula R=1-√(1 / N_scam ∑_((x,y)∈welder area)▒( G(x,y)-G ̄)^2 ) / G ̄ , where R takes a value of 0~1, the closer to 1, the smoother it is, and is used to determine whether there are uneven defects in the weld. G(x,y) is the gray value at (x,y) in the weld area, and G ̄ represents the average gray value of the weld area, with the formula G ̄=1 / N_seam ∑_((x,y)∈welder area)▒G(x,y); Edge continuity C is determined by obtaining the weld edge pixel set P_edge through Canny edge detection and calculating the proportion of continuous edge segments. The formula is C=L_cont / L_total, where C is the edge continuity, which takes a value of 0~1. The closer it is to 1, the more continuous it is. It is used to determine whether there is a break in the weld. L_total is the total length of the weld edge, and L_cont is the total length of continuous edge segments with a length ≥ the preset length threshold L_thr. The surface defect area S is calculated using the defect mask image Mask_defect output by the network. The pixel values ​​for pores / cracks / spatter residue regions are set to 2 / 3 / 4 respectively, with a background value of 0. The area is calculated according to defect type using the formula: S_type = ∑_(x,y)▒[Mask_defect(x,y)=type]×K_size^2, where S_type is the actual area of ​​a certain type of defect (type=2 corresponds to pores, type=3 to cracks, and type=4 to spatter residue), and S={S_2,S_3,S_4} is the set of defect areas used to determine whether the defect exceeds the limit. The improved ResNet-18 uses a multi-task loss function, with the formula Loss = d1⋅Loss_D + d2⋅Loss_R + d3⋅Loss_C + d4⋅Loss_S, where Loss_D, Loss_R, and Loss_C are... Loss_D and Loss_S are the regression / classification losses for weld nugget diameter, weld smoothness, edge continuity, and defect area, respectively. Loss_D / Loss_S uses MSE loss, and Loss_R / Loss_C uses MAE loss. d1, d2, d3, and d4 are the corresponding weight coefficients. The multi-source data fusion module is used to perform correlation analysis between the process parameters of the parameter sensing unit, the clamping force data of the mechanical sensing unit, and the weld feature data of the feature extraction module, and to establish a mapping relationship between process parameters, positioning status, and weld quality. The specific steps are as follows: Based on the process parameters of the parameter sensing unit, the clamping force data of the mechanical sensing unit, and the weld feature data of the feature extraction module, a subset is constructed, including the process parameter subset A={I_cal,U_cal,T_1,T_2,T_3}, the positioning data subset B={F,σ_F}, and the weld feature subset E={D,R,C,S}. Based on the degree of influence of each parameter on welding quality, the weights can be determined by the analytic hierarchy process (AHP), each parameter can be assigned a weight, and each parameter can be normalized to the [0,1] interval. The normalization formula is x_i^'=(x_i-x_(i,min)) / (x_(i,max)-x_(i,min)), where x_i^' is the normalized value of parameter x_i, and x_(i,max) and x_(i,min) are the upper and lower limits of the acceptable range of parameter x_i. The basic probability assignment function is calculated for the normalized subset data, and then fused using the DS synthesis rule. The formula is m(Θ)=1 / (1-P) ∑_(A_1∩A_2∩A_3=Θ)▒m_1 (A_1)⋅m_2 (A_2)⋅m_3 (E_3), where Θ={qualified, unqualified} is the recognition frame, m_1 / m_2 / m_3 are the BPA functions of subsets A / B / E respectively, such as m_1 (qualified)=∑▒w_i⋅x_(i_n)^', m_1 (unqualified)=1-m_1 (qualified), and P is the conflict coefficient, used to judge data consistency, expressed as K=∑_(A_1∩A_2∩A_3=∅)▒m_1 (A_1)⋅m_2 (A_2)⋅m_3 (E_3), after fusion, output the comprehensive confidence levels m(qualified) and m(unqualified), and establish the mapping relationship between A, B, E and m(qualified) and m(unqualified); The quality assessment module has a built-in preset welding quality standard library Lib_std={Lib_para,Lib_feat}, which includes process parameter thresholds Lib_para and weld feature thresholds Lib_feat for welded parts of different materials and thicknesses. By comparing and fusing the data with the standard library thresholds, it outputs the judgment result of whether the welding quality is qualified, and classifies the defect types of unqualified items, specifically: Each parameter is checked individually to see if it is within the range of Lib_std, such as I_cal∈[I_min,I_max], D∈[D_min,D_max], S_3≤S_(3,max); the number of parameters exceeding the standard is counted and recorded as N_over; a confidence threshold m_thr is set. If m(qualified)≥m_thr and N_over=0, the judgment result Q=1, that is, qualified; if m(qualified)≥m_thr and N_over≥2, the judgment result Q=0, that is, unqualified; if it is between the two, a secondary verification is triggered, such as re-collecting data once and then judging. Classify the defect types according to the out-of-specification parameters and the defect area S according to preset rules, such as: If T_2 < T_(2,min) and D < D_min, then determine that the defect type Type = "too small fusion zone (insufficient welding time)"; If σ_F > ε and C < C_min, then determine that Type = "weld offset (unstable clamping)"; If S_3 > S_(3,max) and the fluctuation value of U_cal (max(U_cal) - min(U_cal)) > U_fluc, then determine that Type = "crack (voltage fluctuation)"; Finally, output the judgment result Q (qualified / unqualified) and Type (defect type, "none" when there is no defect), which are used for visual feedback and warning triggering in the result interaction layer; classify all out-of-specification parameters according to the above rules, which can achieve precise traceability, targeted intervention and full-process closed-loop control of welding quality anomalies.

[0025] Specifically, the result interaction layer includes a visual feedback module and a hierarchical warning module: The visual feedback module outputs the judgment result, weld feature image, and key process parameters of the intelligent analysis layer to the visualization terminal 2 in a chart form, supporting the switching display of single-weld spot quality details and multi-weld spot statistical data; The hierarchical warning module triggers warnings of different levels according to the quality judgment result: Level 1 warning: Triggered when there are minor defects in a single weld spot, such as the porosity area not exceeding the critical value and the weld bead flatness being slightly low. Through the yellow light and sound prompt of the visualization terminal 2, that is, the configured LED light is always on combined with intermittent beeping to remind the operator to pay attention, and at the same time, a pop-up window marks the defect location and a preliminary impact description; Level 2 warning: Triggered when a continuously set number of weld spots are judged unqualified or the key process parameters are close to the threshold, such as the welding time being slightly shorter than the standard lower limit. Pause the welding operation of the next weld spot, and pop up a window to output suggestions for fine-tuning the process parameters. Resume the welding process after the operator confirms the adjustment; Level 3 warning: Triggered when there are fatal defects in a single weld spot, such as the crack area exceeding the standard, the clamping stability being seriously insufficient, or equipment anomalies such as the welding voltage fluctuating beyond the safe range. Directly cut off the power control signal of the welding actuator to force a shutdown. The visualization terminal 2 starts a red light and sound warning, that is, the configured LED light flashes combined with continuous beeping, and outputs a fault troubleshooting guide, specifically such as "crack defect: check the stability of the welding circuit and the cleanliness of the base material surface" "clamping anomaly: repair the elastic buffer structure of the positioning mechanism".

[0026] Specifically, the positioning mechanism includes a DC motor 6 installed inside the transmission box 4. A lead screw 8 is installed at the output end of the DC motor 6 via a coupling. A first guide rod 7 is installed on the top surface of the transmission box 4. A third guide rail 9 is threaded onto the lead screw 8. The first guide rod 7 is inserted into the third guide rail 9. A third slider 10 is slidably connected to the third guide rail 9. A first slide rail is installed on the first linear motor 3. A first slider 11 is slidably connected to the first slide rail. Limiting components are installed on both the first slider 11 and the third slider 10.

[0027] Specifically, the limiting assembly includes a base plate 12 installed on the first slider 11 and the third slider 10. Limiting plates 13 are provided on both sides of the base plate 12. Multiple insertion rods are fixed to the inner side of the limiting plates 13. Multiple slots for cooperating with the insertion rods are equally spaced on both sides of the base plate 12. A compression spring 14 is installed in the slot. The other end of the compression spring 14 is connected to one end of the insertion rod.

[0028] Specifically, the welding actuator includes a second guide rail mounted on one side of the second linear motor 5, a mounting plate 15 slidably connected to the second guide rail, a first electric push rod 16 and two second guide rods 17 mounted on the mounting plate 15, the two second guide rods 17 being located on both sides of the first electric push rod 16, and a translation block 18 being mounted on the output end of the first electric push rod 16.

[0029] Specifically, the translation block 18 is slidably connected to the mounting plate 15, the second guide rod 17 is inserted into the translation block 18, two mounting rods 19 are fixedly connected to the inner side of the translation block 18, and a second electric push rod 20 is installed inside the translation block 18. A welding head 21 is installed at the output end of the second electric push rod 20.

[0030] Specifically, racks 22 are fixed to both sides of the welding head 21, and visual monitoring instruments 23 are provided on both sides of the welding head 21. A gear 24 is fixed to one end of the visual monitoring instrument 23. The gear 24 meshes with the rack 22 for transmission, and the gear 24 is rotatably connected to the mounting rod 19 through the central shaft. The focal points of the two visual monitoring instruments 23 and the welding head 21 are consistent.

[0031] The working principle of the resistance spot welding machine for automatically monitoring welding quality provided by this invention is as follows: Positioning and clamping of welded parts: The welded parts are placed on the base plate 12. The DC motor 6 drives the lead screw 8 to drive the third guide rail 9 to adjust to a suitable height along the first guide rod 7. The first slider 11 is used to make the limiting component abut against the welded parts. The first linear motor 3 drives the first slider 11 to adjust the front and rear position of the welded parts. The limiting plate 13 retracts and clamps the welded parts under the action of the compression spring 14 to prevent shaking.

[0032] Welding execution: The second linear motor 5 adjusts the height of the welding execution mechanism, the first electric push rod 16 pushes the translation block 18 to move along the second guide rod 17, and the second electric push rod 20 pushes the welding head 21 forward. Together with the linear motor, the welding execution mechanism and the welding workpiece are moved to achieve double-sided electric welding.

[0033] Visual monitoring and reset: When the welding head 21 extends, the rack 22 meshes with the gear 24 to drive the visual monitoring instrument 23 to rotate to both sides to avoid damage; after welding, the welding head 21 retracts, the instrument resets and focuses on the weld point to monitor the quality, and the data is transmitted to the visualization terminal 2; after welding is completed, each component is reset to prepare for the next operation.

[0034] During the welding process, data is collected synchronously through a multi-dimensional perception layer. This includes the parameter perception unit collecting welding current, voltage, and timing parameters; the mechanical perception unit collecting clamping force and stability data; and the visual perception unit preprocessing weld images collected by the instrument. The intelligent analysis layer extracts weld features (melt nugget diameter, defects, etc.), integrates multi-source data, and compares them with a quality standard library to determine the quality. The result interaction layer displays the judgment results and parameters in charts on the visualization terminal 2, and triggers graded warnings according to the degree of quality anomalies. After welding is completed, all components are reset.

[0035] Formula Explanation: The formulas involved in this solution, such as welding current filtering, clamping force calculation, and weld feature extraction, all employ dimensionless numerical calculations. Dimensionlessness can be achieved through standardization and other means (the specific process is not detailed here). The formulas are obtained through software simulation and optimization based on a large amount of welding condition data (such as process parameters and defect sample data of welded parts of different materials / thicknesses), which closely reflects the actual welding quality judgment requirements. The preset parameters in the formulas (such as sliding window length, edge detection threshold, stability threshold, etc.) can be flexibly set by those skilled in the art based on the material, thickness, and production conditions of the welded parts.

[0036] Implementation: This solution embodiment can be implemented through software, hardware, firmware, or any combination thereof. If implemented in software, it can form a computer program product. After the computer instructions contained therein are loaded and executed, they can drive the device to complete functions such as multi-dimensional perception layer data acquisition (current, voltage, clamping force, weld image), intelligent analysis layer multi-source data fusion and quality judgment, and result interaction layer visualization feedback and hierarchical early warning. The computer instructions can be transmitted between devices via wired or wireless means and stored in computer-readable storage media (such as USB flash drive, ROM, RAM, magnetic disk, optical disk, etc.).

[0037] Other notes: The execution order of each module in this solution is determined by the functional logic and is not related to the process description sequence number; the functional implementation method (hardware / software) depends on the specific application scenario and design constraints; the communication connection between the coupled modules in the system can be achieved through interfaces, supporting the integration of functional units (such as the integration of each unit in the sensing layer into the same acquisition module), and selecting some or all units as needed can meet the actual welding quality monitoring requirements.

[0038] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0039] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A resistance spot welding machine for automatically monitoring welding quality, characterized in that, The cabinet (1) is characterized in that: a control panel and a visualization terminal (2) are installed on the top of the cabinet (1), a first linear motor (3) and a transmission box (4) are installed inside the cabinet (1), the transmission box (4) is located at the rear end of the first linear motor (3), a positioning mechanism is installed on the transmission box (4), and a second linear motor (5) is provided on both sides of the transmission box (4), and a welding execution mechanism is installed on the second linear motor (5); The cabinet (1) is equipped with an automatic welding quality monitoring system. The automatic welding quality monitoring system establishes communication connections with the visualization terminal (2), the welding execution mechanism, and the positioning mechanism. The automatic welding quality monitoring system includes a multi-dimensional perception layer, an intelligent analysis layer, and a result interaction layer, wherein: The multi-dimensional perception layer is used to synchronously collect process parameters, weldment positioning status data, and visual data of the welding area during the welding process. The intelligent analysis layer is used to fuse and process multi-source data collected by the perception layer to achieve weld feature recognition, defect judgment and process parameter compliance verification. The result interaction layer is used to feed back the analysis results to the visualization terminal (2) and trigger an early warning mechanism when there is an abnormality in quality.

2. The resistance spot welding machine for automatically monitoring welding quality according to claim 1, characterized in that, The multi-dimensional perception layer includes a parameter perception unit, a visual perception unit, and a mechanical perception unit, specifically: The parameter sensing unit is used to collect the welding current, welding voltage and welding timing parameters output by the welding actuator, including pre-pressure time, welding time and holding pressure time; The visual perception unit has a built-in image preprocessing subunit, which is used to reduce noise, enhance and extract regions of interest from the original image of the welding area, and output standardized visual data. The mechanical sensing unit is used to collect clamping force feedback data during the clamping process of the welded part. The mechanical sensing unit adopts a pressure sensing component and is linked with the elastic buffer structure of the welded part positioning mechanism to obtain clamping stability data in real time.

3. The resistance spot welding machine for automatically monitoring welding quality according to claim 2, characterized in that, The intelligent analysis layer includes a feature extraction module, a multi-source data fusion module, and a quality judgment module, specifically: The feature extraction module uses a deep learning algorithm to extract weld features from the standardized visual data output by the visual perception unit. The weld features include weld nugget diameter, weld smoothness, edge continuity, and geometric parameters of surface defects, including porosity, cracks, and spatter residue. The multi-source data fusion module is used to perform correlation analysis on the process parameters of the parameter sensing unit, the clamping force data of the mechanical sensing unit, and the weld feature data of the feature extraction module to establish a mapping relationship between process parameters, positioning status and weld quality. The quality judgment module has a built-in preset welding quality standard library, which includes process parameter thresholds and weld feature thresholds for welded parts of different materials and thicknesses. By comparing and fusing the data with the standard library thresholds, it outputs the judgment result of whether the welding quality is qualified and classifies the defect types of unqualified items.

4. The resistance spot welding machine for automatically monitoring welding quality according to claim 3, characterized in that, The result interaction layer includes a visualization feedback module and a tiered early warning module: The visualization feedback module outputs the judgment results of the intelligent analysis layer, weld feature images, and key process parameters to the visualization terminal (2) in a graphical form, and supports switching between single weld point quality details and multi-weld point statistical data display; The graded early warning module triggers different levels of early warnings based on the quality assessment results: Level 1 warning alerts operators with audible and visual cues; Level 2 warning triggers suggestions for fine-tuning process parameters and suspends welding at the next weld point; Level 3 warning directly cuts off the working signal of the welding actuator, forcibly stops the welding process, and outputs troubleshooting guidance.

5. The resistance spot welding machine for automatically monitoring welding quality according to claim 1, characterized in that, The positioning mechanism includes a DC motor (6) installed inside the transmission box (4). The output end of the DC motor (6) is equipped with a lead screw (8) via a coupling. A first guide rod (7) is installed on the top surface of the transmission box (4). A third guide rail (9) is threaded onto the lead screw (8). The first guide rod (7) is inserted into the third guide rail (9). A third slider (10) is slidably connected to the third guide rail (9). A first slide rail is installed on the first linear motor (3). A first slider (11) is slidably connected to the first slide rail. Limiting components are installed on both the first slider (11) and the third slider (10).

6. The resistance spot welding machine for automatically monitoring welding quality according to claim 5, characterized in that, The limiting assembly includes a base plate (12) installed on the first slider (11) and the third slider (10). Limiting plates (13) are provided on both sides of the base plate (12). Multiple insertion rods are fixed to the inner side of the limiting plate (13). Multiple slots for cooperating insertion rods are equally spaced on both sides of the base plate (12). A compression spring (14) is installed in the slot. The other end of the compression spring (14) is connected to one end of the insertion rod.

7. The resistance spot welding machine for automatically monitoring welding quality according to claim 1, characterized in that, The welding actuator includes a second guide rail mounted on one side of the second linear motor (5), and a mounting plate (15) is slidably connected on the second guide rail. A first electric push rod (16) and two second guide rods (17) are mounted on the mounting plate (15). The two second guide rods (17) are located on both sides of the first electric push rod (16). A translation block (18) is installed at the output end of the first electric push rod (16).

8. A resistance spot welding machine for automatically monitoring welding quality according to claim 7, characterized in that, The translation block (18) is slidably connected to the mounting plate (15), the second guide rod (17) is inserted into the translation block (18), two mounting rods (19) are fixedly connected to the inner side of the translation block (18), and a second electric push rod (20) is installed inside the translation block (18), and a welding head (21) is installed at the output end of the second electric push rod (20).

9. A resistance spot welding machine for automatically monitoring welding quality according to claim 8, characterized in that, Both sides of the welding head (21) are fixedly connected to racks (22), and both sides of the welding head (21) are provided with visual monitoring instruments (23). One end of the visual monitoring instrument (23) is fixedly connected to a gear (24). The gear (24) meshes with the rack (22) for transmission, and the gear (24) is rotatably connected to the mounting rod (19) through a central shaft. The two visual monitoring instruments (23) and the welding head (21) have the same focal point.

Images