System and method for calculating effective productivity of submerged arc welding equipment based on multi-view visual identification

Abstract

The invention discloses a submerged arc welding equipment effective productivity calculation system and method based on multi-view visual identification, and belongs to the technical field of steel structure intelligent manufacturing and industrial efficiency evaluation. According to the method, a multi-view industrial camera is deployed to collect an H-shaped steel member image, geometric parameters of the H-shaped steel member image are regressed by using a pre-trained lightweight convolutional neural network, and member identity perception without physical identification is realized; machining starting and ending time is accurately recorded through a virtual detection line, and working condition data are collected from a submerged arc welding machine PLC in real time; identifying effective welding segments based on strict multi-condition screening rules (equipment operation state, parameter deviation, duration and no serious fault); calculating the theoretical weight of the member qualified in quality inspection; and finally, the accumulated qualified weight is divided by the accumulated effective welding time to obtain an effective productivity index of a standardized unit (ton / hour), and the effective productivity index is transmitted back to the MES system. According to the method, the problems that a traditional method depends on vulnerable identification, statistics is extensive, and units are disordered are solved, and accurate, automatic and comparable productivity evaluation and production management closed loop are achieved.

Description

A System and Method for Calculating Effective Capacity of Submerged Arc Welding Equipment Based on Multi-View Visual Recognition Technical Field

[0001] This invention relates to the field of industrial intelligent manufacturing and efficiency evaluation technology, and more specifically to a system and method for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition. Background Technology

[0002] In the steel structure manufacturing industry, submerged arc welding is one of the core processing steps, and accurate assessment of its capacity is crucial for production scheduling, resource allocation, and performance management. Currently, mainstream capacity measurement methods have the following shortcomings:

[0003] High dependence on identification and low reliability: QR codes or RFID tags are widely used to bind component information. However, in the high temperature, spatter, oil, and friction environment of submerged arc welding operations, physical tags are easily damaged or soiled, resulting in a high identification failure rate, requiring frequent manual intervention, and increasing management costs.

[0004] The statistical methods are crude and lack accuracy: Traditional equipment overall efficiency (OEE) statistics or manual reporting usually use "equipment uptime" or "number of completed workpieces" as the benchmark, which cannot effectively eliminate invalid working time such as equipment idling, parameter adjustment, arc ignition failure, and welding parameter deviation, resulting in inflated production capacity data.

[0005] The evaluation units are inconsistent and the comparability is weak: There are many inconsistent units of measurement for production capacity in the industry, such as "pieces / hour", "meters / hour", and "tons / hour". For steel structure components with huge differences in specifications and weight, the first two units cannot objectively reflect the true value output. This makes it impossible to make fair and scientific horizontal comparisons of production capacity between different orders, different work groups and even different equipment, which restricts the implementation of refined management.

[0006] Therefore, there is an urgent need to propose an intelligent capacity calculation method that can break free from dependence on physical identifiers, accurately identify ineffective work time, and quantify output with a unified value unit, so as to achieve objective evaluation and data-driven management of the true effectiveness of welding processes. Summary of the Invention

[0007] The purpose of this invention is to provide a method and system for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition and MES system linkage. This invention achieves accurate, automatic, and real-time measurement of component-level effective production capacity through a technical approach of "visual self-recognition to obtain component identity, rigorous screening of effective time periods based on equipment operating conditions, and precise calculation of output using standardized weights."

[0008] To achieve the above objectives, the present invention adopts the following technical solution: a method for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition, comprising the following steps: S1: Deploying at least two industrial cameras at the production line entrance to acquire side view and top view images of the H-beam steel component to be welded, respectively; S2: Inputting the acquired images into a pre-trained lightweight convolutional neural network model, and regressing to output the key geometric parameter vector of the component, including the section height H, flange width B, and web thickness t. w flange thickness t f S3: Based on the geometric parameter vector, perform fuzzy matching in the work order database of the MES system to obtain the corresponding material grade, theoretical weight calculation rules and standard welding process parameters; S4: Preset a virtual detection line in the image coordinate system. When the front contour of the component crosses the line for the first time, record the first high-precision timestamp as the processing start time Tstart; S5: Collect welding process data stream from the submerged arc welding machine programmable logic controller in real time through industrial communication protocol. The data stream includes at least welding current, welding voltage, wire feeding speed, equipment operating status flag and fault code; S6: Deploy a verification camera at the production line exit. When the tail end of the component leaves the welding area and crosses the exit virtual detection line, record the second high-precision timestamp as the processing end time Tend; S7: Analyze the welding process data stream within the time interval [Tstart, Tend], and filter out all continuous time periods that simultaneously meet the following conditions, which are recorded as valid welding segments: (1) The equipment operating status flag is "running"; (2) The welding current and voltage values ​​are within the allowable deviation range of the standard welding process parameters; (3) (4) No preset severe fault code occurs; S8: If the component is deemed qualified by the final quality inspection, the theoretical weight W of the component is calculated according to its geometric parameter vector and the theoretical weight calculation formula corresponding to the material grade; S9: The sum of the theoretical weights of all qualified components within the preset time period is accumulated and divided by the total duration of the effective welding segments corresponding to all qualified components to obtain the effective production capacity index P in "tons / hour". eff S10: The effective production capacity index P eff The data is fed back to the MES system for dynamic production scheduling optimization and production performance management.

[0009] Furthermore, the lightweight convolutional neural network model in step S2 is based on the EfficientNet architecture and is fine-tuned on a labeled image dataset containing various H-beam specifications through transfer learning, balancing recognition accuracy and real-time requirements.

[0010] Furthermore, in steps S4 and S6, the image coordinate system is aligned with the production line machine coordinate system through camera calibration, and the position of the virtual detection line is precisely defined; the component crossing time is determined by sub-pixel level Canny edge detection combined with linear interpolation algorithm to ensure that the timestamp accuracy is not less than ±10 milliseconds.

[0011] Furthermore, in step S5, Modbus TCP or Profinet protocol is used to communicate with the PLC of the submerged arc welding machine, and the data acquisition frequency is not less than 1Hz to ensure the time resolution of the working condition data.

[0012] Furthermore, in step S7, the preset threshold is set to 30 seconds to filter out short-term start-stop or interference; the severe fault codes include at least E01, which indicates arc ignition failure, and E05, which indicates wire feed blockage.

[0013] Furthermore, in step S8, the formula for calculating the theoretical weight W is: W = L × [2 × B × t] f +(H-2×t f )×t w ]×ρ, where ρ is the density of steel, typically taken as 7.85×10 -6 kg / mm³. This formula is derived from the calculation of the cross-sectional area and density conversion of H-beams, and the result is expressed in tons (t).

[0014] Furthermore, in step S9, the effective production capacity index P eff The calculation formula is:

[0015] P eff =∑Wi / ∑Ti, where the unit is t / h.

[0016] Where Wi is the theoretical weight (t) of the i-th qualified component, and Ti is the total effective welding time (h) of the i-th component.

[0017] Furthermore, the present invention also provides a system for performing the above-described method, comprising:

[0018] Image acquisition modules are deployed at the entrance and exit of the production line to acquire multi-view images of components.

[0019] A visual recognition module, connected to the image acquisition module, has a built-in pre-trained neural network model for regressing the geometric parameters of components from images;

[0020] The data acquisition module communicates with the PLC of the submerged arc welding machine to collect equipment operating data in real time; the MES interface module is used to interact with the MES system to realize work order information matching and result feedback.

[0021] The data processing and analysis module is connected to the visual recognition module, data acquisition module and MES interface module respectively, and is used to perform timestamp recording, effective welding segment screening, theoretical weight calculation and effective production capacity index calculation.

[0022] The modules work together to automate the entire process from data perception, processing and analysis to result feedback.

[0023] The present invention has the following beneficial effects:

[0024] High-reliability perception, free from the constraints of physical identification: By using multi-view vision and deep learning to directly identify the geometric features of components, the problem of identification failure and maintenance costs caused by easily damaged identification such as QR codes is completely avoided. It adapts to the complex and harsh environment of the welding workshop and achieves highly robust automatic identification of component identity.

[0025] Refined chemical condition screening enables the identification of true production capacity: The innovative definition of four strict criteria for "effective welding segments" refines the production capacity statistics benchmark from the general "equipment start-up time" to "quality stable operation time that meets process specifications", effectively eliminating various invalid and low-quality operating conditions and ensuring the authenticity and accuracy of production capacity data.

[0026] Standardized value measurement establishes a fair benchmark for comparison: By using "ton" as the unit of output weight and "ton / hour" as the effective capacity indicator, it reflects the actual material consumption and value output of components, providing a scientific and fair benchmark for performance comparison of products of different specifications and weights, and promoting refined performance management across departments and cycles.

[0027] Forming a data-driven closed loop to empower intelligent production management: Real-time calculated effective capacity indicators are fed back to the MES system, providing direct and objective data support for dynamic adjustments to production plans (such as scheduling optimization), accurate assessment of personnel teams (such as KPI calculation), and predictive maintenance of equipment (such as based on effective working hours analysis), thus realizing a management closed loop from perception, analysis to decision optimization. Attached Figure Description

[0028] Figure 1 is a flowchart of the calculation method according to an embodiment of the present invention;

[0029] Figure 2 is a schematic diagram of the hardware deployment and data flow of the system involved in the embodiment of the present invention. Detailed Implementation

[0030] The following describes in further detail the specific implementation of the effective capacity calculation system and method for submerged arc welding equipment based on multi-view visual recognition according to the present invention, with reference to Figures 1 and 2.

[0031] The effective capacity calculation system for submerged arc welding equipment based on multi-view visual recognition of this invention mainly includes the following parts:

[0032] The image acquisition module is deployed at the entrance and exit of the production line to acquire multi-view images of the components; at least two industrial cameras (such as one side-view camera and one top-view camera) are used. The cameras are preferably equipped with protective covers and active lighting devices to adapt to changes in workshop dust, vibration and lighting.

[0033] A visual recognition module, connected to the image acquisition module, has a built-in pre-trained neural network model for regressing the geometric parameters of components from images;

[0034] The data acquisition module communicates with the PLC of the submerged arc welding machine to collect equipment operating data in real time.

[0035] The MES interface module is used to interact with the MES system to achieve work order information matching and result feedback.

[0036] The data processing and analysis module is connected to the vision recognition module, data acquisition module, and MES interface module, respectively, and is used to perform timestamp recording, effective welding segment screening, theoretical weight calculation, and effective production capacity index calculation. It is typically an edge computing server or industrial computer deployed on-site in the workshop, integrating vision recognition, data analysis, and logic control software.

[0037] A method for calculating the effective capacity of submerged arc welding equipment based on multi-view visual recognition, combined with an embodiment applied to an H-beam submerged arc welding production line, is described below:

[0038] Step 1: Visual recognition and parameter acquisition (corresponding to claims S1-S2)

[0039] A camera assembly is set up at a suitable location in front of the welding station entrance. When the H-beam enters the shooting area, the camera is triggered synchronously to capture side and top view images respectively. The captured images are sent to the data processing unit.

[0040] The data processing unit runs a pre-trained lightweight convolutional neural network model (e.g., a model trained using a large number of H-beam sample images and corresponding geometric parameter annotations based on the EfficientNet architecture). This model processes the input images and directly regresses the key geometric parameter vector (H, B, t) of the output component. w ,t f ,L).

[0041] Step 2: Dynamic matching of process information (corresponding to claim S3)

[0042] After obtaining the geometric parameter vector, the system initiates a query request to the MES system through the MES interface module. The query condition is the work order record that is closest to the geometric parameter vector (fuzzy matching can be performed using Euclidean distance, setting percentage tolerance, etc.). The MES system returns the matched work order information, which includes at least: the component material (e.g., Q355B), the corresponding theoretical weight calculation formula (or density parameter), and the standard current and voltage values ​​and their deviation range allowed for welding components of this specification.

[0043] Step 3: Defining the processing time window (corresponding to claims S4 and S6)

[0044] In the image processing software, a "virtual detection line" perpendicular to the component's travel direction is preset in the image coordinate system based on camera calibration parameters. When the front contour of the component is detected to have completely crossed this line for the first time, the system records a high-precision timestamp as the processing start time Tstart for that component in this process. Similarly, another virtual detection line is set in the verification camera view at the production line exit. When the tail end of the component leaves the welding area and crosses this line, the processing end time Tend is recorded. Through sub-pixel level edge detection algorithms and linear interpolation, the timestamp accuracy can reach the millisecond level.

[0045] Step 4: Real-time acquisition of operating data and screening of "effective segments" (corresponding to claims S5 and S7)

[0046] Within the [Tstart, Tend] time interval, the system continuously acquires data streams from the PLC of the submerged arc welding machine at a frequency of not less than 1Hz using industrial protocols such as Modbus TCP. The data items include at least: welding current I, welding voltage U, equipment operating status flags, and fault codes.

[0047] The data stream is fed into a sliding time window for analysis. The system filters out consecutive time periods that simultaneously meet all of the following conditions, defining them as valid welding segments:

[0048] The equipment's operating status is marked as "Running".

[0049] The real-time current I and voltage U are within the allowable deviation range (e.g., ±10%) of the standard process parameters obtained in step two;

[0050] The duration of this continuous state is not less than a preset threshold (e.g., 30 seconds, to filter out brief starts, stops, or disturbances).

[0051] During this period, no pre-set serious faults, such as arc ignition failure (E01) or wire feed blockage (E05), were reported.

[0052] The sum of the durations of all selected valid welding segments is the effective welding time for that component.

[0053] Step 5: Output accounting and capacity calculation (corresponding to claims S8 and S9)

[0054] The component proceeds to subsequent quality inspection processes. If it is ultimately determined to be a qualified product, output accounting is initiated. This is based on its geometric parameters (H, B, t). w ,t f Calculate the theoretical weight W using the material density ρ obtained from MES (L) and L. The preferred calculation formula is: W = L × [2 × B × t] f +(H-2×t f )×t w ]×ρ, where ρ is usually taken as 7.85×10 -6 kg / mm³, the calculation result is in tons (t).

[0055] To calculate the effective production capacity P over a specific time period (such as a shift or a day) eff The system accumulates the sum of the theoretical weights ΣWi of all qualified components within this time period, and the sum of the effective welding times ΣTi (units converted to hours) for these components. Therefore, the effective production capacity is:

[0056] P eff =ΣWi / ΣTi (Unit: tons / hour, t / h)

[0057] Step Six: Data Feedback and Application (corresponding to claim S10)

[0058] Calculated P eff The metrics are transmitted back and written to the MES database via the MES interface module. This data can be used for:

[0059] Dynamic production scheduling: Planners adjust the timing of subsequent work orders based on the real-time and actual output capacity of the equipment.

[0060] Precise performance appraisal: based on objective, value-quantifiable P... eff It serves as the core basis for the KPIs of work teams or machines.

[0061] Equipment efficiency analysis: By combining the ratio of effective welding time to total uptime, analyze equipment utilization and identify the causes of non-value-added time.

[0062] Results of this embodiment: After implementing this invention in a steel structure factory, compared with the traditional statistical method based on operating time, the accuracy of effective production capacity data was significantly improved. Actual tests show that the system can automatically identify component parameters even without QR code identification, with an accuracy rate of over 96% for identifying effective welding time. The unified effective production capacity indicator measured in "tons / hour" provides a fair benchmark for comparing the production efficiency of products of different specifications, resulting in an increase in the plan achievement rate of approximately 5 percentage points and enhanced management precision.

[0063] It should be noted that the above embodiments, using H-beams as an example, are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments, and is also applicable to the welding capacity measurement of other steel structural components with regular cross-sections (such as box columns, steel pipes, etc.). All technical solutions falling within the scope of the present invention's concept are within the protection scope of the present invention. Any improvements and modifications made by those skilled in the art without departing from the principles of the present invention, such as adaptive adjustments and replacements to camera deployment methods, neural network model architectures, communication protocols, and screening thresholds, should also be considered within the scope of protection of the present invention.

Claims

1. A method for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition, characterized in that, Includes the following steps: S1. Deploy at least two industrial cameras at the production line entrance to acquire side and top view images of the H-beam steel component to be welded, respectively. S2. Input the acquired images into a pre-trained lightweight convolutional neural network model to regress and output the key geometric parameter vectors of the component, including section height H, flange width B, web thickness tw, flange thickness tf, and component length L. S3. Based on the geometric parameter vectors, perform fuzzy matching in the work order database of the MES system to obtain the corresponding material grade, theoretical weight calculation rules, and standard welding process parameters. S4. Preset a virtual detection line in the image coordinate system; when the front contour of the component first crosses this line, record the first high-precision measurement. The timestamp is used as the processing start time Tstart; S5, the welding process data stream is collected in real time from the submerged arc welding machine programmable logic controller through the industrial communication protocol. The data stream includes at least welding current, welding voltage, wire feeding speed, equipment operating status flag and fault code; S6, a verification camera is deployed at the production line exit. When the tail end of the component leaves the welding area and crosses the exit virtual detection line, the second high-precision timestamp is recorded as the processing end time Tend; S7, the welding process data stream is analyzed within the time interval [Tstart, Tend], and all continuous time periods that simultaneously meet the following conditions are selected as valid welding segments: (1) the equipment operating status flag is "running"; (2) the welding current and voltage values ​​are within the allowable deviation range of the standard welding process parameters; (3) the continuous duration is not less than the preset threshold. (4) No preset severity fault code occurred; S8. If the component is deemed qualified by the final quality inspection, the theoretical weight W of the component is calculated according to its geometric parameter vector and the theoretical weight calculation formula corresponding to the material grade. S9. The sum of the theoretical weights of all qualified components within a preset time period is accumulated and divided by the total duration of the effective welding segments corresponding to all qualified components to obtain the effective production capacity index P in "tons / hour". eff S10, the effective production capacity index P eff The data is fed back to the MES system for dynamic production scheduling optimization and production performance management.

2. The method for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition according to claim 1, characterized in that, In step S2, the lightweight convolutional neural network model is a model obtained by transfer learning and fine-tuning based on the EfficientNet architecture.

3. The method for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition according to claim 1, characterized in that, In steps S4 and S6, the position of the virtual detection line is aligned with the production line mechanical coordinate system through camera calibration, and the component crossing time is determined by sub-pixel level edge detection and linear interpolation, with a timestamp accuracy of not less than ±10 milliseconds.

4. The method for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition according to claim 1, characterized in that, In step S5, the industrial communication protocol is Modbus TCP or Profinet, and the data acquisition frequency is not less than 1Hz.

5. The method for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition according to claim 1, characterized in that, In step S7, the preset threshold is 30 seconds; the severe fault codes include at least the fault codes corresponding to arc ignition failure and wire feed blockage.

6. The method for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition according to claim 1, characterized in that, In step S8, the formula for calculating the theoretical weight W is: W = L × [2 × B × t] f +(H-2t f )×t w ]×ρ, where ρ is the density of steel, taken as 7.85×10 -6 kg / mm 3 .

7. The method for calculating the effective production capacity of submerged arc welding equipment based on multi-view visual recognition according to claim 1, characterized in that, In step S9, the effective production capacity index P eff The calculation formula is: P eff =∑Wi / ∑Ti, in t / h, where Wi is the theoretical weight of the i-th qualified component and Ti is the total effective welding time of the i-th component.

8. A submerged arc welding equipment effective capacity calculation system based on multi-view visual recognition, used to execute the method as described in any one of claims 1-7, characterized in that, include: Image acquisition modules are deployed at the entrance and exit of the production line to acquire multi-view images of components. A visual recognition module, connected to the image acquisition module, has a built-in pre-trained neural network model for regressing the geometric parameters of components from images; The data acquisition module communicates with the PLC of the submerged arc welding machine to collect equipment operating data in real time. The MES interface module is used to interact with the MES system to achieve work order information matching and result feedback. The data processing and analysis module is connected to the visual recognition module, data acquisition module and MES interface module respectively, and is used to perform timestamp recording, effective welding segment screening, theoretical weight calculation and effective production capacity index calculation.

Images