A method and system for processing multi-source sensing data of a welding operation

Abstract

This invention relates to the field of industrial data processing technology, specifically to a method and system for processing multi-source sensor data in welding operations. The method includes the following steps: acquiring terminal messages, calculating timing deviations to generate a synchronization message chain; filtering multi-pose recorded information based on this chain to generate welding judgment criteria; calling records to compare the relationship between oscillation and interference gaps to generate operation tags; judging the relationship between tags and voice feedback to generate broadcast records; and generating a processing scheme based on the order of multiple occupancy adjustments. In this invention, the underlying timing consistency is constructed by accurately calculating the equipment feedback time difference; the true state is locked by deeply stripping away secondary shaft interference and meaningless flipping features, completely breaking the limitations of misjudgment in a single motion cycle; the instruction occupancy status of concurrent multi-source events is analyzed synchronously; and the global attribution and triggering order are dynamically adjusted to resolve signal conflicts and delays, ensuring the accuracy of multi-source sensor information fusion judgment under complex working conditions.

Description

Technical Field

[0001] This invention relates to the field of industrial data processing technology, and in particular to a method and system for processing multi-source sensor data of welding operations. Background Technology

[0002] The field of industrial data processing technology involves the collection, organization, transformation, storage, and retrieval of operational data from industrial equipment, personnel operations, environmental conditions, and sensor sampling data. Its core aspects include uniformly recording data from diverse sensors, controllers, and operational terminals according to sampling time, data type, equipment number, work object, and operating condition. This data is then processed through field matching, format conversion, outlier removal, time series alignment, and threshold determination to create data content that can be read and compared later. This technology typically establishes data processing rules based on the action status, temperature status, continuous operation time, equipment signals, and on-site environmental information during industrial production processes, enabling data from diverse sources to enter databases or local storage units according to a unified standard. The traditional multi-source sensor data processing method and system for welding operations refers to the processing method for collecting and organizing multiple types of data during welding operations, including hand movements, continuous working time, and ambient temperature. The technical aspects it addresses include posture sampling data, temperature sampling data, timing data, and voice prompt data from the welding personnel's worn equipment. Typically, an ESP32-C3 main controller reads hand movement data from an MPU6050 six-axis posture sensor, determining a valid working state based on regular movements lasting more than ten minutes, and accumulating continuous working time. A DS18B20 temperature sensor collects ambient temperature data at fixed intervals, comparing the sampled values ​​with thresholds of 65 degrees Celsius and 90 degrees Celsius. A JQ8900 voice chip reads preset voice content based on working time and temperature thresholds and broadcasts it through a speaker. Simultaneously, a sleep state is entered based on a 15-minute period of inactivity, and the timing record is restarted upon the detection of renewed hand movement.

[0003] Traditional welding multi-source data processing relies on independent equipment to read the action and ambient temperature at predetermined cycles. This decentralized operation mechanism results in inherent time differences when multiple types of data are transmitted back, causing discontinuities in the time series between the action status and the temperature rise record. Judging the status solely based on continuous motion characteristics can easily lead to misjudging on-site interference as valid time periods. Mechanized comparison and triggering mechanisms often ignore the positional conflicts of multiple on-site commands, resulting in data omissions and broadcast delays. This makes it difficult for the entire processing system to build accurate judgment criteria under complex and variable working conditions. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a multi-source sensor data processing method for welding operations, comprising the following steps: S1: Obtain the welding station start message, collect the welding station multi-source terminal messages, calculate the timing deviation between the sampling timestamp and the edge clock, merge the message order according to the timing deviation, and generate a synchronization message chain; S2: Based on the synchronous message chain, filter the bevel oscillation, grip area heating and voice reply records in the multi-posture action message and heating message to generate the welding judgment caliber; S3: Based on the welding determination caliber, call the records of bevel oscillation, axial flipping, secondary shaft interference, back of hand tilt angle and communication gap, compare the relationship between oscillation and interference gap, and generate a job identification label; S4: Based on the operation identification tag, call the holding area temperature rise record and voice reply record, determine the relationship between welding swing tag, temperature rise, voice occupation and voice reply, and generate temperature broadcast record; S5: Based on the temperature broadcast records, adjust the position order according to the multiple positions in the work window, and the relationship between voice assignment and high temperature assignment to generate a multi-source sensor data processing scheme.

[0005] As a further aspect of the present invention, the synchronization message chain includes calibrated node timestamps, reconstructed terminal message sequences, and compensated clock differences; the welding judgment caliber includes a preset welding posture deflection threshold, a quantified bevel swing range, and a defined heating rate standard; the operation identification tag includes a matched abnormal flip category, a classified axial interference type, and a marked communication disconnection period; the temperature broadcast record includes the recorded peak temperature of the gripping area, the output voice warning text, and the feedback broadcast response delay; and the multi-source sensor data processing scheme includes an adjusted sensor acquisition frequency, an allocated system alarm weight, and an optimized data cache size.

[0006] As a further aspect of the present invention, the timing deviation merged message order refers to the synchronization message order formed after time calibration, sorting and merging of multi-source terminal messages based on the timing deviation between the sampling timestamps of multiple messages and the edge clock.

[0007] As a further aspect of the present invention, the relationship between the swing and the interference gap refers to the correlation formed by comparing the axial flipping and back-of-hand tilting action characteristics in the bevel swing with the secondary shaft interference and communication gap abnormal characteristics.

[0008] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Obtain welding station start message, collect and parse multi-source terminal messages, obtain node, type and time stamp field, remove missing time stamp messages, check duplicate messages by node number and sampled time stamp, and obtain message time stamp field set; S102: Based on the message timestamp field set, call the node sampling timestamp and the edge clock time, calculate the time difference in the same time dimension, mark the sampling ahead state and the sampling lag state, associate the message identifier field with the timing deviation amplitude, and establish a timing deviation index table. S103: According to the timing deviation index table, merge the messages according to the message arrival sequence number and the difference magnitude. For messages with the same difference magnitude, determine the parallel order according to the edge clock time, map the node number and the load position, and generate a synchronization message chain.

[0009] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the synchronization message chain, filter multiple types of welding action messages according to message type identifier, verify the continuous correspondence between message time sequence number and load position, associate action attributes, node identifier and time interval, and generate action message field set; S202: Based on the action message field set, match the records of bevel oscillation, grip area heating, and voice reply in the heating message, compare the action category with the heating message sequence number, remove fields with inconsistent time periods, and establish a welding record mapping table. S203: Based on the welding record mapping table, determine the consistency relationship between the action type, bevel swing amplitude, holding area temperature rise amplitude, and voice reply status, associate the judgment identifier with the message node number, and generate the welding judgment caliber.

[0010] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Based on the welding judgment caliber, call the records of axial flip, secondary shaft interference, back of hand tilt angle, and communication gap, match the swing direction, wrist flip mark, posture change mark, and message gap position according to the time interval, remove the time interval misalignment field, and generate the bevel action field table. S302: Based on the bevel action field table, compare the correspondence between the swing action characteristics and axial flip, back of hand tilt angle change, and secondary shaft interference, associate the message gap position with the communication gap record, and establish an operation correspondence table; S303: Based on the job correspondence table, determine the caliber number, swing direction, wrist flip mark, posture change mark, message gap position and job category mark according to the same time period interval mapping, and generate job identification label.

[0011] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Based on the job identification tag, call the heating record of the welding torch holding area and the voice broadcast reply record, match the welding swing, temperature sampling, broadcast occupation and reply time according to the tag time period, and remove misaligned data to obtain the broadcast temperature field table. S402: Based on the broadcast temperature field table, determine the temperature increase record based on the difference of adjacent temperature sampling values ​​with the same dimension, compare the welding swing label with the time sequence number of the temperature increase record, associate the broadcast occupation identifier and the receipt time, and establish a swing temperature rise mapping table. S403: Based on the swing temperature rise mapping table, establish the correspondence between welding swing, temperature increase, voice broadcast occupation and voice receipt according to the same label time period, map the operation category, temperature rise status, broadcast status and receipt status, and generate temperature broadcast records.

[0012] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Based on the temperature broadcast record, match welding swing occupants, pause occupants, interference occupants and high temperature rejection occupants according to the window number, compare the overlapping state of the start and end times of the occupants, reject the occupant items outside the window, and generate an occupant registration table. S502: Based on the occupancy registration table, correct the window occupancy according to the mutual exclusion relationship between swinging occupancy and paused occupancy, compare the overlapping state of interference occupancy and high temperature rejection occupancy, associate risk categories, and establish a risk attribution table. S503: Based on the risk attribution table, map window number, placeholder category, risk category and temperature broadcast record sequence number, adjust the placeholder registration and risk attribution relationship of the operation window, and generate a multi-source sensor data processing scheme.

[0013] A multi-source sensor data processing system for welding operations includes: The multi-source message synchronization module is used to implement S1: acquiring the welding station start message, collecting multi-source terminal messages of the welding station, calculating the timing deviation between the sampling timestamp and the edge clock, merging the message order according to the timing deviation, and generating a synchronization message chain; The welding judgment generation module is used to implement S2: based on the synchronous message chain, filter the bevel swing, grip area heating and voice reply records in the multi-posture action message and heating message to generate the welding judgment caliber; The swing interference identification module is used to implement S3: based on the welding judgment caliber, call the axial flip, secondary shaft interference, hand back tilt angle and communication gap record in the bevel swing record, compare the relationship between swing and interference gap, and generate a job identification label; The temperature voice broadcast module is used to implement S4: based on the operation identification tag, call the holding area temperature rise record and voice reply record, determine the relationship between welding swing tag, temperature rise, voice occupation and voice reply, and generate temperature broadcast record; The sensor data processing module is used to implement S5: based on the temperature broadcast record, according to the multiple occupants in the work window, adjust the occupant order, and the relationship between voice attribution and high temperature attribution, generate a multi-source sensor data processing scheme.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, multi-source terminal messages are acquired and the timing deviation between the sampling timestamp and the edge clock is calculated and merged to effectively eliminate the backhaul time difference between different devices and build underlying timing consistency. Based on this, multiple types of recorded information are simultaneously screened and the relationship between oscillation and interference gaps is compared. By deeply stripping the secondary shaft interference and meaningless flip features, the true working state is accurately locked, breaking the limitation of misjudgment in a single motion cycle. Multiple heating and voice feedback records are simultaneously invoked to deeply analyze the occupancy status of multi-source events. The attribution and triggering order are dynamically adjusted according to the multiple concurrent occupancy in the actual working window, completely resolving the instruction conflict and delay problem caused by fixed-cycle comparison. This multi-dimensionally ensures the accuracy of multi-source sensor information fusion and judgment under complex working conditions. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0019] This embodiment provides a multi-source sensor data processing method for welding operations. In practical applications, such as in an edge collaborative processing environment at a welding station, the station control terminal, posture acquisition terminal, grip area temperature acquisition terminal, voice broadcast terminal, and edge processing node continuously interact around the welding operation. The station control terminal outputs a start message when the welding torch enters the operation state. The posture acquisition terminal records flat welding, vertical welding, overhead welding, hand raising, wrist turning, and posture changing actions. The grip area temperature acquisition terminal records the heating status of the welding torch grip area. The voice broadcast terminal records the warning broadcast occupancy and acknowledgment status. The edge processing node continuously processes the message timing, action correspondence, heating correspondence, voice acknowledgment, and operation window occupancy.

[0020] Please see Figure 1 and Figure 2 The specific steps of S1 are as follows: Obtain the welding station start message, collect multi-source terminal messages from the welding station, obtain the timing deviation between the node sampling timestamp and the edge clock time, merge the message arrival order according to the timing deviation, and generate a synchronization message chain. The welding station start message is a communication message sent by the station control terminal to the edge processing node in the operation start state, carrying the station identifier, operation start state, edge clock time, and acquisition permission state. The multi-source terminal messages are structured messages uploaded by the attitude acquisition terminal, grip area temperature acquisition terminal, voice broadcast terminal, and communication gateway, carrying the node identifier, message type, node sampling timestamp, edge reception time, load position, and message status. The node sampling timestamp is the time stamp written to the message load by the terminal when acquiring action, temperature, or voice status. The edge clock time is the time reference used by the edge processing node when receiving or calibrating messages. The timing deviation is the relative difference between the node sampling timestamp and the edge clock time, used to determine the sampling ahead state, sampling behind state, and synchronization state. The synchronization message chain is an ordered data object output by the edge processing node. It carries the calibrated node timestamp, rearranged terminal message sequence, compensated clock difference, node number mapping relationship and load position mapping relationship, which are used for subsequent welding action screening and temperature rise record matching.

[0021] S101: Obtain the welding station start message, collect and parse multi-source terminal messages, obtain node, type, and time stamp fields, remove messages with missing timestamps, and verify duplicate messages by node number and sampling timestamp to obtain the message timestamp field set. The message timestamp field set is a set of fields formed after parsing and cleaning, carrying node number, message type, sampling timestamp, edge reception time, load position, and abnormal status identifier. After receiving the start message, the edge processing node opens the message receiving channel and parses the fields of messages uploaded by the attitude acquisition terminal, grip area temperature acquisition terminal, voice broadcast terminal, and communication gateway. Messages that fail to be parsed are marked as having format errors and written to the traceability record, and are not included in the timing merging. Messages with missing sampling timestamps are marked as having a missing timestamp status. The traceability record saves the source node, message type, and reception status of the message. Subsequent steps only receive messages with retained timestamp fields. For duplicate messages with the same node number and sampling timestamp, messages with complete edge reception status and resolvable payload location are retained. The duplicate status is written into the message timestamp field set so that the downstream timing processing does not change the message arrival order due to duplicate uploads.

[0022] S102: Based on the message timestamp field set, retrieve the node sampling timestamp and the edge clock time, obtain the time difference using the same time base, mark the sampling ahead state and sampling behind state, associate the message identifier field with the timing deviation amplitude, and establish a timing deviation index table. The timing deviation index table is an index object oriented towards message sorting, carrying the message identifier, node number, message type, sampling timestamp, edge clock time, ahead or behind state, difference amplitude state, and payload location. The edge processing node first converts the node sampling timestamp and edge clock time in the message timestamp field set to a consistent time base, and then marks the sampling ahead or behind state according to the chronological relationship between the two types of times. The message identifier field is used to maintain the traceability relationship of messages between the acquisition end, gateway, and edge processing node, and the timing deviation amplitude is used to express the degree to which the message deviates from the edge clock base. For messages that cannot complete the time base conversion, write a conversion exception status, retain their message identifier and source node, do not participate in the synchronous message chain sorting, but can be retrieved by the traceability record.

[0023] S103: Based on the timing deviation index table, messages are merged according to their arrival sequence number and difference magnitude. For messages with the same difference magnitude, their parallel order is determined by the edge clock time. Node numbers and load positions are mapped to generate a synchronization message chain. The message arrival sequence number is the sequential identifier formed when the edge processing node receives the message. The load position is the storage location of the action, temperature, voice, or communication status fields in the message load within the message structure. The edge processing node first maintains the message transmission order based on the message arrival sequence number, and then corrects the misalignment caused by inconsistent sampling timing between terminals based on the difference magnitude. When the difference magnitude is the same, the edge clock time is used as the basis for determining the parallel order. The generated synchronization message chain retains the original message identifier, the calibrated node timestamp, the rearranged terminal message sequence, and the compensated clock difference, and maps the node number to the corresponding load position. In the subsequent welding action screening stage, the message type, calibration timing, and load position in the synchronization message chain are directly read to avoid mismatches in action and temperature records caused by inconsistent arrival times of cross-terminal messages.

[0024] Please see Figure 1 and Figure 3 The specific steps of S2 are as follows: Based on the synchronous message chain, filter the bevel oscillation, grip area temperature rise, and voice feedback records from flat welding, vertical welding, overhead welding, hand raising, wrist turning, posture changing action messages and temperature rise messages to generate a welding judgment calibrator. The welding judgment calibrator serves as the judgment benchmark object for the operation identification stage, carrying the preset welding posture deflection judgment benchmark, the quantified bevel oscillation range, the defined temperature rise change standard, action category, node identifier, and time interval. Bevel oscillation refers to the oscillation state of the welding torch along the bevel direction recorded in the posture acquisition terminal or the associated load of the temperature rise message, carrying the oscillation direction, oscillation range, and corresponding time period. Grip area temperature rise refers to the temperature change record formed by the grip area temperature acquisition terminal, carrying the temperature increase state, peak state, and acquisition time period. Voice feedback records are the response state returned by the voice broadcast terminal after broadcasting the warning message, carrying the broadcast occupation, feedback arrival, and feedback waiting states. The welding judgment calibrator connects the action messages, temperature rise messages, and voice feedback records into judgmentable objects within the same operation time period.

[0025] S201: Based on the synchronization message chain, filter multiple types of welding action messages by message type identifier, verify the continuous correspondence between message sequence number and load position, associate action attributes, node identifiers, and time intervals, and generate an action message field set. The action message field set is a data object used to carry welding posture and hand action information, including flat welding, vertical welding, overhead welding, hand raising, wrist turning, and posture changing action categories, as well as their node sources, calibration sequence, and load positions. Edge processing nodes read the message type identifiers in the synchronization message chain, separate posture-related messages from non-posture-related messages, and verify whether the sequence number of the posture message in the synchronization message chain maintains a continuous correspondence with the load position. Messages whose load positions cannot correspond to action attributes are marked as having abnormal action loads and are subsequently only saved as traceability objects. The action attribute, after being associated with the node identifier and time interval, forms the action message field set, providing action category and time period entry points for matching heating messages.

[0026] S202: Based on the action message field set, match the bevel oscillation, grip area heating, and voice feedback records in the heating message, compare the action category with the heating message sequence number, remove fields with inconsistent time periods, and establish a welding record mapping table. The welding record mapping table is a data object connecting action messages and heating messages, carrying the action category, bevel oscillation record, grip area heating record, voice feedback status, corresponding node, and time interval. The edge processing node uses the action category and time interval in the action message field set as the matching entry point, reads the heating message payload in the synchronization message chain, and aligns the bevel oscillation field, grip area heating field, and voice feedback field with the action time period. Fields with inconsistent action time periods and heating message sequence numbers are marked as time period misalignment states and are not entered into the welding record mapping table, but the source message identifier is retained. After matching is completed, the welding record mapping table outputs the same-segment relationship between actions, oscillation, heating, and feedback to the determination caliber generation stage.

[0027] S203: Based on the welding record mapping table, determine the consistency relationship between action category, bevel oscillation amplitude, grip area temperature rise amplitude, and voice feedback status, associate the judgment identifier with the message node number, and generate the welding judgment caliber. The judgment identifier is a traceable identifier formed by the edge processing node based on the action category, bevel oscillation amplitude status, grip area temperature rise amplitude status, and voice feedback status, pointing to the message node number and time interval involved in the judgment. The preset welding posture deflection judgment benchmark is written into the edge processing node by the station process configuration, used to distinguish whether flat welding, vertical welding, overhead welding, and hand posture changes fall within the process allowable state; the bevel oscillation amplitude range is formed by the oscillation state in the posture load after benchmarking, used to express the oscillation direction and oscillation amplitude category; the temperature rise change standard is formed by the continuous temperature rise record of the grip area temperature acquisition terminal, used to characterize whether the heating change of the grip area corresponds to the welding action. The voice feedback status is used to confirm whether the warning broadcast has been received by the terminal. The welding judgment caliber integrates the above objects into the traceability structure as input for bevel action recognition and communication gap analysis.

[0028] Please see Figure 1 and Figure 4 The specific steps of S3 are as follows: Based on the welding judgment caliber, the axial flip, secondary shaft interference, hand back tilt angle, and communication gap records in the bevel oscillation of the weldment are retrieved. The correspondence between the oscillation direction, wrist flip, posture change, and message gap is compared to generate a job identification tag. The job identification tag is a tag object that classifies the welding job status, carrying the abnormal flip category, axial interference type, communication disconnection period, oscillation direction, wrist flip identifier, posture change identifier, and message gap position. Axial flip is the flip state that occurs along the main axis direction of the weldment in the bevel oscillation record. Secondary shaft interference is an additional oscillation or external disturbance state that is inconsistent with the main axis direction. Hand back tilt angle is the hand back tilt state recorded by the posture acquisition terminal. Communication gap is the disconnection record formed in the synchronous message chain when a valid message is not received at the position where a message should be received. The job identification tag associates the action state, interference state, and communication state into the same job category.

[0029] S301: Based on the welding judgment caliber, the records of axial flip, secondary shaft interference, hand back tilt angle, and communication gap are retrieved. The swing direction, wrist-flipping indicator, posture change indicator, and message gap position are matched according to the time interval. Fields with time interval misalignment are removed, and a bevel action field table is generated. The bevel action field table is a data object carrying information related to bevel swing, recording the judgment caliber number, swing direction, axial flip status, secondary shaft interference status, hand back tilt angle change status, wrist-flipping indicator, posture change indicator, and message gap position. Edge processing nodes use the time interval and message node number in the welding judgment caliber as the retrieval entry point, reading the axial flip and secondary shaft interference fields from the bevel swing records. Simultaneously, the hand back tilt angle field is read from the attitude message load, and the communication gap record is read from the abnormal status record of the synchronization message chain. Fields inconsistent with the judgment caliber time interval are marked as misaligned, their source is retained, and they are not entered into the bevel action field table. The generated bevel action field table maintains the correspondence between swing, wrist-flipping, posture change, and gap position within the same time interval.

[0030] S302: Based on the bevel action field table, compare the correspondence between the swing action characteristics and axial flip, hand back tilt angle change, and secondary shaft interference, associate the message gap position with the communication gap record, and establish a job correspondence table. The job correspondence table is an intermediate data object used to determine the job category, carrying the swing action characteristics, flip category, interference type, hand back tilt angle change state, communication gap position, and judgment caliber number. The edge processing node compares the swing direction with the axial flip state in the bevel action field table to determine whether the flip state is consistent with the current swing direction; it compares the hand back tilt angle change state with the wrist flip icon and posture change icon to confirm whether the hand posture change is consistent with the action message category; it compares the secondary shaft interference state with the swing action characteristics to distinguish between the job state where the welding torch swings along the expected bevel direction and the job state that deviates after being disturbed by the secondary shaft. After associating the message gap position with the communication gap record, the job correspondence table retains the location of the communication disconnection, the involved nodes, and the affected time period, providing a basis for the job category in the tag generation stage.

[0031] S303: Based on the job correspondence table, determine the caliper number, swing direction, wrist-flipping indicator, posture change indicator, message gap position, and job category identifier according to the same time period interval mapping, and generate a job identification tag. The job category identifier is a classification identifier formed by the edge processing node according to the swing, wrist-flipping, posture change, interference, and gap relationships in the job correspondence table, pointing to the abnormal flipping category, axial interference type, and communication disconnection period. When generating the job identification tag, the caliper number is used to trace back the action, temperature rise, and voice feedback source; the swing direction is used to determine the welding swing state; the wrist-flipping and posture change indicators are used to limit changes in hand movements; and the message gap position is used to limit the impact of communication anomalies on the current time period determination. The job identification tag is output to the temperature voice broadcast processing stage, enabling the temperature rise record of the holding area and the voice broadcast feedback to be associated according to the job category.

[0032] Please see Figure 1 and Figure 5 The specific steps of S4 are as follows: Based on the job identification tag, the heating record of the welding torch holding area and the voice broadcast receipt record are retrieved. The correspondence between the welding oscillation tag, temperature increase record, voice broadcast occupancy record, and voice broadcast receipt record is determined, and a temperature broadcast record is generated. The temperature broadcast record is a data object oriented towards the risk handling stage, carrying the peak temperature status of the holding area, the voice warning text, the broadcast occupancy status, the voice receipt status, the receipt response interval, the job category, and the temperature rise status. The welding oscillation tag is the tag field related to bevel oscillation in the job identification tag. The temperature increase record is the heating direction record formed by continuous acquisition by the holding area temperature acquisition terminal. The voice broadcast occupancy record is the channel status of the voice broadcast terminal that is broadcasting or waiting to broadcast. The voice broadcast receipt record is the receipt status formed after the voice broadcast terminal receives a confirmation response. The temperature broadcast record connects the job tag, temperature status, and voice response, providing input for job window occupancy and risk attribution.

[0033] S401: Based on the job identification tag, retrieve the welding torch holding area temperature rise record and voice broadcast receipt record. Match welding oscillation, temperature sampling, broadcast occupation, and receipt time according to the tag time period, and remove misaligned data to obtain the broadcast temperature field table. The broadcast temperature field table is a data object carrying temperature and voice status, recording the job identification tag, temperature sampling status, temperature increase status, broadcast occupation status, receipt time, and voice warning text. The edge processing node uses the tag time period and job category in the job identification tag as the retrieval entry point, reads the temperature sampling status and peak temperature status from the holding area temperature rise record, and reads the broadcast occupation status, broadcast text status, and receipt arrival status from the voice broadcast receipt record. Temperature sampling or voice receipts that are inconsistent with the tag time period are marked as misaligned and retained in the traceability record, and are not entered into the broadcast temperature field table. The broadcast temperature field table is output to the temperature rise mapping stage.

[0034] S402: Based on the broadcast temperature field table, determine the temperature increase record based on the difference between adjacent temperature sampling values ​​and the baseline. Compare the welding swing label with the time sequence number of the temperature increase record, associate the broadcast occupation identifier and the receipt time, and establish a swing temperature rise mapping table. The swing temperature rise mapping table is a data object connecting welding swing and holding area temperature rise, carrying the welding swing label, temperature increase status, broadcast occupation identifier, receipt time, and work category. The edge processing node converts adjacent temperature sampling states to a consistent temperature baseline and forms temperature increase records according to the direction of change of the sampling states. It does not output temperature values, nor does it use numerical results as implementation content. Subsequently, the time sequence number of the welding swing label is compared with the time sequence number of the temperature increase record to confirm whether the swing state and the temperature rise state are in the same label period. The broadcast occupation identifier is used to indicate whether the voice channel is occupied by the current warning, and the receipt time is used to express the status that the voice terminal feedback has arrived. The swing temperature rise mapping table connects swing, temperature rise, broadcast, and receipt into the same data structure.

[0035] S403: Based on the oscillation temperature rise mapping table, establish the correspondence between welding oscillation, temperature increase, voice broadcast occupation, and voice receipt according to the same label time period, mapping the job category, temperature rise status, broadcast status, and receipt status, and generating temperature broadcast records. Edge processing nodes read the job category according to the label time period in the oscillation temperature rise mapping table, bind the welding oscillation label to the temperature increase status, and bind the voice broadcast occupation status to the receipt status. The temperature rise status indicates whether the temperature change in the gripping area corresponds to the welding oscillation; the broadcast status indicates whether the voice warning text is in an occupied or waiting state; and the receipt status indicates whether the broadcast response has arrived. The voice warning text is jointly pointed to by the job category, temperature rise status, and broadcast status, pointing to a preset text set. The text set only stores the text content and trigger category, not the numerical temperature instance. The temperature broadcast record is output to the job window processing stage, retaining the message identifier, label time period, and receipt status for traceability.

[0036] Please see Figure 1 and Figure 6The specific steps of S5 are as follows: Based on temperature broadcast records, and according to welding oscillation, pause, interference, and high-temperature rejection positions, adjust the work window position registration and risk attribution relationships to generate a multi-source sensor data processing scheme. The multi-source sensor data processing scheme is the processing result object output by the edge processing node, carrying sensor acquisition rhythm adjustment, alarm weight allocation relationships, data buffer size adjustment, work window position registration, and risk attribution relationships. Work window position is the window state object where the edge processing node registers welding oscillation, pause, interference, and high-temperature rejection states within the same work window segment; welding oscillation position indicates an effective oscillation state, pause position indicates an intermittent action state, interference position indicates a state affected by secondary shaft interference or communication gaps, and high-temperature rejection position indicates a state where the holding area's temperature rise is determined to be excluded from the regular work window. This processing scheme connects window position, risk category, and temperature broadcast records into an executable acquisition and alarm adjustment relationship.

[0037] S501: Based on temperature broadcast records, match welding oscillation occupants, pause occupants, interference occupants, and high-temperature rejection occupants by window number, compare the overlap of occupant start and end times, reject occupants outside the window, and generate an occupant registration table. The occupant registration table is a data object carrying the work window status, recording the window number, occupant category, start and end time status, corresponding temperature broadcast record number, and source tag. Edge processing nodes use the work category and tag time period in the temperature broadcast records as entry points to read the registered welding oscillation occupants, pause occupants, interference occupants, and high-temperature rejection occupants. Occupant items whose start and end times do not correspond to the window number are marked as outside the window status and do not participate in risk attribution, but traceability records are retained. For occupant items with overlapping start and end times, the edge processing node records their overlap status and source category, and outputs the occupant registration table for mutual exclusion correction and risk attribution processing.

[0038] S502: Based on the occupancy registration table, window occupancy is corrected according to the mutual exclusion relationship between swing occupancy and pause occupancy. The overlapping status of interference occupancy and high temperature rejection occupancy is compared, and risk categories are associated to establish a risk attribution table. The risk attribution table is a data object that carries the relationship between window occupancy and risk categories, recording the window number, corrected occupancy category, interference impact status, high temperature rejection status, voice receipt status, and risk category. Swing occupancy and pause occupancy are set to mutual exclusion in the same window. Edge processing nodes correct window occupancy based on the continuous status of actions and the job category in the temperature broadcast record, so that effective swing and pause are not registered simultaneously in the same window segment. When interference occupancy and high temperature rejection occupancy overlap in time, the risk category is associated with secondary axis interference, communication gap, grip area temperature rise, and voice receipt status; occupancy items that do not overlap retain individual risk relationships according to the source label. The risk attribution table is output to the scheme generation stage as the basis for adjusting the acquisition rhythm, alarm weight, and buffer size.

[0039] S503: Based on the risk attribution table, map the window number, placeholder category, risk category, and temperature broadcast record sequence number, adjust the work window placeholder registration and risk attribution relationship, and generate a multi-source sensor data processing scheme. The edge processing node reads the risk attribution table, binds the window number to the placeholder category, binds the risk category to the corresponding temperature broadcast record sequence number, and adjusts the sensor acquisition rhythm, alarm weight allocation relationship, and data buffer size according to the risk source. The sensor acquisition rhythm adjustment is used to change the subsequent message acquisition arrangement when the welding swing, interference, or high temperature rejection status changes; the alarm weight allocation relationship is used to classify voice broadcast, holding area temperature rise, and communication gap into the corresponding risk category; the data buffer size adjustment is used to reserve matching buffer space for the synchronization message chain, temperature broadcast record, and risk attribution table. If the risk attribution table has missing receipts, cache status invalidation, or placeholder registration failure, the edge processing node writes the abnormal status and maintains the original window placeholder registration, and re-verifies the source message and placeholder relationship when the next round of synchronization message chain input occurs. The generated multi-source sensor data processing solution is output to the configuration channel of the workstation edge processing node and synchronously written to the traceability record, so that the operation window, temperature broadcast, communication gap and risk attribution are kept in a closed loop.

[0040] Please see Figure 7 A multi-source sensor data processing system for welding operations, comprising: Multi-source message synchronization module: This module is a message timing processing component deployed in the edge processing node. Its input interface receives welding station start messages, attitude acquisition terminal messages, grip area temperature acquisition terminal messages, voice broadcast terminal messages, and communication gateway messages. Its output interface sends a synchronization message chain to the welding judgment generation module. The module's responsibilities are limited to message parsing, timestamp cleaning, duplicate verification, timing deviation indexing, and synchronization message chain output. It does not handle welding action classification, temperature broadcast judgment, or risk attribution. Upon receiving the start message, the multi-source message synchronization module parses the node, type, and timestamp fields. For messages with missing timestamps, abnormal formats, duplicate uploads, or failed time base conversions, it writes a traceability status. Then, it forms a synchronization message chain according to the message arrival order, timing deviation magnitude, and edge clock time, and passes the calibration node timestamp, rearranged terminal message sequence, and compensated clock difference to downstream modules.

[0041] Welding Judgment Generation Module: This module processes the correspondence between action messages and temperature rise messages. Its input interface receives the synchronization message chain output by the multi-source message synchronization module, and its output interface sends the welding judgment criteria to the oscillation interference identification module. The module's responsibilities are limited to filtering action messages for flat welding, vertical welding, overhead welding, hand lifting, wrist turning, and posture changing; matching the bevel oscillation, grip area temperature rise, and voice feedback records in the temperature rise message; and generating preset welding posture deflection judgment benchmarks, bevel oscillation ranges, and temperature rise change standards. It does not handle axial interference identification or window risk attribution. Upon receiving the synchronization message chain, this module verifies the continuous correspondence between the message sequence number and the load position, constructs an action message field set from the action attributes, node identifiers, and time intervals, aligns the action category with the temperature rise message sequence number, removes fields with inconsistent time periods, and outputs the welding judgment criteria.

[0042] Swing Interference Identification Module: This module generates job status tags. Its input interface receives the welding judgment criteria from the welding judgment generation module, and its output interface sends job identification tags to the temperature voice broadcast module. The module's responsibilities are limited to retrieving records of axial rotation, secondary shaft interference, hand back tilt angle, and communication gaps; establishing a bevel action field table and a job correspondence table; and generating abnormal rotation categories, axial interference types, and communication disconnection cycles. It does not handle the judgment of temperature increases in the gripping area or the generation of voice broadcast records. After the welding judgment criteria are received, this module reads the swing direction, wrist flip indicator, posture change indicator, and message gap position according to the time interval in the criteria. It removes time-period misalignment fields, compares the swing action characteristics with the correspondence between axial rotation, hand back tilt angle changes, and secondary shaft interference, and outputs the job identification tag.

[0043] Temperature Voice Broadcast Module: This module connects the job identification tag, the temperature rise of the gripping area, and the voice feedback. Its input interface receives the job identification tag output by the oscillation interference identification module, along with the gripping area temperature rise record and the voice broadcast feedback record. Its output interface sends the temperature broadcast record to the sensor data processing module. The module's responsibilities are limited to matching welding oscillation, temperature sampling, broadcast occupancy, and feedback times according to the tag's time period, establishing an oscillation temperature rise mapping table, and outputting the peak temperature status of the gripping area, the voice warning text, and the feedback response interval. It does not handle job window occupancy correction. After the job identification tag enters, this module removes misaligned time period data, records the increasing temperature of adjacent temperature sampling states under the same benchmark, and then associates the broadcast occupancy identifier and feedback time to form the temperature broadcast record.

[0044] Sensor Data Processing Module: This module handles job window placement and risk attribution. Its input interface receives temperature broadcast records from the temperature voice broadcast module, and its output interface sends multi-source sensor data processing solutions to the workstation edge configuration channel and traceability record channel. This module's responsibilities are limited to matching welding oscillation placements, pause placements, interference placements, and high-temperature rejection placements; establishing placement registration and risk attribution tables; and adjusting job window placement registration, risk attribution relationships, sensor acquisition rhythm, alarm weight allocation, and data buffer size. It does not handle front-end message parsing or action determination. Upon receiving temperature broadcast records, this module filters out placement items outside the window by window number, corrects placements based on the mutual exclusion relationship between oscillation and pause placements, associates risk categories based on the overlapping time sequences of interference and high-temperature rejection placements, and writes abnormal statuses when there are missing receipts, cache status failures, or placement registration failures, before outputting the multi-source sensor data processing solution.

[0045] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the technical solution.

Claims

1. A method for processing multi-source sensor data in welding operations, characterized in that, Includes the following steps: S1: Obtain the welding station start message, collect the welding station multi-source terminal messages, calculate the timing deviation between the sampling timestamp and the edge clock, merge the message order according to the timing deviation, and generate a synchronization message chain; S2: Based on the synchronous message chain, filter the bevel oscillation, grip area heating and voice reply records in the multi-posture action message and heating message to generate the welding judgment caliber; S3: Based on the welding determination caliber, call the records of bevel oscillation, axial flipping, secondary shaft interference, back of hand tilt angle and communication gap, compare the relationship between oscillation and interference gap, and generate a job identification label; S4: Based on the operation identification tag, call the holding area temperature rise record and voice reply record, determine the relationship between welding swing tag, temperature rise, voice occupation and voice reply, and generate temperature broadcast record; S5: Based on the temperature broadcast records, adjust the position order according to the multiple positions in the work window, and the relationship between voice assignment and high temperature assignment to generate a multi-source sensor data processing scheme.

2. The multi-source sensor data processing method for welding operations according to claim 1, characterized in that, The synchronization message chain includes calibrated node timestamps, reconstructed terminal message sequences, and compensated clock differences. The welding judgment criteria include a preset welding posture deflection threshold, a quantified bevel swing range, and a defined heating rate standard. The operation identification tag includes a matched abnormal flip category, a classified axial interference type, and a marked communication disconnection period. The temperature broadcast record includes the recorded peak temperature of the grip area, the output voice warning text, and the feedback broadcast response delay. The multi-source sensor data processing scheme includes adjusted sensor acquisition frequency, allocated system alarm weights, and optimized data cache size.

3. The multi-source sensor data processing method for welding operations according to claim 1, characterized in that, The timing deviation merged message order refers to the synchronization message order formed after time calibration, sorting and merging of multi-source terminal messages based on the timing deviation between the sampling timestamps of multiple messages and the edge clock.

4. The multi-source sensor data processing method for welding operations according to claim 1, characterized in that, The relationship between the swing and the interference gap refers to the correlation formed by comparing the axial flipping and back-of-hand tilting motion characteristics in the bevel swing with the secondary shaft interference and communication gap abnormal characteristics.

5. The multi-source sensor data processing method for welding operations according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Obtain welding station start message, collect and parse multi-source terminal messages, obtain node, type and time stamp field, remove missing time stamp messages, check duplicate messages by node number and sampled time stamp, and obtain message time stamp field set; S102: Based on the message timestamp field set, call the node sampling timestamp and the edge clock time, calculate the time difference in the same time dimension, mark the sampling ahead state and the sampling lag state, associate the message identifier field with the timing deviation amplitude, and establish a timing deviation index table. S103: According to the timing deviation index table, merge the messages according to the message arrival sequence number and the difference magnitude. For messages with the same difference magnitude, determine the parallel order according to the edge clock time, map the node number and the load position, and generate a synchronization message chain.

6. The multi-source sensor data processing method for welding operations according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the synchronization message chain, filter multiple types of welding action messages according to message type identifier, verify the continuous correspondence between message time sequence number and load position, associate action attributes, node identifier and time interval, and generate action message field set; S202: Based on the action message field set, match the records of bevel oscillation, grip area heating, and voice reply in the heating message, compare the action category with the heating message sequence number, remove fields with inconsistent time periods, and establish a welding record mapping table. S203: Based on the welding record mapping table, determine the consistency relationship between the action type, bevel swing amplitude, holding area temperature rise amplitude, and voice reply status, associate the judgment identifier with the message node number, and generate the welding judgment caliber.

7. The multi-source sensor data processing method for welding operations according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: Based on the welding judgment caliber, call the records of axial flip, secondary shaft interference, back of hand tilt angle, and communication gap, match the swing direction, wrist flip mark, posture change mark, and message gap position according to the time interval, remove the time interval misalignment field, and generate the bevel action field table. S302: Based on the bevel action field table, compare the correspondence between the swing action characteristics and axial flip, back of hand tilt angle change, and secondary shaft interference, associate the message gap position with the communication gap record, and establish an operation correspondence table; S303: Based on the job correspondence table, determine the caliber number, swing direction, wrist flip mark, posture change mark, message gap position and job category mark according to the same time period interval mapping, and generate job identification label.

8. The multi-source sensor data processing method for welding operations according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the job identification tag, call the heating record of the welding torch holding area and the voice broadcast reply record, match the welding swing, temperature sampling, broadcast occupation and reply time according to the tag time period, and remove misaligned data to obtain the broadcast temperature field table. S402: Based on the broadcast temperature field table, determine the temperature increase record based on the difference of adjacent temperature sampling values ​​with the same dimension, compare the welding swing label with the time sequence number of the temperature increase record, associate the broadcast occupation identifier and the receipt time, and establish a swing temperature rise mapping table. S403: Based on the swing temperature rise mapping table, establish the correspondence between welding swing, temperature increase, voice broadcast occupation and voice receipt according to the same label time period, map the operation category, temperature rise status, broadcast status and receipt status, and generate temperature broadcast records.

9. The multi-source sensor data processing method for welding operations according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Based on the temperature broadcast record, match welding swing occupants, pause occupants, interference occupants and high temperature rejection occupants according to the window number, compare the overlapping state of the start and end times of the occupants, reject the occupant items outside the window, and generate an occupant registration table. S502: Based on the occupancy registration table, correct the window occupancy according to the mutual exclusion relationship between swinging occupancy and paused occupancy, compare the overlapping state of interference occupancy and high temperature rejection occupancy, associate risk categories, and establish a risk attribution table. S503: Based on the risk attribution table, map window number, placeholder category, risk category and temperature broadcast record sequence number, adjust the placeholder registration and risk attribution relationship of the operation window, and generate a multi-source sensor data processing scheme.

10. A multi-source sensor data processing system for welding operations, characterized in that, The system is used to implement the welding operation multi-source sensor data processing method according to any one of claims 1-9, including: The multi-source message synchronization module is used to implement S1: acquiring the welding station start message, collecting multi-source terminal messages of the welding station, calculating the timing deviation between the sampling timestamp and the edge clock, merging the message order according to the timing deviation, and generating a synchronization message chain; The welding judgment generation module is used to implement S2: based on the synchronous message chain, filter the bevel swing, grip area heating and voice reply records in the multi-posture action message and heating message to generate the welding judgment caliber; The swing interference identification module is used to implement S3: based on the welding judgment caliber, call the axial flip, secondary shaft interference, hand back tilt angle and communication gap record in the bevel swing record, compare the relationship between swing and interference gap, and generate a job identification label; The temperature voice broadcast module is used to implement S4: based on the operation identification tag, call the holding area temperature rise record and voice reply record, determine the relationship between welding swing tag, temperature rise, voice occupation and voice reply, and generate temperature broadcast record; The sensor data processing module is used to implement S5: based on the temperature broadcast record, according to the multiple occupants in the work window, adjust the occupant order, and the relationship between voice attribution and high temperature attribution, generate a multi-source sensor data processing scheme.

Images