Adaptive control method, device and equipment of multi-axis soldering machine and storage medium

Abstract

The application provides a kind of adaptive control method, device, equipment and storage medium of multi-axis soldering machine, the method includes: by temperature sensor, vision sensor and environmental sensor, current soldering head temperature, solder joint position information, PCB board characteristic information and environmental parameter are collected, and current state data is obtained;Inference calculation is carried out by main controller according to current state data, and the welding process parameters for current solder joint are obtained;Control command is generated by motion controller according to welding process parameters and sent to soldering shaft unit, and adjusted execution parameter is obtained;After welding, the solder joint is detected by vision sensor, and solder joint quality evaluation data is obtained, and state data, process parameters and quality evaluation data are stored for subsequent optimization.The application can dynamically adjust welding parameters according to real-time working condition, and improve the stability of welding quality.

Description

Technical Field

[0001] This invention relates to the field of equipment control technology, and in particular to an adaptive control method, device, equipment and storage medium for a multi-axis soldering machine. Background Technology

[0002] Multi-axis soldering machines are widely used automated soldering equipment in the electronics manufacturing industry. They achieve precise soldering of solder joints on PCB boards through the coordinated control of multiple motion axes. Traditional multi-axis soldering machines typically use a fixed parameter control method, that is, the soldering process parameters such as temperature, speed, and time are preset based on experience or experiments, and these parameters are stored in the control system for use during soldering.

[0003] However, in actual production, factors such as PCB material, solder joint location, component type, and ambient temperature and humidity often vary and change. Fixed parameter control methods cannot dynamically adjust welding parameters according to real-time working conditions, resulting in unstable welding quality and defects such as cold solder joints and missing solder joints, which affect product reliability and production efficiency. Summary of the Invention

[0004] The main objective of this invention is to solve the technical problem of unstable welding quality caused by the use of fixed parameter control in existing multi-axis soldering machines; This invention provides an adaptive control method for a multi-axis soldering machine. The multi-axis soldering machine includes a soldering axis unit, a temperature sensor, a vision sensor, a motion controller, an environmental sensor, and a main controller. The adaptive control method includes: The current status data is obtained by acquiring the real-time temperature of the solder head through the temperature sensor, the position information of the solder joint to be soldered and the feature information of the PCB board through the vision sensor, and the ambient temperature and humidity parameters through the environmental sensor. Based on the current status data, the main controller performs inference calculations to obtain the welding process parameters for the current weld point; Based on the welding process parameters, the motion controller generates control commands and sends the control commands to the soldering shaft unit to obtain the adjusted execution parameters of the soldering shaft unit. After welding is completed, the weld point is detected by the vision sensor to obtain weld point quality evaluation data. The current status data, the welding process parameters, and the weld point quality evaluation data are stored in the main controller for subsequent inference calculation optimization.

[0005] The present invention also provides an adaptive control device for a multi-axis soldering machine, the multi-axis soldering machine comprising a soldering axis unit, a temperature sensor, a vision sensor, a motion controller, an environmental sensor, and a main controller, the adaptive control device for the multi-axis soldering machine comprising: The data acquisition module is used to acquire the real-time temperature of the current solder head through the temperature sensor, acquire the position information of the solder joint to be soldered and the feature information of the PCB board through the vision sensor, and acquire the ambient temperature and humidity parameters through the environmental sensor to obtain the current status data. The parameter reasoning module is used to perform reasoning calculations by the main controller based on the current state data to obtain the welding process parameters for the current weld point. The instruction generation module is used to generate control instructions from the motion controller based on the welding process parameters, and send the control instructions to the soldering shaft unit respectively to obtain the adjusted execution parameters of the soldering shaft unit; The quality feedback module is used to detect the weld joint through the vision sensor after welding is completed, obtain weld joint quality evaluation data, and store the current status data, welding process parameters and weld joint quality evaluation data in the main controller for subsequent inference calculation optimization.

[0006] The present invention also provides an adaptive control device for a multi-axis soldering machine, comprising: a memory and at least one processor, wherein the memory stores instructions, and the memory and the at least one processor are interconnected by a circuit; the at least one processor invokes the instructions in the memory to cause the adaptive control device of the multi-axis soldering machine to perform the steps of the adaptive control method of the multi-axis soldering machine described above.

[0007] The present invention also provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the steps of the adaptive control method for the multi-axis soldering machine described above.

[0008] The aforementioned adaptive control method, device, equipment, and storage medium for a multi-axis soldering machine collects current solder head temperature, solder joint position information, PCB board feature information, and environmental parameters through temperature sensors, vision sensors, and environmental sensors to obtain current state data. The main controller performs inference calculations based on the current state data to obtain soldering process parameters for the current solder joint. The motion controller generates control commands based on the soldering process parameters and sends them to the solder axis unit to obtain adjusted execution parameters. After soldering is completed, the solder joint is inspected by a vision sensor to obtain solder joint quality evaluation data. The state data, process parameters, and quality evaluation data are stored for subsequent optimization. This invention can dynamically adjust soldering parameters according to real-time operating conditions, improving the stability of soldering quality.

[0009] Beneficial effects: This invention achieves adaptive adjustment of welding parameters as working conditions change through a closed-loop mechanism of real-time status perception, dynamic parameter reasoning, precise command control, and quality feedback optimization. This effectively solves the problem of unstable welding quality caused by fixed parameter control, and improves the consistency of welding quality and the reliability of the production process.

[0010] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0011] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the first embodiment of the adaptive control method for a multi-axis soldering machine in this invention. Figure 2 This is a schematic diagram of a second embodiment of the adaptive control method for a multi-axis soldering machine according to the present invention; Figure 3 This is a schematic diagram of an embodiment of the adaptive control device for a multi-axis soldering machine according to the present invention; Figure 4 This is a schematic diagram of an embodiment of the adaptive control device for a multi-axis soldering machine according to the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] The terms "comprising" and "having," and any variations thereof, used in the embodiments of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0015] To facilitate understanding of this embodiment, a detailed description of an adaptive control method for a multi-axis soldering machine disclosed in this embodiment of the invention will be provided first. This multi-axis soldering machine includes a soldering axis unit, a temperature sensor, a vision sensor, a motion controller, an environmental sensor, and a main controller. For example... Figure 1 As shown, this method includes the following steps: 101. The real-time temperature of the solder head is collected by the temperature sensor, the position information of the solder joint to be soldered and the feature information of the PCB board are collected by the vision sensor, and the ambient temperature and humidity parameters are obtained by the environmental sensor to obtain the current status data. In this embodiment, the multi-axis soldering machine needs to collect real-time operating condition information through multiple sensors when performing soldering tasks. Specifically, the temperature sensor can be a thermocouple or an infrared thermometer, installed near the solder head to monitor the surface temperature of the solder head in real time. The vision sensor can be an industrial camera, which uses image processing algorithms to identify the position coordinates of the solder joint to be soldered on the PCB board, and simultaneously extract feature information such as the type, size, and number of solder pads of the PCB board. The environmental sensor can be a temperature and humidity sensor, set in the working area of ​​the soldering machine to monitor changes in the ambient temperature and humidity of the workshop.

[0016] Furthermore, the main controller can aggregate and integrate the collected real-time temperature data, solder joint location information, PCB board feature information, and environmental parameters to form multi-dimensional current status data. This current status data comprehensively reflects the current welding conditions and can be used as input for subsequent inference calculations.

[0017] It should be noted that in practical applications, the current status data can also include equipment status information such as the current position and movement speed of the soldering shaft unit, so as to more comprehensively characterize the real-time status of the soldering system.

[0018] 102. Based on the current status data, the main controller performs inference calculations to obtain the welding process parameters for the current weld point; In this embodiment, after acquiring the current state data, the main controller can perform inference calculations based on a pre-trained control strategy to dynamically generate welding process parameters adapted to the current operating conditions. Specifically, the welding process parameters may include key parameters such as welding temperature, welding time, solder feeding speed, and movement speed, which directly affect welding quality and efficiency.

[0019] Traditional multi-axis soldering machines typically use fixed parameter tables, meaning several sets of fixed parameters are pre-set for different types of PCBs, and the corresponding parameter set is called during soldering based on the PCB type. However, this method cannot cope with changes in actual production conditions, such as ambient temperature fluctuations, solder head temperature drift, and differences between different batches of PCBs.

[0020] The inference calculation method in this embodiment can adjust the welding process parameters in real time based on the current state data. Furthermore, the main controller can input the current state data into the inference model, which can be implemented based on algorithms such as reinforcement learning, neural networks, or fuzzy control. The inference model comprehensively considers factors such as the current solder head temperature, solder joint position, PCB board characteristics, and environmental parameters, and outputs the optimal welding process parameters for the current solder joint.

[0021] For example, when the ambient temperature is detected to be low, the inference model may output a higher soldering temperature and a longer preheating time to compensate for the impact of ambient heat dissipation; when the PCB board is detected to be a multilayer board, the inference model may output a longer soldering time to ensure that the solder melts fully and penetrates into each layer of pads.

[0022] Understandably, the core of the inference calculation process lies in establishing a mapping relationship between the current state and the optimal process parameters. This mapping relationship can be obtained through offline training, that is, by training the inference model using historical welding data, enabling it to learn the optimal parameter selection strategy under different working conditions. In actual welding, the inference model can complete calculations in milliseconds, achieving real-time parameter adjustments.

[0023] It should be noted that the welding process parameters obtained through inference calculations may include multiple dimensions such as temperature parameters, time parameters, and speed parameters. These parameters will be converted into specific control commands in subsequent steps and sent to the soldering shaft unit for execution.

[0024] 103. Based on the welding process parameters, the motion controller generates control commands and sends the control commands to the soldering shaft unit respectively to obtain the adjusted execution parameters of the soldering shaft unit; In this embodiment, each soldering axis unit includes a solder head, a heater, a solder feeding mechanism, and a drive motor. The process of generating control commands by the motion controller based on the welding process parameters and sending these commands to the corresponding soldering axis units to obtain adjusted execution parameters for each soldering axis unit includes: determining whether each parameter is within a preset safety range based on the welding process parameters, performing boundary correction on parameters exceeding the range, and obtaining safety-verified welding process parameters; planning the motion trajectory of each motion axis from its current position to the target solder point based on the position information of the solder joint to be welded and the current position of the soldering axis unit, and calculating the motion speed of each motion axis based on the safety-verified welding process parameters to obtain a multi-axis linkage motion scheme; generating control signals for the heater, the solder feeding mechanism, and each motion axis based on the multi-axis linkage motion scheme and the safety-verified welding process parameters to obtain hardware control commands; and sending the hardware control commands to the soldering axis units by the motion controller to control each motion axis to move collaboratively to the target solder point and perform the welding operation, thereby obtaining the adjusted execution parameters for the soldering axis units.

[0025] Specifically, after receiving the welding process parameters output by the main controller, the motion controller needs to convert these abstract process parameters into control commands that can be recognized and executed by the various hardware components of the soldering axis unit. Specifically, each soldering axis unit includes multiple motion axes (such as X-axis, Y-axis, Z-axis, R-axis, etc.), solder head, heater, solder feeding mechanism, and drive motor. These components need to complete the welding action under the coordination of the motion controller.

[0026] The motion controller performs a safety check on the soldering process parameters calculated through inference. Since the inference model may output parameter values ​​exceeding the physical limitations of the equipment under certain extreme conditions, it is necessary to determine whether parameters such as temperature, speed, and time are within preset safety ranges. For example, soldering temperature is typically limited to the range of 250-450℃, and solder feed speed is limited to the range of 1-10mm / s. If a parameter exceeds the safety range, it is truncated to the nearest boundary value to avoid equipment damage or soldering failure, resulting in the safety-checked soldering process parameters.

[0027] After completing the safety verification, the motion controller can plan the motion trajectory of each motion axis based on the position information of the solder joint to be soldered and the current position of the solder axis unit. Specifically, it can calculate the displacement vector from the current position to the target solder joint and generate a sequence of trajectory points using linear interpolation or circular interpolation algorithms. Simultaneously, based on the time parameter in the soldering process parameters, it calculates the required motion speed of each motion axis. For example, if the requirement is to reach a solder joint 100mm from the current position within 2 seconds, the combined speed of all motion axes should be 50mm / s. The speed of each axis can be proportionally allocated based on the displacement components in the X, Y, and Z directions. This speed allocation method ensures that all motion axes reach the target position synchronously, forming a multi-axis linkage motion scheme.

[0028] The motion controller generates specific hardware control commands based on the multi-axis linkage motion scheme and the welding process parameters after safety verification. Specifically, temperature parameters are converted into heater power control signals, such as using a PID control algorithm to adjust the heater's output power to stabilize the solder head temperature at the target value; solder feeding speed parameters are converted into speed control signals for the stepper motor or servo motor in the solder feeding mechanism; and the speed and position parameters of each motion axis are converted into pulse signals or position commands for the drive motors. These control signals together constitute the hardware control commands.

[0029] The motion controller sends hardware control commands to each component of the soldering axis unit via a communication bus (such as CAN bus, EtherCAT, etc.). Upon receiving the commands, each motion axis begins coordinated movement, driving the solder head along a planned trajectory to the target solder joint; the heater adjusts the solder head temperature according to the power control signal; and the solder feeding mechanism delivers solder wire at a set speed upon reaching the solder joint. Throughout the process, the execution status of each component is fed back to the motion controller in real time, forming adjusted execution parameters for the soldering axis unit, used to monitor whether the soldering process proceeds as expected.

[0030] Furthermore, the step of planning the motion trajectory of each motion axis from the current position to the target solder joint based on the position information of the solder joint to be soldered and the current position of the solder axis unit, and calculating the motion speed of each motion axis based on the welding process parameters after safety verification to obtain a multi-axis linkage motion scheme includes: calculating the displacement from the current position to the target position based on the position information of the solder joint to be soldered and the current position of the solder axis unit to obtain motion path parameters; generating the motion trajectory of each motion axis based on the motion path parameters to obtain a multi-axis motion trajectory; calculating the motion speed of each motion axis based on the time parameter in the welding process parameters after safety verification and the multi-axis motion trajectory to obtain the target speed of each motion axis; and generating a smooth acceleration / deceleration curve for each motion axis based on the target speed to obtain a multi-axis linkage motion scheme.

[0031] Specifically, the motion controller can acquire the three-dimensional coordinates of the solder joint to be soldered in the working coordinate system, as well as the current position coordinates of the solder axis unit. By subtracting the two coordinates, the required displacement of each axis can be obtained, including the X-direction displacement, Y-direction displacement, and Z-direction displacement. These displacements constitute a displacement vector, and the length of this displacement vector is the total distance the solder head needs to move. The aforementioned displacements and the total distance together constitute the motion path parameters.

[0032] The motion controller can use a linear interpolation algorithm to generate the motion trajectory of each motion axis. Specifically, the interpolation period can be set, for example, calculating the position coordinates that each axis should reach every 1 millisecond. Based on the total motion time and the interpolation period, the motion controller determines the number of interpolation points to be generated and calculates the position coordinates of the X, Y, and Z axes at each interpolation moment according to the displacement ratio. These position coordinate sequences arranged in chronological order constitute the motion trajectory of each motion axis.

[0033] It should be noted that in certain soldering scenarios, the solder tip needs to contact the solder joint at a specific angle, in which case the rotary axis also needs to participate in motion planning. The motion controller can calculate the target angle to which the rotary axis needs to rotate based on the direction vector of the solder joint or the normal vector of the PCB board, and generate the corresponding angle change trajectory.

[0034] After generating the multi-axis motion trajectory, the motion controller calculates the motion speed of each axis based on the time parameter in the welding process parameters. The time parameter specifies the maximum allowable time for the solder head to move from its current position to the target solder joint. The motion controller can calculate the average motion speed based on the total displacement distance and the time parameter. This average speed needs to be decomposed onto each axis. Since the displacement of each axis is different, the speed allocation is based on the displacement ratio. Specifically, axes with larger displacements are assigned higher speeds, and axes with smaller displacements are assigned lower speeds, but the motion time of each axis remains consistent, thereby ensuring that each axis can reach the target position synchronously.

[0035] However, if each motion axis starts and stops directly at the target speed, it will cause instantaneous acceleration changes, leading to mechanical shock and vibration. Therefore, the motion controller needs to generate smooth acceleration and deceleration curves for each motion axis. Commonly used acceleration and deceleration curves include trapezoidal acceleration and deceleration curves and S-shaped acceleration and deceleration curves. S-shaped acceleration and deceleration curves, due to the continuous change in acceleration, can effectively reduce mechanical shock.

[0036] Specifically, the S-shaped acceleration / deceleration curve can be divided into an acceleration phase, a constant speed phase, and a deceleration phase. In the acceleration phase, the speed smoothly increases from zero to the target speed; in the constant speed phase, the target speed is maintained; and in the deceleration phase, the speed smoothly decreases from the target speed to zero. The motion controller can calculate the time allocation and speed change curves for each phase based on preset acceleration and deceleration time constants. These acceleration / deceleration curves, together with the aforementioned motion trajectory and target speed, constitute a complete multi-axis linkage motion scheme.

[0037] 104. After welding is completed, the weld point is detected by the vision sensor to obtain weld point quality evaluation data, and the current status data, the welding process parameters and the weld point quality evaluation data are stored in the main controller for subsequent inference calculation optimization.

[0038] In this embodiment, after the soldering spindle unit completes the soldering operation, the vision sensor can perform quality inspection on the solder joints. Specifically, the vision sensor acquires images of the solder joint area and analyzes the morphological characteristics of the solder joints using image processing algorithms, such as the fullness of the solder joints, wetting angle, uniformity of solder distribution, and the presence of pores or cracks. These characteristics can be used to determine whether the soldering quality is acceptable.

[0039] Solder joint quality evaluation data can be represented in different ways. In some embodiments, a binary evaluation, i.e., "pass" or "fail", can be used. In other embodiments, a score system can be used, such as 0-100 points, where a higher score indicates better soldering quality. Evaluation can also be categorized according to the specific defect type of the solder joint, such as "cold solder joint", "insufficient solder", or "solder ball". These evaluation results constitute the solder joint quality evaluation data.

[0040] After obtaining the solder joint quality evaluation data, the main controller packages and stores the current status data, welding process parameters, and solder joint quality evaluation data for this welding process. Specifically, the current status data includes solder head temperature, solder joint location, PCB board features, environmental parameters, etc., collected before welding; the welding process parameters include parameters such as temperature, time, and speed output by inference calculations; and the solder joint quality evaluation data reflects the welding effect produced by the combination of these parameters. These three sets of data together constitute a complete welding record.

[0041] These welding records are continuously accumulated in the main controller's storage unit for subsequent inference calculation optimization. Referring to the explanation in step 102 above, the inference calculation is implemented based on a pre-trained inference model. This inference model initially learns the mapping relationship between operating conditions and process parameters through offline training data. However, with changes in the actual production environment, such as solder head aging, seasonal fluctuations in ambient temperature, and differences in the characteristics of different batches of PCBs, the initially trained inference model may gradually exhibit prediction bias.

[0042] Therefore, the main controller can periodically perform incremental updates to the inference model using newly accumulated welding records. Specifically, when the number of stored welding records reaches a certain threshold, such as 500 or 1000 new records, the main controller can trigger a model update process. This update process extracts state data, process parameters, and quality evaluation data from these new records and calculates reward values ​​based on the quality evaluation data. Records that meet quality standards are assigned positive rewards, while records that do not meet quality standards are assigned negative rewards. Through this new data with reward labels, the inference model can undergo incremental training, adjusting its internal parameters so that its output process parameters better adapt to the current actual working conditions.

[0043] In this embodiment, temperature sensors, vision sensors, and environmental sensors collect current solder head temperature, solder joint position information, PCB board feature information, and environmental parameters to obtain current status data. The main controller performs inference calculations based on the current status data to obtain soldering process parameters for the current solder joint. The motion controller generates control commands based on the soldering process parameters and sends them to the solder axis unit to obtain adjusted execution parameters. After soldering is completed, the solder joint is inspected by a vision sensor to obtain solder joint quality evaluation data. The status data, process parameters, and quality evaluation data are stored for subsequent optimization. This invention can dynamically adjust soldering parameters according to real-time operating conditions, improving the stability of soldering quality.

[0044] Please see Figure 2 Another embodiment of the adaptive control method for a multi-axis soldering machine in this application includes: 201. The real-time temperature of the solder head is collected by the temperature sensor, the position information of the solder joint to be soldered and the feature information of the PCB board are collected by the vision sensor, and the ambient temperature and humidity parameters are obtained by the environmental sensor to obtain the current status data. In this embodiment, step 201 is similar to step 101, and will not be described again here.

[0045] 202. Normalize the real-time temperature, location information, feature information, and environmental parameters in the current state data to obtain a standardized state vector; In this embodiment, the current state data collected by the main controller includes data on various physical quantities, with significant differences in dimensions and numerical ranges. For example, the solder head temperature may be in the range of 250-450℃, the solder joint coordinates may be in the range of 0-300mm, and the ambient humidity may be in the range of 30%-70%. These differences in numerical ranges will affect the calculation performance of the subsequent inference model, therefore, these data need to be normalized.

[0046] Normalization can map data with different dimensions to the same numerical range, such as between 0 and 1. Specifically, for temperature data, a min-max normalization method can be used, where the minimum temperature value is subtracted from the temperature value and then divided by the temperature range to obtain the normalized temperature value. The same method can also be applied to other data such as location coordinates and environmental parameters. After normalization, all data are unified to the same numerical scale.

[0047] These normalized data are arranged in a certain order to form a multi-dimensional vector. For example, the normalized temperature, X coordinate, Y coordinate, Z coordinate, PCB board characteristic parameters, ambient temperature, and ambient humidity can be arranged in sequence to form a seven-dimensional state vector. This vector is the standardized state vector, which can be used as input to the inference model.

[0048] 203. Input the standardized state vector into a pre-trained Q-value evaluation function, and calculate the corresponding value score for each of the preset candidate welding process parameters; In this embodiment, the main controller stores a pre-trained Q-value evaluation function. This Q-value evaluation function is a function model trained based on a reinforcement learning algorithm, and its function is to evaluate the long-term value of taking a certain welding process parameter under specific conditions. The so-called long-term value considers not only the welding quality of the current weld joint, but also the impact of the parameter selection on the subsequent welding process.

[0049] The Q-value evaluation function is typically implemented using a neural network structure, which includes an input layer, multiple hidden layers, and an output layer. The standardized state vector is fed into the input layer of the neural network as input data. Simultaneously, the main controller pre-sets multiple sets of candidate welding process parameters. These candidate parameters can be several typical parameter combinations preset based on experience, such as "high temperature short time," "low temperature long time," and "standard parameters." Each set of candidate parameters can also be encoded and input into the neural network along with the state vector.

[0050] After the neural network performs multi-layer nonlinear transformations on the input data, it outputs a Q-value. This Q-value represents the expected performance score of the candidate parameter under the current state; a higher Q-value indicates that the parameter is more suitable for the current operating condition. The main controller sequentially inputs each set of candidate parameters and state vectors into the Q-value evaluation function to calculate the corresponding Q-values. These Q-values ​​constitute the value score of each candidate welding process parameter.

[0051] It should be noted that the number of candidate parameters can be set according to actual needs. The more candidate parameters there are, the greater the range of choices, but the computational load also increases accordingly. In practical applications, an appropriate number of candidate parameters, such as 10-50 groups, can be selected based on computational capabilities and response time requirements.

[0052] 204. Based on the value score, select the candidate welding process parameter with the largest Q value as the preliminary welding process parameter; In this embodiment, after obtaining the value scores of each candidate welding process parameter, the main controller can compare the magnitudes of these Q values. Since the Q value reflects the expected effect of the parameter, the candidate parameter with the largest Q value means that the best welding quality is expected to be obtained under the current operating conditions. Therefore, the main controller selects the set of candidate parameters with the largest Q value as the initial welding process parameters. This initial parameter is the optimal selection calculated based on the inference model.

[0053] 205. Calculate the deviation between the preliminary welding process parameters and the parameter distribution center of historical successful welding samples; In this embodiment, although the initial welding process parameters are the optimal parameters evaluated by the inference model, these parameters may still carry certain risks. Since the inference model is trained on historical data, its predictive ability is primarily concentrated within the parameter range covered by that historical data. If the initial parameters deviate significantly from the historical data distribution, the inference model's predictions may be unreliable.

[0054] The main controller stores parameter distribution information for historical successful welding samples. These historical samples are welding records that were rated as qualified in past welding processes. By statistically analyzing the process parameters of these samples, the central location and range of the parameter distribution can be obtained. The center of the parameter distribution can be represented by the average values ​​of each parameter dimension, such as the average temperature, average time, and average speed of the historical samples.

[0055] The main controller can calculate the distance between the initial welding process parameters and the center of that parameter distribution; this distance is called the deviation. The deviation can be calculated using methods such as Euclidean distance or Mahalanobis distance. The larger the deviation, the further the initial parameters deviate from historical experience ranges, and the higher the risk of using these parameters.

[0056] 206. When the deviation exceeds the preset safety threshold, the preliminary welding process parameters are adjusted along the gradient direction to the boundary range of the parameter distribution to obtain the welding process parameters for the current weld point. When the deviation does not exceed the preset safety threshold, the preliminary welding process parameters are used as the welding process parameters for the current weld point. In this embodiment, the main controller compares the calculated deviation with a preset safety threshold. This safety threshold is preset based on the distribution characteristics of historical data and is used to determine whether the initial parameters are within an acceptable range.

[0057] If the deviation does not exceed the safety threshold, it means that the preliminary parameters are within the coverage of historical data, and the prediction of the inference model is relatively reliable. At this time, the preliminary welding process parameters can be directly used as the final welding process parameters for output.

[0058] If the deviation exceeds the safety threshold, it indicates that the initial parameters deviate significantly from historical experience, indicating high uncertainty. In this case, the main controller needs to conservatively correct the initial parameters. Specifically, a direction vector pointing from the parameter distribution center to the initial parameters can be calculated. A point on the parameter distribution boundary can be found along this direction, and the initial parameters can be adjusted to the position of that boundary point. This adjustment method preserves the directionality of the initial parameters (i.e., which direction the inference model believes the parameters should be adjusted) while ensuring that the parameters do not deviate excessively from historical experience, reducing the risk of welding failure. The adjusted parameters are then output as the final welding process parameters.

[0059] Furthermore, before the main controller performs inference calculations based on the current state data to obtain the welding process parameters for the current weld point, the method further includes: extracting historical welding records from the main controller's storage unit, each record including historical state data, corresponding welding process parameters, and weld point quality evaluation data, and constructing an offline training dataset; establishing a safe parameter space for welding process parameters based on the offline training dataset, the safe parameter space being used to constrain the parameter range during the training process; initializing the network parameters of the Q-value evaluation function, and iteratively updating the network parameters by minimizing the loss function to obtain the trained Q-value evaluation function.

[0060] Specifically, in this embodiment, the Q-value evaluation function needs to undergo an offline training process before practical application. The main controller can extract historically accumulated soldering record data from the storage unit. These historical records originate from the actual production process of the multi-axis soldering machine over a period of time, with each record fully documenting all information of a soldering operation. Historical status data includes operating condition information such as solder head temperature, solder joint location, PCB board characteristics, and environmental parameters before soldering; soldering process parameters include control parameters such as temperature, time, and speed used at the time; and solder joint quality evaluation data records the quality inspection results of that soldering operation. After summarizing and organizing these historical records, an offline training dataset can be constructed.

[0061] Based on the offline training dataset, the main controller can analyze the distribution of welding process parameters in these historical records. Since the historical records are all from actual production processes, the process parameters corresponding to the records with acceptable quality reflect the feasible parameter range of the equipment under actual operating conditions. By statistically analyzing the value ranges of these parameters, the upper and lower boundaries of each parameter dimension can be determined. For example, the value range of the temperature parameter in the historical records may be between 280-420℃, and the value range of the time parameter may be between 1.5-4 seconds. These parameter value ranges together constitute the safety parameter space. The role of this safety parameter space is to constrain the learning range of the Q-value evaluation function during subsequent training, preventing the model from learning parameter combinations that have never occurred in actual production or are obviously infeasible.

[0062] After initializing the network parameters of the Q-value evaluation function, the main controller begins model training. The training process employs an iterative optimization approach. In each iteration, the main controller randomly selects a batch of sample data from the offline training dataset, inputs the historical state data from these samples into the Q-value evaluation function, and calculates the predicted Q-value. Simultaneously, based on the solder joint quality evaluation data and the information of the next state in the samples, the target Q-value for that state-action pair can be calculated. Comparing the predicted Q-value with the target Q-value yields the prediction error, which is part of the loss function.

[0063] In addition to prediction error, the loss function also includes a conservative constraint. This constraint penalizes parameter selections that exceed the safe parameter space. Specifically, during training, the main controller, besides evaluating the parameters actually used in the samples, samples candidate parameters both inside and outside the safe parameter space and calculates their Q-values. For candidate parameters within the space, a higher Q-value is desired; for candidate parameters outside the space, a lower Q-value is desired. In this way, the model is guided to learn a conservative strategy, tending to select parameters within the historical data coverage rather than risking parameters from unknown regions.

[0064] The main controller updates the network parameters based on the loss function using gradient descent, gradually decreasing the value of the loss function. This process is repeated multiple times, each iteration traversing the entire training dataset. When the rate of decrease in the loss function becomes very slow, or when the performance on the validation dataset reaches a preset standard, the training process is considered to have converged, and the Q-value evaluation function for training is obtained.

[0065] Furthermore, the initialization of the network parameters for the Q-value evaluation function, and the iterative updating of the network parameters by minimizing the loss function to obtain the trained Q-value evaluation function, includes: initializing the network parameters for the Q-value evaluation function, and performing data cleaning and reward labeling on the offline training dataset, removing unqualified samples and calculating reward values ​​for the remaining samples to obtain a labeled training dataset; based on the labeled training dataset, extracting the distribution range of welding process parameters to construct the boundary of the safety parameter space; for each training sample in the labeled training dataset, calculating a loss value including a Bellman error term and a conservative constraint term, wherein the Bellman error term is obtained by comparing the predicted Q-value with the target Q-value, and the conservative constraint term is obtained by comparing the candidate parameter Q-values ​​inside and outside the safety parameter space; and iteratively updating the network parameters based on the loss value using gradient descent until the loss value converges to obtain the trained Q-value evaluation function.

[0066] Specifically, the main controller needs to preprocess the offline training dataset. Since historical welding records originate from actual production processes, they may contain some substandard or abnormal samples, such as welding failures due to equipment malfunctions. Directly using these substandard samples for training would mislead the model into learning incorrect parameter selection strategies. Therefore, the main controller can filter the samples based on weld quality evaluation data, removing those that are deemed substandard.

[0067] For the retained qualified samples, the main controller needs to calculate a reward value for them. The reward value quantifies the quality of the process parameter selection in that sample. Solder joint quality evaluation data may be discrete levels (e.g., excellent, good, acceptable) or continuous scores (e.g., 0-100). The main controller can design a reward function based on this quality evaluation data, converting the quality evaluation into a numerical reward. For example, samples with higher quality scores receive higher reward values. After data cleaning and reward labeling, the labeled training dataset is obtained.

[0068] The main controller, based on the labeled training dataset, statistically analyzes the distribution characteristics of welding process parameters. Specifically, it can calculate the minimum and maximum values ​​for each parameter dimension. For example, it iterates through all samples of temperature parameters to find the lowest and highest temperatures; similarly, it processes other parameter dimensions such as time and speed. The minimum and maximum values ​​of these parameter dimensions collectively define a multi-dimensional hyperrectangular region, which serves as the boundary of the safety parameter space. This boundary ensures that the parameters learned by the model during training do not exceed the empirical range of historical data.

[0069] During training, the main controller calculates a loss value for each sample in the labeled training dataset. The loss value consists of two parts. The Bellman error term is calculated by inputting the sample's historical state data into the Q-value evaluation function to obtain the predicted Q-value, and simultaneously calculating the target Q-value based on the sample's reward value and subsequent state. The squared difference between the two constitutes the Bellman error. This error term measures the accuracy of the model's predictions.

[0070] The calculation of the conservative constraint term is slightly more complex. The master controller randomly samples several sets of candidate parameters within the safety parameter space, calculates the Q-values ​​of these candidate parameters in the current state, and averages them to obtain the Q-value within the space. Simultaneously, it also randomly samples several sets of candidate parameters outside the safety parameter space, calculates their Q-values, and averages them to obtain the Q-value outside the space. The conservative constraint term aims for the Q-value within the space to be higher than the Q-value outside the space; if a Q-value outside the space is higher than the Q-value within the space, a penalty is imposed. This constraint mechanism encourages the model to tend to select parameters within the safety parameter space.

[0071] The main controller weights and sums the Bellman error term and the conservative constraint term to obtain the total loss value. The gradient of the loss value with respect to the network parameters is calculated using gradient descent, and the network parameters are updated according to the gradient direction. This process is repeated, decreasing the loss value with each iteration. When the change in the loss value is less than a preset convergence threshold, such as a loss value change rate of less than 0.1% over multiple iterations, the model is considered to have converged, and the Q-value evaluation function for training is obtained.

[0072] 207. Based on the welding process parameters, the motion controller generates control commands and sends the control commands to the soldering shaft unit to obtain the adjusted execution parameters of the soldering shaft unit; In this embodiment, step 207 is similar to step 103 in the first embodiment, and will not be described again here.

[0073] 208. After welding is completed, the weld point is detected by the vision sensor to obtain weld point quality evaluation data, and the current status data, the welding process parameters and the weld point quality evaluation data are stored in the main controller for subsequent inference calculation optimization.

[0074] In this embodiment, after storing the current state data, the welding process parameters, and the weld quality evaluation data in the main controller, the method further includes: determining whether the number of welding records stored in the main controller has reached a preset update threshold; when the number of welding records reaches the update threshold, extracting newly added welding records to construct an incremental training dataset; performing incremental training on the Q-value evaluation function based on the incremental training dataset, updating the network parameters of the Q-value evaluation function, and obtaining the updated Q-value evaluation function.

[0075] Specifically, multi-axis soldering machines continuously accumulate soldering records during actual production. The main controller can maintain a counter to record the number of new soldering records since the last model update. This counter automatically increments after each soldering operation is completed and the data is stored. When the counter reaches a preset update threshold, such as accumulating 500 or 1000 new records, the main controller triggers the model update process.

[0076] Setting the update threshold requires balancing the update frequency and the amount of data. If the threshold is set too low, such as updating the model every 10 records, updates will be too frequent, increasing the computational burden while the amount of data updated each time is too small, resulting in limited effectiveness. If the threshold is set too high, such as requiring 10,000 records to accumulate before updating, although the data volume is sufficient, the update cycle is too long, and the model cannot adapt to changes in operating conditions in a timely manner. Therefore, the update threshold is usually set based on the actual production rhythm and the equipment's computing power.

[0077] Once the number of welding records reaches the update threshold, the main controller retrieves these newly added welding records from the storage unit. These records constitute the incremental training dataset, containing the latest welding conditions and parameter selection information. Compared to the initial offline training dataset, the incremental training dataset reflects the recent operating status of the equipment and may include new operating condition characteristics or parameter trends.

[0078] The main controller incrementally trains the Q-value evaluation function based on the incremental training dataset. Incremental training is similar to the initial training process, but instead of training the network from scratch, it fine-tunes the existing network parameters. The main controller inputs incremental data samples into the Q-value evaluation function, calculates the loss value, and updates the network parameters using gradient descent. Since the network parameters have already learned the basic parameter selection strategy during initial training, incremental training converges with only a few iterations. After incremental training, the updated Q-value evaluation function is obtained, which replaces the original function in subsequent inference calculations.

[0079] In this embodiment, temperature sensors, vision sensors, and environmental sensors collect current solder head temperature, solder joint position information, PCB board feature information, and environmental parameters to obtain current status data. The main controller performs inference calculations based on the current status data to obtain soldering process parameters for the current solder joint. The motion controller generates control commands based on the soldering process parameters and sends them to the solder axis unit to obtain adjusted execution parameters. After soldering is completed, the solder joint is inspected by a vision sensor to obtain solder joint quality evaluation data. The status data, process parameters, and quality evaluation data are stored for subsequent optimization. This invention can dynamically adjust soldering parameters according to real-time operating conditions, improving the stability of soldering quality.

[0080] The adaptive control method for a multi-axis soldering machine in this embodiment of the invention has been described above. The adaptive control device for a multi-axis soldering machine in this embodiment of the invention is described below. The multi-axis soldering machine includes a soldering axis unit, a temperature sensor, a vision sensor, a motion controller, an environmental sensor, and a main controller. For details on the adaptive control device for this multi-axis soldering machine, please refer to [link to relevant documentation]. Figure 3 One embodiment of the adaptive control device for a multi-axis soldering machine in this invention includes: The data acquisition module 301 is used to acquire the real-time temperature of the current solder head through the temperature sensor, acquire the position information of the solder joint to be soldered and the feature information of the PCB board through the vision sensor, and acquire the ambient temperature and humidity parameters through the environmental sensor to obtain the current status data. The parameter reasoning module 302 is used to perform reasoning calculations by the main controller based on the current state data to obtain welding process parameters for the current weld point; The instruction generation module 303 is used to generate control instructions from the motion controller based on the welding process parameters, and send the control instructions to the soldering shaft unit respectively to obtain the adjusted execution parameters of the soldering shaft unit; The quality feedback module 304 is used to detect the weld point through the vision sensor after welding is completed, obtain weld point quality evaluation data, and store the current status data, the welding process parameters and the weld point quality evaluation data to the main controller.

[0081] In this embodiment of the invention, the adaptive control device of the multi-axis soldering machine operates the aforementioned adaptive control method for the multi-axis soldering machine. The adaptive control device collects current solder head temperature, solder joint position information, PCB board feature information, and environmental parameters through temperature sensors, vision sensors, and environmental sensors to obtain current state data. The main controller performs inference calculations based on the current state data to obtain soldering process parameters for the current solder joint. The motion controller generates control commands based on the soldering process parameters and sends them to the soldering axis unit to obtain adjusted execution parameters. After soldering is completed, the solder joint is detected by a vision sensor to obtain solder joint quality evaluation data. The state data, process parameters, and quality evaluation data are stored for subsequent optimization. This invention can dynamically adjust soldering parameters according to real-time operating conditions, improving the stability of soldering quality.

[0082] above Figure 3 The adaptive control device of the multi-axis soldering machine in the embodiments of the present invention will be described in detail from the perspective of unitized functional entities. The adaptive control equipment of the multi-axis soldering machine in the embodiments of the present invention will be described in detail from the perspective of hardware processing.

[0083] Figure 4This is a schematic diagram of the structure of an adaptive control device for a multi-axis soldering machine according to an embodiment of the present invention. The adaptive control device 400 for the multi-axis soldering machine can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 410 (e.g., one or more processors) and a memory 420, and one or more storage media 430 (e.g., one or more mass storage devices) storing application programs 433 or data 432. The memory 420 and storage media 430 can be temporary or persistent storage. The program stored in the storage media 430 may include one or more units (not shown in the diagram), each unit may include a series of instruction operations on the adaptive control device 400 of the multi-axis soldering machine. Furthermore, the processor 410 may be configured to communicate with the storage media 430 and execute the series of instruction operations in the storage media 430 on the adaptive control device 400 of the multi-axis soldering machine to implement the steps of the adaptive control method for the multi-axis soldering machine described above.

[0084] The adaptive control device 400 of the multi-axis soldering machine may also include one or more power supplies 440, one or more wired or wireless network interfaces 450, one or more input / output interfaces 460, and / or one or more operating systems 431, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 4 The adaptive control device structure of the multi-axis soldering machine shown does not constitute a limitation on the adaptive control device of the multi-axis soldering machine provided by the present invention. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0085] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the adaptive control method of the multi-axis soldering machine.

[0086] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0087] 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 the present invention, 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 the present invention. 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.

[0088] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention 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 the present invention.

Claims

1. An adaptive control method for a multi-axis soldering machine, characterized in that, The multi-axis soldering machine includes a soldering axis unit, a temperature sensor, a vision sensor, a motion controller, an environmental sensor, and a main controller. The adaptive control method includes: The current status data is obtained by acquiring the real-time temperature of the solder head through the temperature sensor, the position information of the solder joint to be soldered and the feature information of the PCB board through the vision sensor, and the ambient temperature and humidity parameters through the environmental sensor. Based on the current status data, the main controller performs inference calculations to obtain the welding process parameters for the current weld point; Based on the welding process parameters, the motion controller generates control commands and sends the control commands to the soldering shaft unit to obtain the adjusted execution parameters of the soldering shaft unit. After welding is completed, the weld point is detected by the vision sensor to obtain weld point quality evaluation data. The current status data, the welding process parameters, and the weld point quality evaluation data are stored in the main controller for subsequent inference calculation optimization.

2. The adaptive control method according to claim 1, characterized in that, The step of obtaining welding process parameters for the current weld point by inference calculation performed by the main controller based on the current state data includes: The real-time temperature, location information, feature information, and environmental parameters in the current state data are normalized to obtain a standardized state vector; The standardized state vector is input into a pre-trained Q-value evaluation function, and the corresponding value score is calculated for each of the preset candidate welding process parameters. Based on the value score, the candidate welding process parameter with the largest Q value is selected as the initial welding process parameter; Calculate the deviation between the preliminary welding process parameters and the parameter distribution center of historical successful welding samples; When the deviation exceeds the preset safety threshold, the preliminary welding process parameters are adjusted along the gradient direction to the boundary range of the parameter distribution to obtain the welding process parameters for the current weld point. When the deviation does not exceed the preset safety threshold, the preliminary welding process parameters are used as the welding process parameters for the current weld point.

3. The adaptive control method according to claim 2, characterized in that, Before the main controller performs inference calculations based on the current state data to obtain the welding process parameters for the current weld point, the method further includes: Historical welding records are extracted from the storage unit of the main controller. Each record includes historical status data, corresponding welding process parameters, and weld quality evaluation data to construct an offline training dataset. A safety parameter space for welding process parameters is established based on the offline training dataset. The safety parameter space is used to constrain the parameter range during the training process. The network parameters of the Q-value evaluation function are initialized, and the network parameters are iteratively updated by minimizing the loss function to obtain the trained Q-value evaluation function.

4. The adaptive control method according to claim 3, characterized in that, The network parameters for initializing the Q-value evaluation function are iteratively updated by minimizing the loss function to obtain the trained Q-value evaluation function, which includes: Initialize the network parameters of the Q-value evaluation function, and perform data cleaning and reward labeling on the offline training dataset. Remove unqualified samples and calculate reward values ​​for the remaining samples to obtain the labeled training dataset. Based on the labeled training dataset, the distribution range of welding process parameters is extracted, and the boundary of the safety parameter space is constructed. For each training sample in the labeled training dataset, a loss value including a Bellman error term and a conservative constraint term is calculated, wherein the Bellman error term is obtained by comparing the predicted Q value with the target Q value, and the conservative constraint term is obtained by comparing the candidate parameter Q values ​​inside and outside the safety parameter space. The network parameters are iteratively updated based on the loss value using gradient descent until the loss value converges, thus obtaining the Q-value evaluation function after training is complete.

5. The adaptive control method according to claim 1, characterized in that, Each soldering shaft unit includes a solder head, a heater, a solder feeding mechanism, and a drive motor; The motion controller generates control commands based on the welding process parameters and sends these commands to the corresponding soldering axis units to obtain adjusted execution parameters for each soldering axis unit, including: Based on the welding process parameters, determine whether each parameter is within the preset safety range, and perform boundary correction on parameters that exceed the range to obtain the welding process parameters after safety verification. Based on the position information of the solder joint to be soldered and the current position of the solder axis unit, the motion trajectory of each motion axis from the current position to the target solder joint is planned, and the motion speed of each motion axis is calculated based on the welding process parameters after safety verification, so as to obtain a multi-axis linkage motion scheme. Based on the multi-axis linkage motion scheme and the welding process parameters after safety verification, control signals for the heater, solder feeding mechanism and each motion axis are generated to obtain hardware control instructions. The motion controller sends the hardware control command to the solder axis unit, controls each motion axis to move in coordination to the target solder joint and perform the soldering operation, and obtains the adjusted execution parameters of the solder axis unit.

6. The adaptive control method according to claim 5, characterized in that, The process of planning the motion trajectory of each motion axis from its current position to the target solder joint based on the position information of the solder joint to be soldered and the current position of the solder axis unit, and calculating the motion speed of each motion axis based on the welding process parameters after safety verification, to obtain a multi-axis linkage motion scheme includes: Based on the position information of the solder joint to be soldered and the current position of the solder shaft unit, the displacement from the current position to the target position is calculated to obtain the motion path parameters; Based on the motion path parameters, the motion trajectory of each motion axis is generated to obtain the multi-axis motion trajectory; Based on the time parameter in the welding process parameters after the safety verification and the multi-axis motion trajectory, the motion speed of each motion axis is calculated to obtain the target speed of each motion axis. Based on the target speed, smooth acceleration and deceleration curves are generated for each motion axis to obtain a multi-axis linkage motion scheme.

7. The adaptive control method according to claim 2, characterized in that, After storing the current status data, the welding process parameters, and the weld quality evaluation data into the main controller, the method further includes: Determine whether the number of welding records stored in the main controller has reached a preset update threshold; When the number of welding records reaches the update threshold, the newly added welding records are extracted to construct an incremental training dataset; The Q-value evaluation function is incrementally trained based on the incremental training dataset, and the network parameters of the Q-value evaluation function are updated to obtain the updated Q-value evaluation function.

8. An adaptive control device for a multi-axis soldering machine, characterized in that, The multi-axis soldering machine includes a soldering axis unit, a temperature sensor, a vision sensor, a motion controller, an environmental sensor, and a main controller. The adaptive control device of the multi-axis soldering machine includes: The data acquisition module is used to acquire the real-time temperature of the current solder head through the temperature sensor, acquire the position information of the solder joint to be soldered and the feature information of the PCB board through the vision sensor, and acquire the ambient temperature and humidity parameters through the environmental sensor to obtain the current status data. The parameter reasoning module is used to perform reasoning calculations by the main controller based on the current state data to obtain the welding process parameters for the current weld point. The instruction generation module is used to generate control instructions from the motion controller based on the welding process parameters, and send the control instructions to the soldering shaft unit respectively to obtain the adjusted execution parameters of the soldering shaft unit; The quality feedback module is used to detect the weld joint through the vision sensor after welding is completed, obtain weld joint quality evaluation data, and store the current status data, welding process parameters and weld joint quality evaluation data in the main controller for subsequent inference calculation optimization.

9. An adaptive control device for a multi-axis soldering machine, characterized in that, The adaptive control device of the multi-axis soldering machine includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the adaptive control device of the multi-axis soldering machine to perform the steps of the adaptive control method of the multi-axis soldering machine as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the steps of the adaptive control method for the multi-axis soldering machine as described in any one of claims 1-7.

Images