A multi-sensor fusion welding temperature monitoring method and related device

Abstract

The application provides a multi-sensor fusion welding temperature monitoring method and related device, which comprises the following steps: dividing a welding area into a plurality of sub-areas according to the spatial distribution of the welding area, and obtaining initial temperature data corresponding to each sub-area according to a plurality of temperature sensors; time aligning the initial temperature data through difference resampling of different sampling points to generate a time series temperature matrix; performing real-time quality detection on the time series temperature matrix to generate a real-time quality score, and weighting and combining the real-time quality score with a pre-set confidence parameter table to obtain a dynamic confidence result set; fusing effective temperature data in the dynamic confidence result set that exceeds a pre-set confidence threshold according to a fusion rule corresponding to each sub-area to obtain a fusion temperature result; and time series splicing the fusion temperature results of each sub-area to obtain a global temperature curve of the welding process, which can effectively improve the accuracy of temperature detection in the welding process.

Description

Technical Field

[0001] This application relates to the field of welding technology, and in particular to a welding temperature monitoring method and related apparatus using multi-sensor fusion. Background Technology

[0002] In the welding process, the temperature field distribution directly affects the microstructure and mechanical properties of the joint. Taking the automated welding of large pressure vessels as an example, there is often a significant spatial temperature gradient in the welding area. The temperature in the molten pool is extremely high, the temperature in the heat-affected zone changes rapidly, while the base material area is at a relatively low temperature. To achieve precise monitoring of the welding process, various types of temperature sensors, such as thermocouples, infrared sensors, and fiber optic sensors, are typically placed in the welding area.

[0003] However, due to factors such as sensor installation location, sampling frequency differences, and environmental interference, data acquired by different sensors within the same time period suffers from inconsistent timestamps, fluctuating readings, and significant deviations. Current technologies for processing multi-sensor data often rely on single-sensor or simple averaging methods. This simplistic approach cannot effectively handle complex, non-linear interference. When a sensor fails or is severely affected by interference, it directly leads to erroneous measurement results. Summary of the Invention

[0004] This application provides a welding temperature monitoring method and related apparatus based on multi-sensor fusion, which solves the problem that related technologies cannot obtain accurate temperature measurement data when processing data from multiple temperature sensors.

[0005] The first aspect of this application provides a welding temperature monitoring method based on multi-sensor fusion, the welding temperature monitoring method based on multi-sensor fusion includes: The welding area is divided into several sub-areas based on its spatial distribution, and the initial temperature data of each sub-area is obtained using multiple types of temperature sensors. The initial temperature data is time-aligned by resampling the difference between different sampling points to generate a time-series temperature matrix. The time-series temperature matrix is ​​subjected to real-time quality detection to generate a real-time quality score. The real-time quality score is then weighted and combined with a pre-set confidence parameter table to obtain a dynamic confidence result set. According to the fusion rules corresponding to each sub-region, the valid temperature data exceeding the preset confidence threshold in the dynamic confidence result set are fused to obtain the fused temperature result; The fusion temperature results of each sub-region are spliced ​​together over time to obtain the global temperature curve of the welding process.

[0006] Optionally, in the first implementation of the first aspect of this application, the step of dividing the welding area into several sub-regions according to the spatial distribution of the welding area, and obtaining the initial temperature data corresponding to each sub-region according to multiple types of temperature sensors, includes: Obtain the temperature gradient distribution of the welding area corresponding to the welding type during the welding process; By mapping the temperature gradient distribution to the spatial distribution of the welding area, the welding area is divided into three sub-regions: a high-temperature region, a heat-affected region, and a substrate region, and a region division table is generated. The raw temperature data corresponding to each sub-region is obtained by multiple types of temperature sensors, and the raw temperature data is labeled according to the region division table to generate initial temperature data.

[0007] Optionally, in the second implementation of the first aspect of this application, the step of performing time alignment on the initial temperature data through differential resampling of different sampling points to generate a time-series temperature matrix includes: Based on the sampling frequency and timestamp of multiple types of temperature sensors, determine the time difference sequence between sampling points of different sensors; By performing minimum time interval analysis on the time difference sequence, a unified target sampling step size is generated, and a reference sampling time axis is constructed based on the target sampling step size; Based on the reference sampling time axis, interpolation is performed on the initial temperature data of multiple types of temperature sensors to generate supplementary data points at missing time points, resulting in a complete multi-channel temperature sequence. By performing time index alignment on the completed multi-channel temperature sequence, the temperature readings of different temperature sensors at the same time are merged into a single record, resulting in a time-series temperature matrix.

[0008] Optionally, in the third implementation of the first aspect of this application, the step of performing real-time quality detection on the time-series temperature matrix, generating a real-time quality score, and weighting and combining the real-time quality score with a pre-set confidence parameter table to obtain a dynamic confidence result set includes: Based on the continuous temperature readings of each temperature sensor in the time-series temperature matrix, the temperature fluctuation index is calculated, and a fluctuation score matrix is ​​generated by comparing the fluctuation index with a preset threshold. Based on the temperature change sequence of each temperature sensor at adjacent time points in the time-series temperature matrix, the data continuity status is determined, and a continuity score matrix is ​​generated by statistical analysis of the continuity status. Based on the temperature differences of different temperature sensors in the same region of the time-series temperature matrix, the relative deviation of the temperature values ​​is compared with the average temperature in the region to generate a difference scoring matrix. By weighting and combining the volatility score matrix, the continuity score matrix, and the difference score matrix, and combining them with the baseline confidence of the corresponding temperature sensor in the preset confidence parameter table, a dynamic confidence result set for each sub-region is generated.

[0009] Optionally, in the fourth implementation of the first aspect of this application, the step of fusing the valid temperature data exceeding a preset confidence threshold in the dynamic confidence result set according to the fusion rules corresponding to each sub-region to obtain the fused temperature result includes: Sensor data exceeding a preset confidence threshold are filtered from the dynamic confidence result set to obtain effective temperature data for each sub-region; Based on the fusion rules corresponding to each sub-region, the weight allocation method of the effective temperature data is determined, and a preliminary fusion temperature result is generated by weighting the effective temperature data. The preliminary fusion temperature result is compared with the outlier determination criteria, the abnormal data in the preliminary fusion temperature result is removed, and the weighted fusion value is recalculated to generate the corrected fusion temperature result. The final fusion temperature result is determined by matching the corrected fusion temperature result with the historical fusion temperature sequence.

[0010] Optionally, in the fifth implementation of the first aspect of this application, before the step of comparing the preliminary fusion temperature result with the outlier determination criteria, removing outlier data from the preliminary fusion temperature result, and recalculating the weighted fusion value to generate the corrected fusion temperature result, the method further includes: Based on the continuous time series of the preliminary fused temperature results, the temperature change rate between adjacent time points is extracted, and a change rate determination matrix is ​​generated by comparing the temperature change rate with a preset change threshold. Based on the difference between the preliminary fusion temperature result and the historical fusion temperature sequence in the corresponding sub-region, a deviation index is calculated, and a deviation judgment matrix is ​​generated by comparing the deviation index with a preset deviation threshold. By jointly analyzing the rate of change determination matrix and the deviation determination matrix, the set of abnormal data points in the preliminary fused temperature results is determined; Based on the set of abnormal data points, the corresponding data is removed, and the remaining preliminary fusion temperature results are weighted to generate the corrected fusion temperature results.

[0011] Optionally, in the sixth implementation of the first aspect of this application, the step of stitching together the fusion temperature results of each sub-region over time to obtain the global temperature curve of the welding process includes: Based on the fusion temperature results and the corresponding timestamp information, a fusion temperature sequence for the sub-region is constructed, and the fusion temperature sequence is divided into time ranges to generate a sub-region time index table. By sorting and comparing the time index table of the sub-regions, the time boundary points between adjacent sub-regions are determined, and interpolation is performed at the boundary points to generate a transition temperature sequence. Based on the fused temperature sequence and the transition temperature sequence, the temperature data of each time segment are spliced ​​together to obtain a continuous global spliced ​​temperature sequence. By smoothing the global splicing temperature sequence, a global temperature curve for the welding process is generated.

[0012] A second aspect of this application provides a multi-sensor fusion welding temperature monitoring device, which is used to implement a multi-sensor fusion welding temperature monitoring method. The multi-sensor fusion welding temperature monitoring device includes: The acquisition module is used to divide the welding area into several sub-regions according to the spatial distribution of the welding area, and to acquire the initial temperature data corresponding to each sub-region according to multiple types of temperature sensors. The generation module is used to perform time alignment on the initial temperature data by resampling the difference between different sampling points to generate a time-series temperature matrix; The detection module is used to perform real-time quality detection on the time-series temperature matrix, generate a real-time quality score, and weight and combine the real-time quality score with a pre-set confidence parameter table to obtain a dynamic confidence result set. The fusion module is used to fuse the valid temperature data exceeding the preset confidence threshold in the dynamic confidence result set according to the fusion rules corresponding to each sub-region, so as to obtain the fused temperature result; The splicing module is used to splice the fusion temperature results of each sub-region over time to obtain the global temperature curve of the welding process.

[0013] A third aspect of this application provides an electronic device, including a memory and a processor, wherein the processor is configured to execute a computer program stored in the memory, and when the processor executes the computer program, it implements the steps of the multi-sensor fusion welding temperature monitoring method provided in the first aspect of this application.

[0014] The fourth aspect of this application provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the steps of the multi-sensor fusion welding temperature monitoring method provided in the first aspect of this application.

[0015] In summary, the welding temperature monitoring method and related apparatus based on multi-sensor fusion provided in this application establish a multi-sub-region temperature monitoring model within the welding area. Data from multiple types of temperature sensors are introduced into each sub-region, and a time-series temperature matrix is ​​generated by combining differential resampling. Then, the sensor data is dynamically weighted based on real-time quality detection and a pre-set confidence parameter table. This allows for the acquisition of stable fused temperature results for each sub-region under fusion rules, which are then stitched together to form a complete global temperature curve. This application can obtain a global temperature curve that accurately reflects the temperature evolution relationship in different regions, even when sensor data exhibits asynchronous sampling, fluctuations, and regional biases, effectively improving the accuracy of temperature detection during the welding process. Attached Figure Description

[0016] Figure 1 A schematic flowchart illustrating the multi-sensor fusion welding temperature monitoring method provided in this application embodiment; Figure 2 A schematic diagram of the program modules of the welding temperature monitoring device with multi-sensor fusion provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0017] To make the inventive objectives, features, and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] To address the problem that related technologies cannot obtain accurate temperature measurement data when processing data from multiple temperature sensors, embodiments of this application provide a welding temperature monitoring method that integrates multiple sensors, such as... Figure 1 This is a flowchart illustrating the multi-sensor fusion welding temperature monitoring method provided in this embodiment. The multi-sensor fusion welding temperature monitoring method includes the following steps: Step 110: Divide the welding area into several sub-areas according to the spatial distribution of the welding area, and obtain the initial temperature data corresponding to each sub-area according to multiple types of temperature sensors.

[0019] Specifically, during the welding process, the spatial distribution characteristics of the welding area are rationally divided to enable targeted monitoring of the thermal field evolution in different regions. Because the temperature gradients in the molten pool, heat-affected zone, and base material zone differ significantly, directly using temperature data from a single region cannot reflect the overall characteristics. Therefore, it is necessary to divide the welding area into multiple sub-regions based on process characteristics and deploy various types of temperature sensors within each sub-region. Through the combined acquisition of thermocouples, infrared sensors, and fiber optic sensors, temperature data from different physical mechanisms can be obtained within the same spatial range.

[0020] In one optional embodiment of this example, the steps of dividing the welding area into several sub-regions according to the spatial distribution of the welding area and obtaining the initial temperature data corresponding to each sub-region according to multiple types of temperature sensors include: obtaining the temperature gradient distribution of the welding area corresponding to the welding type during the welding process; dividing the welding area into three sub-regions—a high-temperature zone, a heat-affected zone, and a substrate zone—by mapping the temperature gradient distribution to the spatial distribution of the welding area, and generating a region division table; obtaining the original temperature data corresponding to each sub-region according to multiple types of temperature sensors, and labeling the original temperature data according to the region division table to generate initial temperature data.

[0021] In this embodiment, by acquiring the thermal input parameters and process boundary conditions of the welding type, and combining them with historical monitoring data or numerical simulation results, a spatiotemporal distribution map of the temperature field during the welding process is generated. The temperature gradient distribution refers to the rate of temperature change along the spatial coordinates on the welding cross section, i.e., the derivative of the temperature field with respect to the spatial coordinates (e.g., ∂T / ∂x, ∂T / ∂y). This distribution characterizes the strength of heat transfer from the weld zone outwards. Subsequently, the temperature gradient distribution is mapped onto the welding geometric coordinate system. A unified reference frame is established between the coordinates of the simulation mesh or historical measurement points and the on-site geometric model and sensor installation coordinates through coordinate registration. Coordinate registration can be completed based on reference point positioning or CAD model alignment, and the evolution of the gradient isosurface is recorded in the time dimension to capture the migration of isotherms over time. Next, the mapped temperature field is divided into regions using a combination of isotherm extraction and threshold segmentation. The threshold can be set based on the phase transition temperature of the material, the metallurgical critical temperature, or the peak temperature quantile obtained from simulation and experimental statistics. This divides the spatial domain into a high-temperature region, a heat-affected region, and a matrix region. The boundaries of the division are represented in the form of polygons or grid cells, and the boundary coordinates and corresponding isothermal values ​​are recorded over time, thereby generating a region division table containing region identifiers, boundary coordinates, isothermal thresholds, and time window information. Subsequently, the raw readings of various types of temperature sensors are spatially labeled according to the region division table. The labeling process is achieved by projecting the sensor measuring points or sensor fields of view onto the region polygons in the reference coordinate system. For point sensors, the region label is directly matched with the measuring point coordinates, while for area sensors or sensors with fields of view (such as the detection field of view of an infrared thermometer or the equivalent projection of distributed optical fibers in space), the region is determined by the center of the field of view or the weighted projection position. Then, the sensor identifier, measurement position, timestamp, and corresponding region label are combined and formatted to form an initial temperature data table containing spatial labeling information.

[0022] Step 120: Time-align the initial temperature data by resampling the difference between different sampling points to generate a time-series temperature matrix.

[0023] Specifically, since thermocouples often output at low frequencies, while fiber optic and infrared sensors may acquire data at the millisecond level, direct comparison can lead to time axis misalignment. Therefore, by resampling the differences between different sampling points, a unified time reference can be established, and the temperature readings of each sensor at the same moment can be combined into a unified record, thereby generating a time-series temperature matrix.

[0024] In one optional implementation of this embodiment, the step of time-aligning the initial temperature data by resampling the difference between different sampling points to generate a time-series temperature matrix includes: determining the time difference sequence between sampling points of different sensors based on the sampling frequency and timestamp of multiple types of temperature sensors; generating a unified target sampling step size by performing minimum time interval analysis on the time difference sequence, and constructing a reference sampling time axis based on the target sampling step size; performing interpolation on the initial temperature data of multiple types of temperature sensors based on the reference sampling time axis, generating supplementary data points at missing time points to obtain a completed multi-channel temperature sequence; and merging the temperature readings of different temperature sensors at the same time into a single record by performing time index alignment on the completed multi-channel temperature sequence to obtain a time-series temperature matrix.

[0025] In this embodiment, based on the sampling frequency and timestamps of multiple types of temperature sensors, the original sampling time points of each sensor are first extracted and serialized in chronological order. Then, the time difference is calculated for any adjacent sampling points to obtain a time difference value sequence. The sampling frequency refers to the number of samples per unit time, while the timestamp is used to accurately locate the moment of a single sampling. The time difference value sequence reflects sampling non-uniformity and clock jitter. Subsequently, minimum time interval analysis is performed on the time difference value sequence. The so-called minimum time interval analysis can be performed by converting the timestamps to integer microseconds or milliseconds and calculating the least common divisor of the sequence, or by selecting the lower percentile and introducing a clock jitter margin, thereby generating a unified target sampling step size. The target sampling step size should balance data accuracy and computational cost. For example, in the mixed case of thermocouple 10Hz, infrared 100Hz, and fiber optic 1000Hz, 1ms or 10ms can be selected as the target step size to take into account transient information and interpolation stability. Then, a reference sampling time axis is constructed based on the target sampling step size. This reference axis is aligned with a unified starting point and generates time points at equal intervals according to the target step size to avoid cumulative drift. Allowable time offset windows can be recorded during the axis generation process to facilitate interpolation error tolerance. Next, interpolation is performed on the initial temperature data based on the reference sampling time axis to fill in the missing time points. The interpolation method can be selected according to the sensor response characteristics, such as linear interpolation, cubic spline interpolation, or smooth interpolation based on Kalman filtering. During the interpolation process, quantization error, sample-and-hold effect, and anti-aliasing filter (a low-pass filter that removes high frequencies before sampling) need to be considered to reduce high-frequency artifacts. For example, linear or zero-order hold is used to fit the instantaneous readings for slow-responding thermocouples, while spline is used for high-frequency fiber optic signals to retain rapid changes. Subsequently, the completed multi-channel temperature sequence is aligned with the time index according to the reference time axis. That is, the interpolation results of each channel are extracted at each reference time and the multi-source readings at the same time are merged into a single record. Finally, a time-series temperature matrix is ​​formed with the row index being time and the column index being the sensor or sensing channel. Confidence labels are added to the points obtained by large-interval interpolation in the matrix for subsequent quality assessment and weighted fusion, thus completing the conversion from asynchronous sampling to a unified time-series representation.

[0026] Step 130: Perform real-time quality detection on the time-series temperature matrix, generate a real-time quality score, and weight and combine the real-time quality score with a pre-set confidence parameter table to obtain a dynamic confidence result set.

[0027] Specifically, after obtaining the time-series temperature matrix, dynamic monitoring of data quality is required. Due to arc interference, environmental radiation, and localized sensor damage during welding, the raw data may exhibit abnormal fluctuations, discontinuities, and numerical deviations. To address this, the volatility of continuous readings needs to be analyzed, identifying abnormal fluctuations by calculating the temperature change range between adjacent time points; the intervals between adjacent readings also need to be statistically analyzed to determine if data interruptions exist; furthermore, data collected by different sensors in the same area can be compared horizontally, identifying values ​​with significant deviations by comparing them to the average level within the area. These detection results are then converted into a scoring matrix and combined with a pre-defined confidence parameter table to form a dynamic confidence result set.

[0028] In one optional implementation of this embodiment, the steps of performing real-time quality detection on the time-series temperature matrix, generating a real-time quality score, and weighting and combining the real-time quality score with a preset confidence parameter table to obtain a dynamic confidence result set include: calculating a temperature fluctuation index based on the continuous temperature readings of each temperature sensor in the time-series temperature matrix, and generating a fluctuation score matrix by comparing the fluctuation index with a preset threshold; determining the data continuity status based on the temperature change sequence of each temperature sensor at adjacent time points in the time-series temperature matrix, and generating a continuity score matrix by statistically analyzing the continuity status; generating a difference score matrix by comparing the relative deviation of the temperature values ​​of different temperature sensors in the same area of ​​the time-series temperature matrix with the average temperature in the area; and generating a dynamic confidence result set for each sub-region by weighting and combining the fluctuation score matrix, the continuity score matrix, and the difference score matrix, and combining them with the benchmark confidence of the corresponding temperature sensor in the preset confidence parameter table.

[0029] In this embodiment, the first step in the time-series temperature matrix is ​​to calculate a fluctuation index based on the continuous temperature readings of each sensor. This fluctuation index can be measured using methods such as sliding window standard deviation, root mean square deviation, or first-order difference variance. Its purpose is to assess the stability of a sensor's temperature signal over a short period. If high-frequency jitter or random noise exists in the signal, the fluctuation index value will increase. Subsequently, the obtained fluctuation index is compared point-by-point with a pre-set threshold. The threshold can be selected by referring to the sensor calibration error range or by considering the physical rate of change of the thermal field during welding. This converts the stability of each sensor at different time periods into a score, and a fluctuation score matrix is ​​constructed. This matrix stores the stability score corresponding to each time point, with time as the row and sensor number as the column. Next, the data continuity status needs to be determined by analyzing the temperature change sequence at adjacent time points. The continuity status refers to whether the sensor outputs valid data according to the reference sampling time axis at the expected time point. If there are missing points, breakpoints, or abrupt changes, the continuity will be disrupted. By statistically analyzing the distribution of time intervals and the proportion of missing points, the continuity of each sensor can be quantified, thus generating a continuity scoring matrix. This matrix, also indexed by time and sensor, reflects whether the signal output is stable. Next, the temperature differences between different temperature sensors within the same region need to be calculated. This is done by using the region's average temperature as a reference and calculating the relative deviation between each sensor's temperature value and this average value point by point. If the deviation is too large, it indicates that the sensor has abnormal drift or calibration error. This method generates a difference scoring matrix, where matrix elements represent the reliability of a sensor relative to the overall temperature of the region at a given time. Finally, the fluctuation scoring matrix, continuity scoring matrix, and difference scoring matrix are weighted and combined. The weighting coefficients can be set based on prior knowledge of the importance of different indicators; for example, more emphasis is placed on difference in high-temperature gradient regions, and more emphasis on continuity in stable regions. Then, this is combined with the baseline confidence level in a pre-set confidence parameter table, which is derived from long-term experimental data or comparisons with standard thermometers, to obtain a dynamic confidence result set. This result set not only reflects the instantaneous reliability of each sensor in the time dimension, but also incorporates sub-region division in the spatial dimension, allowing sensors in different regions to dynamically adjust their weights in subsequent fusion. For example, if a fiber optic sensor at the weld pool boundary has a low volatility score but high continuity and difference scores, its dynamic confidence value will remain within a reasonable range after calculation, and it will not be directly eliminated due to a single abnormal indicator. This dynamically generated confidence result set provides a quantitative basis for subsequent multi-sensor weighted fusion and can reflect data quality fluctuations in real time during the welding process.

[0030] Step 140: According to the fusion rules corresponding to each sub-region, the effective temperature data exceeding the preset confidence threshold in the dynamic confidence result set are fused to obtain the fused temperature result.

[0031] Specifically, after obtaining the dynamic confidence score result set, the data needs to be processed according to the preset regional fusion rules. Since the reliability of different sensors in the same area is not consistent, it is necessary to filter the data by a confidence score threshold. For sensors with a confidence score below the threshold, their data will be removed to avoid interfering with the final result. After the valid data is selected, the corresponding weights are calculated based on the dynamic confidence score to complete the weighted fusion. This fusion method can ensure that sensors with high confidence scores occupy a larger proportion in the results, thereby generating the fused temperature values ​​of the sub-regions.

[0032] In one optional implementation of this embodiment, the step of fusing valid temperature data exceeding a preset confidence threshold in the dynamic confidence result set according to the fusion rules corresponding to each sub-region to obtain a fused temperature result includes: filtering sensor data exceeding the preset confidence threshold in the dynamic confidence result set to obtain valid temperature data for each sub-region; determining the weight allocation method of the valid temperature data according to the fusion rules corresponding to each sub-region, and generating a preliminary fused temperature result by weighting the valid temperature data; comparing the preliminary fused temperature result with the outlier judgment conditions, removing outlier data from the preliminary fused temperature result, and recalculating the weighted fusion value to generate a corrected fused temperature result; and determining the final fused temperature result by matching and analyzing the corrected fused temperature result with historical fused temperature sequences.

[0033] In this embodiment, after the dynamic confidence result set has been generated, the data needs to be filtered according to a preset confidence threshold. The confidence threshold is used to distinguish between high and low reliability. If the confidence result of a sensor at a specific time is lower than the threshold, the data at that time is considered to have significant uncertainty and will not participate in subsequent fusion. Only the portion exceeding the threshold is retained, ultimately forming the effective temperature data set for each sub-region. This filtering method ensures that the data input to the fusion stage has undergone confidence quality constraints and will not introduce noise or distorted readings into the results. Next, the weighting method for the effective temperature data needs to be determined according to the fusion rules corresponding to each sub-region. The fusion rules can reflect the differences in physical characteristics and sensor configurations of different sub-regions. For example, infrared sensors are given greater weight in the high-temperature region, while embedded thermocouples are given greater weight in the substrate region. The determination of the weighting method depends not only on the sensor type but can also be coupled with the confidence result, so that data with higher confidence accounts for a larger proportion in the weighted calculation. By weighting the effective temperature data, a preliminary fusion temperature result can be generated. Mathematically, this result represents a weighted average or weighted fit of multi-source data; physically, it represents a comprehensive estimate of the temperature state of the same region. Next, the preliminary fusion temperature result needs to be compared with outlier criteria. These criteria can be set based on physical constraints or statistical characteristics, such as limiting the fusion temperature result at a certain moment to not exceeding the range predicted by the regional heat conduction model, or preventing jumps exceeding a specific threshold between adjacent moments. If outlier data points are detected, they need to be removed, and the weighting operation is re-executed to generate a corrected fusion temperature result. This correction process avoids significant deviations in the overall fusion curve due to local data distortion. Finally, the corrected fusion temperature result needs to be matched with historical fusion temperature sequences. This matching analysis calculates the degree of agreement between the new result and the existing historical sequence in terms of trend and value, confirming whether the corrected result maintains consistency with the continuity of the thermal field evolution during welding. For example, if the result at a certain moment deviates too much from the historical trend, further adjustments to the calculation parameters may be necessary. This process allows for the final determination of the fused temperature results for each sub-region, which are then stably input into the global curve stitching stage. For example, in a high-temperature region, if both infrared sensor and thermocouple data are selected as valid data, but at a certain moment the infrared sensor experiences an abnormal spike due to surface reflection, the initial fusion result may show an abnormal peak. By comparing this with the outlier criteria, this point is discarded, and a smoother and more reasonable correction result is obtained through reweighting. After comparing this result with historical curves to confirm that there is no significant deviation, it is finally used as the fused temperature output for that sub-region. This processing ensures that the temperature estimation for different regions is both real-time and maintains physical rationality in the overall trend.

[0034] In an optional implementation of this embodiment, before the steps of comparing the preliminary fusion temperature result with the outlier determination criteria, removing outlier data from the preliminary fusion temperature result, and recalculating the weighted fusion value to generate the corrected fusion temperature result, the method further includes: extracting the temperature change rate between adjacent time points based on the continuous time series of the preliminary fusion temperature result, and generating a change rate determination matrix by comparing the temperature change rate with a preset change threshold; calculating a deviation index based on the difference between the preliminary fusion temperature result and the historical fusion temperature series in the corresponding sub-region, and generating a deviation determination matrix by comparing the deviation index with a preset deviation threshold; determining the set of outlier data points in the preliminary fusion temperature result through joint analysis of the change rate determination matrix and the deviation determination matrix; removing corresponding data based on the set of outlier data points, and performing a weighted calculation on the remaining preliminary fusion temperature results to generate the corrected fusion temperature result.

[0035] In this embodiment, based on the continuous time series of the preliminary fusion temperature results, the rate of temperature change between adjacent time points is extracted. The rate of change refers to the amplitude of temperature rise or fall per unit time, which can be calculated from the difference between adjacent time points. To reduce instantaneous noise interference, a sliding window mid-range filter or low-pass filter can be applied to the rate of change signal to smooth transient fluctuations and obtain a stable instantaneous change trend. Subsequently, the rate of change of each sensor at each time point is compared with a preset change threshold. The threshold is set based on the thermal response characteristics of the material and the heat input of the welding process. The comparison results can generate a rate of change judgment matrix. This matrix records the intensity level of change of each sensor at each time point, reflecting whether the temperature fluctuation exceeds a reasonable range. At the same time, it is necessary to calculate the deviation index based on the difference between the preliminary fusion temperature results and the historical fusion temperature series in the corresponding sub-region. The deviation index is used to quantify the degree of deviation of the current temperature reading from the historical temperature evolution. It can be obtained by comparing the current value with the historical average or the temperature change trend within the historical time period, and combined with the fluctuation range of the historical series to assess the degree of anomaly. By comparing the deviation with the preset deviation threshold, a deviation judgment matrix can be formed. Each element in the matrix indicates whether the reading of a certain sensor at a certain time exceeds the expected range relative to historical data. Next, the rate of change judgment matrix and the deviation judgment matrix are jointly analyzed to identify the set of outlier data points in the preliminary fused temperature results. The joint analysis considers not only simultaneous anomalies in both rate of change and deviation but also incorporates temporal continuity rules. For example, if an anomalous signal appears in several consecutive sampling points, it is more likely to be identified as a genuine anomaly rather than transient noise. By marking points in the matrices that meet the joint judgment conditions, the final set of outlier data points is formed. Based on the identified set of outlier data points, the corresponding preliminary fused temperature readings are removed, and the remaining valid data are re-weighted to generate a corrected fused temperature result. The weight allocation references the dynamic confidence result set, and the weights are normalized after removing outliers to ensure the stability and continuity of the fused result. For data gaps caused by removal, interpolation of adjacent valid points or short-term predictions from historical sequences can be used to supplement the data, ensuring the integrity of the corrected fused temperature curve in terms of continuity and physical rationality. For example, in the boundary region of the molten pool, if the fiber optic sensor shows a sudden and significant increase that deviates from the historical temperature trend, that moment will be identified as an anomaly and removed. The corrected temperature curve obtained after reweighting and interpolation with nearby time points is both smooth and continuous, maintaining the authenticity of the temperature change characteristics of the sub-region.

[0036] Step 150: Perform time-series splicing of the fusion temperature results of each sub-region to obtain the global temperature curve of the welding process.

[0037] Specifically, since there are boundary areas between the sub-regions, simple splicing would cause discontinuities in the curve. Therefore, transition processing is needed between adjacent regions, such as through linear interpolation or other smoothing methods to ensure a natural connection. After splicing, the temperature evolution trajectory of the entire welding process is fully presented, which can show both the rapid changes in the molten pool and the gradual transition between the heat-affected zone and the base material zone.

[0038] In one optional implementation of this embodiment, the step of splicing the fusion temperature results of each sub-region into a time series to obtain the global temperature curve of the welding process includes: constructing a fusion temperature sequence of the sub-region based on the fusion temperature results and corresponding timestamp information, and dividing the fusion temperature sequence into time ranges to generate a sub-region time index table; determining the time boundary points between adjacent sub-regions by sorting and comparing the sub-region time index table, and performing interpolation operations at the boundary points to generate a transition temperature sequence; splicing the temperature data of each time segment based on the fusion temperature sequence and the transition temperature sequence to obtain a continuous global spliced ​​temperature sequence; and generating the global temperature curve of the welding process by smoothing the global spliced ​​temperature sequence.

[0039] In this embodiment, based on the fusion temperature results and corresponding timestamp information, a fusion temperature sequence can be established in each sub-region. The sequence is indexed by time, recording the fusion temperature corresponding to each time point. The timestamp is used to mark the precise moment of each temperature data point during the welding process, thus forming a continuous temperature record for the sub-region. Based on this, by analyzing the start and end times of the temperature sequence and combining it with the stage division of the welding process, the sequence is divided into time ranges. A mapping relationship is established between the temperature data of each sub-region and its corresponding time interval, generating a sub-region time index table. This table records the start time, end time, and time step information of the temperature data in each sub-region, providing a time reference for subsequent cross-region temperature splicing. Subsequently, by sorting and comparing the sub-region time index tables, the time boundary points between adjacent sub-regions can be identified. These boundary points are key nodes for connecting temperature sequences from different regions, and their positions are determined by the end time and start time of the adjacent regions. At the boundary points, interpolation is applied to generate a transition temperature sequence. Linear interpolation can be used to ensure continuous temperature variation within the transition segment, or spline interpolation can be used to smooth temperature changes in the high-temperature gradient region, thus avoiding jumps or spikes caused by direct splicing of data from adjacent sub-regions. The generated transition temperature sequence is combined with the fused temperature sequences of each sub-region to continuously splice the temperature data from each time segment. By connecting the transition segment and each region segment in chronological order, a global spliced ​​temperature sequence is constructed. During the splicing process, consistency of the time index must be ensured so that each time point corresponds to only one fused temperature value in the global sequence. Possible differences in sampling step size and interpolation accuracy are also considered to reduce numerical errors. Finally, the global spliced ​​temperature sequence is smoothed. Smoothing can be achieved through moving average filtering, weighted sliding windows, or low-pass filters. The aim is to eliminate small spikes caused by interpolation or high-frequency fluctuations, making the temperature curve visually and analytically continuous and stable, while preserving the thermal characteristic changes of the molten pool region, heat-affected zone, and matrix region. For example, at the boundary between the molten pool and the heat-affected zone, the transition temperature sequence generated by interpolation allows the temperature to smoothly transition from the peak value of the molten pool to the gradually decreasing temperature of the heat-affected zone. The smoothed global curve not only shows the details of local temperature changes but also reflects the overall heat distribution trend of the welding process, forming a continuous, physically reasonable global temperature curve that can be used for subsequent welding process analysis and control.

[0040] The welding temperature monitoring method based on multi-sensor fusion provided in this application establishes a multi-sub-region temperature monitoring model within the welding area. Data from multiple temperature sensors are introduced into each sub-region, and a time-series temperature matrix is ​​generated by combining differential resampling. Then, the sensor data is dynamically weighted based on real-time quality detection and a pre-set confidence parameter table. This allows for the acquisition of stable fused temperature results for each sub-region under fusion rules, which are then stitched together to form a complete global temperature curve. This application can obtain a global temperature curve that accurately reflects the temperature evolution relationship in different regions, even when sensor data exhibits asynchronous sampling, fluctuation differences, and regional biases, effectively improving the accuracy of temperature detection during the welding process.

[0041] Figure 2 This application provides a multi-sensor fusion welding temperature monitoring device, which can be used to implement the multi-sensor fusion welding temperature monitoring method described in the foregoing embodiments. For example... Figure 2 As shown, the multi-sensor fusion welding temperature monitoring device mainly includes: The acquisition module 10 is used to divide the welding area into several sub-regions according to the spatial distribution of the welding area, and to acquire the initial temperature data corresponding to each sub-region according to multiple types of temperature sensors. The generation module 20 is used to perform time alignment on the initial temperature data by resampling the difference between different sampling points to generate a time-series temperature matrix; The detection module 30 is used to perform real-time quality detection on the time-series temperature matrix, generate a real-time quality score, and weight and combine the real-time quality score with a pre-set confidence parameter table to obtain a dynamic confidence result set. The fusion module 40 is used to fuse valid temperature data exceeding a preset confidence threshold in the dynamic confidence result set according to the fusion rules corresponding to each sub-region, so as to obtain the fused temperature result; The splicing module 50 is used to splice the fusion temperature results of each sub-region in a time series to obtain the global temperature curve of the welding process.

[0042] In one optional implementation of this embodiment, the acquisition module is specifically used to: acquire the temperature gradient distribution of the welding area corresponding to the welding type during the welding process; divide the welding area into three sub-regions—high temperature zone, heat-affected zone, and substrate zone—by mapping the temperature gradient distribution to the spatial distribution of the welding area, and generate a region division table; acquire the original temperature data corresponding to each sub-region based on multiple types of temperature sensors, and label the original temperature data according to the region division table to generate initial temperature data.

[0043] In one optional implementation of this embodiment, the generation module is specifically used to: determine the time difference sequence between sampling points of different sensors based on the sampling frequency and timestamp of multiple types of temperature sensors; generate a unified target sampling step size by performing minimum time interval analysis on the time difference sequence, and construct a reference sampling time axis based on the target sampling step size; perform interpolation calculation on the initial temperature data of multiple types of temperature sensors based on the reference sampling time axis, generate supplementary data points at missing time points, and obtain a completed multi-channel temperature sequence; and merge the temperature readings of different temperature sensors at the same time into a single record by performing time index alignment on the completed multi-channel temperature sequence, thereby obtaining a time-series temperature matrix.

[0044] In one optional implementation of this embodiment, the detection module is specifically used to: calculate a temperature fluctuation index based on the continuous temperature readings of each temperature sensor in the time-series temperature matrix, and generate a fluctuation scoring matrix by comparing the fluctuation index with a preset threshold; determine the data continuity status based on the temperature change sequence of each temperature sensor at adjacent time points in the time-series temperature matrix, and generate a continuity scoring matrix by performing statistical analysis on the continuity status; generate a difference scoring matrix by comparing the relative deviation of the temperature values ​​of different temperature sensors in the same area of ​​the time-series temperature matrix with the average temperature in the area; and generate a dynamic confidence result set for each sub-region by weighted combination of the fluctuation scoring matrix, the continuity scoring matrix, and the difference scoring matrix, and by combining the benchmark confidence of the corresponding temperature sensor in the preset confidence parameter table.

[0045] In one optional implementation of this embodiment, the fusion module is specifically used to: filter sensor data that exceed a preset confidence threshold in the dynamic confidence result set to obtain effective temperature data for each sub-region; Based on the fusion rules corresponding to each sub-region, the weight allocation method of the effective temperature data is determined, and a preliminary fusion temperature result is generated by weighting the effective temperature data. The preliminary fusion temperature result is compared with the outlier judgment conditions, and the outlier data in the preliminary fusion temperature result is removed. The weighted fusion value is then recalculated to generate a corrected fusion temperature result. The final fusion temperature result is determined by matching the corrected fusion temperature result with the historical fusion temperature sequence.

[0046] In an optional implementation of this embodiment, the fusion module is further configured to: extract the temperature change rate between adjacent time points based on the continuous time series of the preliminary fusion temperature results, and generate a change rate judgment matrix by comparing the temperature change rate with a preset change threshold; calculate a deviation index based on the difference between the preliminary fusion temperature results and the historical fusion temperature series in the corresponding sub-region, and generate a deviation judgment matrix by comparing the deviation index with a preset deviation threshold; determine the set of abnormal data points in the preliminary fusion temperature results by jointly analyzing the change rate judgment matrix and the deviation judgment matrix; remove corresponding data based on the set of abnormal data points, and perform a weighted calculation on the remaining preliminary fusion temperature results to generate a corrected fusion temperature result.

[0047] In one optional implementation of this embodiment, the splicing module is specifically used for: constructing a fusion temperature sequence for sub-regions based on the fusion temperature results and corresponding timestamp information, dividing the fusion temperature sequence into time ranges, and generating a sub-region time index table; determining the time boundary points between adjacent sub-regions by sorting and comparing the sub-region time index tables, and performing interpolation operations at the boundary points to generate a transition temperature sequence; splicing the temperature data of each time segment based on the fusion temperature sequence and the transition temperature sequence to obtain a continuous global spliced ​​temperature sequence; and generating a global temperature curve for the welding process by smoothing the global spliced ​​temperature sequence.

[0048] According to the multi-sensor fusion welding temperature monitoring device provided in this application, a multi-sub-region temperature monitoring model is established in the welding area, and data from multiple types of temperature sensors are introduced into each sub-region. A time-series temperature matrix is ​​generated by combining differential resampling. Then, based on real-time quality detection and a pre-set confidence parameter table, the sensor data is dynamically weighted. This allows for the acquisition of stable fused temperature results for each sub-region under fusion rules, and these results are then stitched together to form a complete global temperature curve. This application can obtain a global temperature curve that accurately reflects the temperature evolution relationship in different regions, even when sensor data exhibits asynchronous sampling, fluctuation differences, and regional biases, effectively improving the accuracy of temperature detection during the welding process.

[0049] According to the scheme provided in this application Figure 3 An electronic device is provided as an embodiment of this application. This electronic device can be used to implement the multi-sensor fusion welding temperature monitoring method described in the foregoing embodiments, and mainly includes: The system includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and executable on the processor 302. The memory 301 and the processor 302 are connected via communication. When the processor 302 executes the computer program 303, it implements the multi-sensor fusion welding temperature monitoring method described in the foregoing embodiments. The number of processors can be one or more.

[0050] The memory 301 can be a high-speed random access memory (RAM) or a non-volatile memory, such as a disk storage device. The memory 301 is used to store executable program code, and the processor 302 is coupled to the memory 301.

[0051] Furthermore, embodiments of this application also provide a computer-readable storage medium, which may be disposed in the electronic device described in the above embodiments, and the computer-readable storage medium may be as described above. Figure 3 The memory in the illustrated embodiment.

[0052] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the multi-sensor fusion welding temperature monitoring method described in the foregoing embodiments. Furthermore, the computer-readable storage medium can also be a USB flash drive, external hard drive, read-only memory (ROM), RAM, magnetic disk, or optical disk, or any other medium capable of storing program code.

[0053] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0054] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0055] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A welding temperature monitoring method using multi-sensor fusion, characterized in that, include: The welding area is divided into several sub-areas based on its spatial distribution, and the initial temperature data of each sub-area is obtained using multiple types of temperature sensors. The initial temperature data is time-aligned by resampling the difference between different sampling points to generate a time-series temperature matrix. Based on the continuous temperature readings of each temperature sensor in the time-series temperature matrix, a temperature fluctuation index is calculated, and a fluctuation score matrix is ​​generated by comparing the fluctuation index with a preset threshold. Based on the temperature change sequence of each temperature sensor at adjacent time points in the time-series temperature matrix, the data continuity status is determined, and a continuity score matrix is ​​generated by statistical analysis of the continuity status. Based on the temperature value differences of different temperature sensors in the same area of ​​the time-series temperature matrix, the relative deviation of the temperature value is compared with the average temperature in the area to generate a difference score matrix. By weighting and combining the fluctuation score matrix, the continuity score matrix, and the difference score matrix, and combining them with the benchmark confidence of the corresponding temperature sensor in the preset confidence parameter table, a dynamic confidence result set for each sub-region is generated. According to the fusion rules corresponding to each sub-region, the valid temperature data exceeding the preset confidence threshold in the dynamic confidence result set are fused to obtain the fused temperature result; The fusion temperature results of each sub-region are spliced ​​together over time to obtain the global temperature curve of the welding process.

2. The welding temperature monitoring method based on multi-sensor fusion according to claim 1, characterized in that, The step of dividing the welding area into several sub-regions according to the spatial distribution of the welding area, and obtaining the initial temperature data corresponding to each sub-region according to multiple types of temperature sensors, includes: Obtain the temperature gradient distribution of the welding area corresponding to the welding type during the welding process; By mapping the temperature gradient distribution to the spatial distribution of the welding area, the welding area is divided into three sub-regions: a high-temperature region, a heat-affected region, and a substrate region, and a region division table is generated. The raw temperature data corresponding to each sub-region is obtained by multiple types of temperature sensors, and the raw temperature data is labeled according to the region division table to generate initial temperature data.

3. The welding temperature monitoring method based on multi-sensor fusion according to claim 1, characterized in that, The step of performing time alignment on the initial temperature data through differential resampling at different sampling points to generate a time-series temperature matrix includes: Based on the sampling frequency and timestamp of multiple types of temperature sensors, determine the time difference sequence between sampling points of different sensors; By performing minimum time interval analysis on the time difference sequence, a unified target sampling step size is generated, and a reference sampling time axis is constructed based on the target sampling step size; Based on the reference sampling time axis, interpolation is performed on the initial temperature data of multiple types of temperature sensors to generate supplementary data points at missing time points, resulting in a complete multi-channel temperature sequence. By performing time index alignment on the completed multi-channel temperature sequence, the temperature readings of different temperature sensors at the same time are merged into a single record, resulting in a time-series temperature matrix.

4. The welding temperature monitoring method based on multi-sensor fusion according to claim 1, characterized in that, The step of fusing valid temperature data exceeding a preset confidence threshold in the dynamic confidence result set according to the fusion rules corresponding to each sub-region to obtain the fused temperature result includes: Sensor data exceeding a preset confidence threshold are filtered from the dynamic confidence result set to obtain effective temperature data for each sub-region; Based on the fusion rules corresponding to each sub-region, the weight allocation method of the effective temperature data is determined, and a preliminary fusion temperature result is generated by weighting the effective temperature data. The preliminary fusion temperature result is compared with the outlier determination criteria, the abnormal data in the preliminary fusion temperature result is removed, and the weighted fusion value is recalculated to generate the corrected fusion temperature result. The final fusion temperature result is determined by matching the corrected fusion temperature result with the historical fusion temperature sequence.

5. The welding temperature monitoring method based on multi-sensor fusion according to claim 4, characterized in that, Before the step of comparing the preliminary fusion temperature result with the outlier determination criteria, removing outlier data from the preliminary fusion temperature result, and recalculating the weighted fusion value to generate the corrected fusion temperature result, the method further includes: Based on the continuous time series of the preliminary fused temperature results, the temperature change rate between adjacent time points is extracted, and a change rate determination matrix is ​​generated by comparing the temperature change rate with a preset change threshold. Based on the difference between the preliminary fusion temperature result and the historical fusion temperature sequence in the corresponding sub-region, a deviation index is calculated, and a deviation judgment matrix is ​​generated by comparing the deviation index with a preset deviation threshold. By jointly analyzing the rate of change determination matrix and the deviation determination matrix, the set of abnormal data points in the preliminary fused temperature results is determined; Based on the set of abnormal data points, the corresponding data is removed, and the remaining preliminary fusion temperature results are weighted to generate the corrected fusion temperature results.

6. The welding temperature monitoring method based on multi-sensor fusion according to claim 4, characterized in that, The step of stitching together the fusion temperature results of each sub-region over time to obtain the global temperature curve of the welding process includes: Based on the fusion temperature results and the corresponding timestamp information, a fusion temperature sequence for the sub-region is constructed, and the fusion temperature sequence is divided into time ranges to generate a sub-region time index table. By sorting and comparing the time index table of the sub-regions, the time boundary points between adjacent sub-regions are determined, and interpolation is performed at the boundary points to generate a transition temperature sequence. Based on the fused temperature sequence and the transition temperature sequence, the temperature data of each time segment are spliced ​​together to obtain a continuous global spliced ​​temperature sequence. By smoothing the global splicing temperature sequence, a global temperature curve for the welding process is generated.

7. A welding temperature monitoring device based on multi-sensor fusion, characterized in that, The multi-sensor fusion welding temperature monitoring device is used to implement the multi-sensor fusion welding temperature monitoring method of claim 1, and the multi-sensor fusion welding temperature monitoring device includes: The acquisition module is used to divide the welding area into several sub-regions according to the spatial distribution of the welding area, and to acquire the initial temperature data corresponding to each sub-region according to multiple types of temperature sensors. The generation module is used to perform time alignment on the initial temperature data by resampling the difference between different sampling points to generate a time-series temperature matrix; The detection module is used to calculate a temperature fluctuation index based on the continuous temperature readings of each temperature sensor in the time-series temperature matrix, and generate a fluctuation score matrix by comparing the fluctuation index with a preset threshold; determine the data continuity status based on the temperature change sequence of each temperature sensor at adjacent time points in the time-series temperature matrix, and generate a continuity score matrix by statistically analyzing the continuity status; generate a difference score matrix by comparing the relative deviation of the temperature values ​​of different temperature sensors in the same area of ​​the time-series temperature matrix with the average temperature in the area; and generate a dynamic confidence result set for each sub-region by weighting and combining the fluctuation score matrix, the continuity score matrix, and the difference score matrix, and combining them with the benchmark confidence of the corresponding temperature sensor in the preset confidence parameter table. The fusion module is used to fuse the valid temperature data exceeding the preset confidence threshold in the dynamic confidence result set according to the fusion rules corresponding to each sub-region, so as to obtain the fused temperature result; The splicing module is used to splice the fusion temperature results of each sub-region in a time series to obtain the global temperature curve of the welding process.

8. An electronic device, characterized in that, Includes memory and processor, of which: The processor is used to execute computer programs stored in the memory; When the processor executes the computer program, it implements the steps in the welding temperature monitoring method with multi-sensor fusion as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps in the welding temperature monitoring method with multi-sensor fusion as described in any one of claims 1 to 6.

Images