Adaptive feedback temperature control method and system for laser soldering

Abstract

The present invention relates to an adaptive feedback temperature control method and system for laser soldering, belonging to the field of welding control. The method comprises: establishing a simulation thermodynamic mathematical model based on a lumped parameter method; introducing a parameter error learning value to adjust the simulation thermodynamic mathematical model; establishing a nonlinear system with the parameter error learning value as input and the difference between the actual solder point temperature and the model simulation temperature as output, designing a parameter error learning expression to reconstruct a thermodynamic mathematical model based on learning support; calculating a compensation control quantity based on learning support; defining a residual modeling error of the laser soldering process based on the actual solder point temperature, the ambient temperature, and the residual parameter error between the parameter error learning value and its ideal value, designing a calculation formula for the adaptive feedback control quantity, and finally obtaining a calculation formula for the total temperature control quantity to control the error between the solder point temperature and the target temperature. The present invention can effectively improve the accuracy of the model in describing temperature changes.

Description

Technical Field

[0001] The present invention relates to the field of welding control, and in particular to a learning-supported adaptive feedback temperature control method and system during the welding process of laser soft soldering equipment. Background Art

[0002] With the rapid development of the microelectronics industry, microelectronic components are trending towards miniaturization, structural complexity, and functional integration. High-density and high-reliability microelectronic assembly technologies have become a worthy research topic. Laser soldering, among other things, has attracted considerable attention due to its precision, controllable properties, rapid heating rate, and minimal heat-affected zone. As a surface mount technology with broad development prospects, laser soldering enables precise welding of tiny and delicate components without affecting surrounding components. This technology uses a high-intensity laser beam as a heat source for efficient welding. During the welding process, the laser beam is focused on the weld point. When the temperature reaches the soldering temperature, the solder material melts and wets the surfaces to be joined. To reduce the difficulty of soldering, a filler metal is typically used, which has a low melting point and can be completed at a lower temperature. Temperature control during laser soldering is crucial to product quality. Excessively high temperatures can cause the filler metal to melt too quickly, resulting in an incomplete weld or a fusion weld. Excessively low temperatures can create voids within the resulting solder joint. Therefore, it is necessary to study how to effectively reduce or eliminate the error between the solder temperature and the target temperature.

[0003] Designing a high-precision temperature control algorithm for laser soldering systems has always been a challenging technical problem due to the inaccuracy of mathematical models and the large disturbances they cause. Currently, the most common temperature control method is a well-tuned PID control method, but this method suffers from poor control accuracy and poor anti-interference capabilities. Summary of the Invention

[0004] To address the challenges of existing technologies, this invention provides an adaptive feedback temperature control method for laser soldering. The online learning algorithm employed in this method effectively improves the accuracy of the existing thermodynamic model in describing temperature changes, thereby enabling the design of an effective model-based closed-loop control method for temperature control. Furthermore, the adaptive feedback control designed in this invention dynamically adjusts the control gain based on the actual system conditions. Compared to existing, well-tuned PID control algorithms, this method offers superior response speed and control accuracy in temperature control, and is robust to system uncertainties and external disturbances.

[0005] The adaptive feedback temperature control method for laser soldering adopted in an embodiment of the present invention includes the following steps:

[0006] S1. Based on the lumped parameter method, a thermodynamic mathematical model for the simulation of the laser soldering heating process is established;

[0007] S2. Introduce a parameter error learning value, adjust the simulated thermodynamic mathematical model, and obtain the actual thermodynamic mathematical model; establish a nonlinear system with the parameter error learning value as input and the difference between the actual solder point temperature and the model simulation temperature as output, and design an online learning algorithm for adjusting the parameter error learning value based on the principle of P-type iterative learning control. Obtain a parameter error learning expression to reconstruct the thermodynamic mathematical model based on learning support;

[0008] S3. Calculate a learning-supported compensation control variable based on the target temperature and the simulated thermodynamic mathematical model after introducing the parameter error learning value; define the residual modeling error of the laser soldering process based on the actual solder point temperature, ambient temperature, and the residual parameter error between the parameter error learning value and its ideal value; and design a calculation formula for the adaptive feedback control variable based on the actual thermodynamic mathematical model and the calculation formula for the learning-supported compensation control variable; derive a calculation formula for the total temperature control variable based on the learning-supported compensation control variable and the adaptive feedback control variable;

[0009] S4. In each sampling cycle, based on the target temperature and the solder point temperature measured in the current cycle, the temperature control input value of the current cycle is calculated by the calculation formula of the total temperature control amount to control the error between the solder point temperature and the target temperature.

[0010] The adaptive feedback temperature control system for laser soldering adopted in the embodiment of the present invention is characterized by comprising the following modules:

[0011] A simulation model building module is used to establish a simulation thermodynamic mathematical model of the laser soldering heating process based on the lumped parameter method;

[0012] The model reconstruction module is used to introduce parameter error learning values, adjust the simulated thermodynamic mathematical model, and obtain the actual thermodynamic mathematical model. A nonlinear system is established with the parameter error learning value as input and the difference between the actual solder point temperature and the model simulation temperature as output. An online learning algorithm for adjusting the parameter error learning value is designed based on the principle of P-type iterative learning control. The parameter error learning expression is obtained to reconstruct the thermodynamic mathematical model based on learning support.

[0013] The control variable calculation formula design module is used to calculate the learning-supported compensation control variable based on the target temperature and the simulated thermodynamic mathematical model after the parameter error learning value is introduced. The residual modeling error of the laser soldering process is defined based on the actual solder point temperature, ambient temperature, and the residual parameter error between the parameter error learning value and its ideal value. The calculation formula of the adaptive feedback control variable is designed based on the actual thermodynamic mathematical model and the calculation formula of the learning-supported compensation control variable. The calculation formula of the total temperature control variable is obtained based on the learning-supported compensation control variable and the adaptive feedback control variable.

[0014] The control quantity calculation module is used to calculate the temperature control input value of the current cycle in each sampling cycle based on the target temperature and the solder point temperature measured in the current cycle through the calculation formula of the total temperature control quantity to control the error between the solder point temperature and the target temperature.

[0015] Compared with the prior art, the present invention has the following beneficial effects:

[0016] 1. In response to the problems in the prior art where the mathematical model is inaccurate and the welding process is sensitive to disturbances, the adaptive feedback temperature control method proposed in the present invention is based on the principle of P-type iterative learning control. By introducing the error learning value of the time-varying parameter on the basis of the existing thermodynamic model, an online learning algorithm for the time-varying parameter error is designed to periodically obtain more accurate time-varying parameters, thereby correcting the mathematical model through online learning and improving its accuracy. To a certain extent, the accuracy of the model in describing the temperature changes during the heating process is improved, thereby improving the control effect of the model feedforward and the control effect of the model-based closed-loop control algorithm.

[0017] 2. The adaptive feedback control algorithm in the temperature control method of the present invention can dynamically adjust the control gain according to the actual situation during the soft soldering process, control the impact of residual modeling errors and external disturbances, improve the robustness of the system to disturbances, and further reduce the error between the actual temperature and the target temperature.

[0018] 3. Compared with the existing PID control method, the present invention has obvious improvements in control accuracy and response speed, and can keep the temperature error in the welding process within a high-precision range. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a schematic diagram of the laser soldering system principle;

[0020] Figure 2 This is a control principle diagram of the adaptive feedback temperature control method according to an embodiment of the present invention;

[0021] Figure 3 The pad radius is 2×10 -3m. Comparison of the experimental temperature curve at a laser power of 8.2 W, the temperature simulation curve of the learning-supported thermodynamic model, and the temperature simulation curve of the original thermodynamic model;

[0022] Figure 4 The pad radius is 1.5×10 -3 m, a temperature error comparison curve of the control method of the present invention and a well-tuned PID control method;

[0023] Figure 5 The pad radius is 2×10 -3 m, a temperature error comparison curve of the control method of the present invention and a well-tuned PID control method;

[0024] Figure 6 The pad radius is 2.5×10 -3 m, a temperature error comparison curve of the control method of the present invention and a well-tuned PID control method. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0026] like Figure 1 As shown, the laser soldering equipment used in the present invention mainly includes modules such as a laser emitter, an infrared temperature measurement system, an error detection system, a control system, an observation system and an optical focusing system. The movement of the workbench is controlled by a motor controller so that the welding spot of the workpiece on the workbench is moved to the bottom of the laser emitter, and then the defocus amount is adjusted and the laser is turned on so that a beam of semiconductor laser is focused on the welding spot through the optical fiber and the focusing system. The actual welding spot temperature is periodically measured by an infrared temperature measurement system (such as an infrared thermometer). The control system outputs a control signal to adjust the laser power according to the error between the actual welding spot temperature and the target temperature, and drives the laser emitter to ensure that the solder paste reaches the target temperature. After the solder paste melts, the welding process is completed. The workbench places the sample plate under the infrared thermometer to detect each position on the surface of the pad. Data acquisition, platform positioning and visual systems are all controlled by a computer, providing a complete system.

[0027] That is, during the laser soldering heating process, the solder paste is heated and melted by the laser beam. The infrared temperature measurement system measures the solder point temperature during each unit cycle, and the error detection system determines whether there is an error between the temperature measurement result of the current cycle and the target temperature. If there is an error, the control system calculates the laser power value based on this error, combined with the temperature measurement data of the current cycle and the target temperature, based on the control method proposed by the present invention, and outputs a corresponding control signal to adjust the laser power, driving the laser generator to heat while measuring the solder point temperature data for the next unit cycle. For each subsequent unit cycle, the error detection system determines whether there is an error between the measured solder point temperature and the target temperature of the corresponding cycle. If there is an error, the above method is used to eliminate the error between the solder point temperature and the target temperature.

[0028] like Figure 2 As shown, the learning-supported adaptive feedback temperature control method for laser soldering in this embodiment includes the following steps:

[0029] S1. Based on the lumped parameter method, a simulation thermodynamic mathematical model of the laser soldering heating process is established.

[0030] This step is based on a lumped parameter thermodynamic model that describes the laser soldering heating process, with laser power as the system input and solder point temperature as the system output. The relationship between input and output is determined to achieve the construction of a simulation thermodynamic mathematical model. The expression of the constructed model is preferably:

[0031]

[0032] Where t is time, P is laser power, α is the absorption rate of solder paste to laser, m is the mass of solder paste, c is the specific heat capacity of solder paste, A is the surface area of ​​solder paste, h is the surface heat transfer coefficient, and T0 is the ambient temperature. To distinguish it from other temperature parameters described later, T m is the simulation temperature of the model. To simplify the subsequent formula expression, define Then the simulation thermodynamic mathematical model is rewritten as:

[0033]

[0034] S2. Introduce the parameter error learning value, adjust the simulated thermodynamic mathematical model, and obtain the actual thermodynamic mathematical model; establish a nonlinear system with the parameter error learning value as input and the difference between the actual solder point temperature and the model simulation temperature as output, and design an online learning algorithm for adjusting the parameter error learning value based on the principle of P-type iterative learning control, and obtain the parameter error learning expression to reconstruct the thermodynamic mathematical model based on learning support.

[0035] Specifically, this step further includes:

[0036] S21. Rewrite the time-varying parameter form in the simulation thermodynamic mathematical model established in step S1 into a summed value consisting of an initial value and a parameter error learning value, define the summed value as an estimate of the true value of the time-varying parameter in the actual project, and adjust the simulation thermodynamic mathematical model.

[0037] The time-varying parameter values ​​in thermodynamic mathematical models are generally difficult to obtain accurately, which can lead to poor control performance of the feedforward compensation control method. Therefore, to accurately obtain the time-varying parameter values ​​in the model, the time-varying parameters in the model are rewritten as the sum of the initial value and the parameter error learning value. The expression is as follows:

[0038]

[0039] where b * and a * are the initial values ​​of b and a respectively, and the corresponding calculation formulas are and Δb and Δa are the parameter error learning values ​​of b and a respectively, and b is defined as * +Δb and a * +Δa are estimates of the true values ​​of b and a in actual engineering.

[0040] Since the main components of the solder paste have been determined, the value of the solder paste's specific heat capacity and its laser absorptivity can be accurately determined. This makes the parameter b in the model * The value of can be directly equivalent to the actual value of the parameter in actual engineering, that is, Δb = 0, then the model expressed by formula (3) can be adjusted as follows:

[0041]

[0042] Based on formula (4), the actual thermodynamic relationship of solder paste (i.e., the actual thermodynamic mathematical model) can be expressed as follows:

[0043]

[0044] Where T p is the actual solder point temperature, e a is the ideal value of the parameter error learning value Δa, a * +e a This is the actual value of parameter a.

[0045] S22. The actual solder point temperature during the welding process is subtracted from the model simulation temperature. Based on the principle of P-type iterative learning control, an online learning algorithm that can periodically learn the parameter error value is designed to obtain the parameter error learning expression and realize the reconstruction of the thermodynamic mathematical model based on learning support.

[0046] This step defines T e =T p -T m , where T e is the difference between the actual solder point temperature and the model simulation temperature. Combining equations (4) and (5), Expressed as:

[0047]

[0048] Formula (6) can be regarded as taking the parameter error learning value Δa as input, and the difference T e In this nonlinear system, the error learning value Δa can be adjusted to make the difference T e tends to 0. To this end, based on the principle of P-type iterative learning control, the online learning algorithm for adjusting Δa can be designed as follows:

[0049] Δa (k) =Δa (k-1) +K p T e(k -1) (7)

[0050] where K p is the learning gain of the P-type iterative learning algorithm, T e is the difference between the actual solder point temperature and the model simulation temperature, that is, T e =T p -T m The subscript (k) indicates that the corresponding parameter is the parameter calculated in the kth unit cycle. At this point, the reconstructed learning-supported thermodynamic mathematical model has been derived.

[0051] S3. Calculate the compensation control quantity based on learning support according to the target temperature and the simulated thermodynamic mathematical model after introducing the parameter error learning value; define the residual modeling error of the laser soft soldering process based on the actual solder joint temperature, ambient temperature, and the residual parameter error between the parameter error learning value and its ideal value, and design the calculation formula of the adaptive feedback control quantity according to the actual thermodynamic mathematical model and the calculation formula of the compensation control quantity based on learning support; obtain the calculation formula of the total temperature control quantity based on the compensation control quantity based on learning support and the adaptive feedback control quantity.

[0052] In this step, a reasonable target temperature curve is first designed to determine the target temperature based on the physical properties of the welding material, such as thermal conductivity and melting point, as well as actual engineering experience and the requirements of different stages of the heating process.

[0053] Then, based on the parameter learning expression and the target temperature curve, a learning-supported compensation control quantity calculation formula is designed; and according to the adaptive feedback control principle, an adaptive feedback control quantity calculation formula is designed. The sum of these two control quantity calculation formulas is the calculation formula for the total system temperature control quantity, which is used to control the target temperature of the actual solder point temperature tracking.

[0054] The calculation formula of the total temperature control amount of the system of the present invention consists of two parts, which are expressed as follows:

[0055] P=P c +P a (8)

[0056] Where P is the total control quantity of the system, P c is the compensation control quantity based on learning support, P a is the adaptive feedback control quantity. Define the target temperature as T d , the model simulation temperature T in (4) m Replace with T d , then the calculation formula of the compensation control amount based on learning support can be recursively deduced as follows:

[0057]

[0058] Because e a is the ideal value of the parameter error learning value Δa, so e a =Δa+ω must hold, where ω represents the residual parameter error. Combining equations (5), (8), and (9), we can obtain the following relationship:

[0059]

[0060] Define E = T p -T d , where E is the difference between the actual solder point temperature and the target temperature, and Φ is defined as -ω(T p -T0), Φ is the residual modeling error of the laser soldering process. Then (10) can be rewritten as:

[0061]

[0062] Based on the above derivation process, this embodiment can adopt an adaptive feedback control method to eliminate the residual modeling error, wherein the calculation formula of the adaptive feedback control amount can be designed as follows:

[0063]

[0064] Where β is the gain of the adaptive feedback control algorithm, and λ is a positive constant.

[0065] S4. In each sampling cycle, based on the target temperature and the solder point temperature measured in the current cycle, the temperature control input value of the current cycle is calculated using the calculation formula of the total temperature control amount, so as to control the error between the solder point temperature and the target temperature so that the temperature error eventually approaches 0.

[0066] This embodiment, based on a thermodynamic model of the laser soldering heating process established using the lumped parameter method, decomposes the time-varying parameters in the model into two parts, namely the initial value and the parameter error learning value, wherein the parameter error learning value is much smaller than the initial value, and defines that the sum of the values ​​of the two parts is equal to the estimate of the actual value of the thermodynamic parameter. The difference between the model output and the actual output is obtained, and based on the P-type iterative learning control principle, an online learning algorithm is designed that can obtain the time-varying parameter error learning value in each sampling cycle, and a self-regulating closed-loop control calculation formula based on the model is designed to control the actual temperature in the welding process to the target temperature. In order to further reduce the error between the actual temperature and the target temperature, this embodiment designs an adaptive control calculation formula that can be used to control the residual modeling error according to the adaptive control principle. Through experiments and simulations, it is confirmed that this control method can effectively eliminate the temperature control error in the laser soldering heating process. Compared with the well-tuned PID control algorithm, this embodiment has better control accuracy and response speed, such as Figure 3-Figure 6 shown.

[0067] In addition, this embodiment also provides an adaptive feedback temperature control system for laser soldering, which specifically includes the following modules:

[0068] A simulation model building module is used to establish a simulation thermodynamic mathematical model of the laser soldering heating process based on the lumped parameter method;

[0069] The model reconstruction module is used to introduce parameter error learning values, adjust the simulated thermodynamic mathematical model, and obtain the actual thermodynamic mathematical model. A nonlinear system is established with the parameter error learning value as input and the difference between the actual solder point temperature and the model simulation temperature as output. An online learning algorithm for adjusting the parameter error learning value is designed based on the principle of P-type iterative learning control. The parameter error learning expression is obtained to reconstruct the thermodynamic mathematical model based on learning support.

[0070] The control variable calculation formula design module is used to calculate the learning-supported compensation control variable based on the target temperature and the simulated thermodynamic mathematical model after the parameter error learning value is introduced. The residual modeling error of the laser soldering process is defined based on the actual solder point temperature, ambient temperature, and the residual parameter error between the parameter error learning value and its ideal value. The calculation formula of the adaptive feedback control variable is designed based on the actual thermodynamic mathematical model and the calculation formula of the learning-supported compensation control variable. The calculation formula of the total temperature control variable is obtained based on the learning-supported compensation control variable and the adaptive feedback control variable.

[0071] The control quantity calculation module is used to calculate the temperature control input value of the current cycle in each sampling cycle based on the target temperature and the solder point temperature measured in the current cycle through the calculation formula of the total temperature control quantity to control the error between the solder point temperature and the target temperature.

[0072] The above modules are respectively used to implement corresponding steps of the temperature control method of this embodiment. The detailed implementation process thereof is shown in steps S1-S4.

[0073] It can be seen that the temperature control method during the welding process of the laser soldering equipment of the present invention is based on the thermodynamic mathematical model established using the lumped parameter method, rewriting the time-varying parameters into the form of the sum of the initial value and the parameter error learning value, and designing an online learning algorithm that can periodically learn the parameter error value based on the P-type iterative learning control algorithm, so as to improve the accuracy of the thermodynamic mathematical model in describing temperature changes and improve the control effect of the model-based compensation control method. As for the residual modeling error with time-varying properties, the present invention adopts an adaptive feedback control algorithm that can dynamically adjust the control gain for control, so as to further reduce the error between the actual solder joint temperature and the target temperature. The control method is simple to implement, has a fast response speed, and high control accuracy. It can effectively eliminate the temperature error during the laser soldering heating process, and the method has a certain degree of robustness to unknown random disturbances, and can be widely used in the laser soldering equipment welding process.

[0074] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An adaptive feedback temperature control method for laser soldering, characterized in that: The following steps are involved: S1. Based on the lumped parameter method, a thermodynamic mathematical model for the simulation of the laser soldering heating process is established; S2. Introduce a parameter error learning value, adjust the simulated thermodynamic mathematical model, and obtain the actual thermodynamic mathematical model; establish a nonlinear system with the parameter error learning value as input and the difference between the actual solder point temperature and the model simulation temperature as output, and design an online learning algorithm for adjusting the parameter error learning value based on the principle of P-type iterative learning control. Obtain a parameter error learning expression to reconstruct the thermodynamic mathematical model based on learning support; S3. Calculating a learning-supported compensation control amount based on the target temperature and the simulated thermodynamic mathematical model after introducing the parameter error learning value; Based on the actual solder point temperature, ambient temperature, and the residual parameter error between the parameter error learning value and its ideal value, the residual modeling error of the laser soldering process is defined. Based on the actual thermodynamic mathematical model and the calculation formula of the compensation control quantity based on learning support, the calculation formula of the adaptive feedback control quantity is designed; According to the compensation control quantity based on learning support and the adaptive feedback control quantity, the calculation formula of the total temperature control quantity is obtained; S4. In each sampling cycle, based on the target temperature and the solder point temperature measured in the current cycle, the temperature control input value of the current cycle is calculated by the calculation formula of the total temperature control amount to control the error between the solder point temperature and the target temperature.

2. The adaptive feedback temperature control method according to claim 1, characterized in that: Step S1 is based on the lumped parameter thermodynamic model describing the laser soldering heating process, takes the laser power as the system input and the solder point temperature as the system output, determines the relationship between the input and output, and constructs a simulation thermodynamic mathematical model.

3. The adaptive feedback temperature control method according to claim 1, characterized in that: Step S2 includes: S21, rewriting the time-varying parameter form in the simulation thermodynamic mathematical model into a summed value consisting of an initial value and a parameter error learning value, defining the summed value as an estimate of the true value of the time-varying parameter in the actual project, and adjusting the simulation thermodynamic mathematical model; S22. The actual solder point temperature during the welding process is subtracted from the model simulation temperature. Based on the principle of P-type iterative learning control, an online learning algorithm that can periodically learn the parameter error value is designed to obtain the parameter error learning expression and realize the reconstruction of the thermodynamic mathematical model based on learning support.

4. The adaptive feedback temperature control method according to claim 3, characterized in that: The simulation thermodynamic mathematical model constructed in step S1 is: Where t is time, P is laser power, α is the absorption rate of solder paste to laser, m is the mass of solder paste, c is the specific heat capacity of solder paste, A is the surface area of ​​solder paste, h is the surface heat transfer coefficient, T0 is the ambient temperature, T m Simulate temperature for the model; definition The simulation thermodynamic mathematical model after adjustment in step S21 is: where b * and a * are the initial values ​​of b and a respectively, and the corresponding calculation formulas are and Δb and Δa are the parameter error learning values ​​of b and a respectively, and b is defined as * +Δb and a * +Δa are the estimates of the true values ​​of b and a in actual engineering; When the main components of solder paste are determined, parameter b * The value of is equivalent to the actual value of this parameter in actual engineering, Δb=0, and the simulation thermodynamic mathematical model is further adjusted to: The actual thermodynamic mathematical model is expressed as: Where T p is the actual solder point temperature, e a is the ideal value of the parameter error learning value Δa, a * +e a That is the actual value of parameter a; In step S22, T is defined e =T p -T m , where T e The difference between the actual solder point temperature and the model simulation temperature is expressed as follows: In the nonlinear system, the difference T is adjusted by adjusting the parameter error learning value Δa. e tends to 0; The parameter error learning expression is designed as: Δa (k) =Δa (k-1) +K p T e(k-1) where K p is the learning gain of the P-type iterative learning algorithm. The subscript k represents that the corresponding parameter is the parameter calculated in the k-th unit cycle.

5. The adaptive feedback temperature control method according to claim 4, characterized in that: In step S3, the target temperature is defined as T d According to the simulation thermodynamic mathematical model established in step S1, the compensation control quantity P based on learning support is c The calculation formula is recursively as follows: Assume e a =Δa+ω, where ω represents the residual parameter error, and the following relationship is derived: Define E = T p -T d , where E is the difference between the actual solder point temperature and the target temperature, and Φ is defined as -ω(T p -T0), Φ is the residual modeling error of the laser soldering process, then: The adaptive feedback control quantity P a The calculation formula is designed as: Where β is the gain of the adaptive feedback control algorithm, and λ is a positive constant.

6. The adaptive feedback temperature control method according to claim 1, characterized in that: Step S3 designs a reasonable target temperature curve to set the target temperature based on the physical properties of the welding material, such as thermal conductivity and melting point, as well as actual engineering experience and the requirements of different stages of the heating process.

7. An adaptive feedback temperature control system for laser soldering, characterized in that: Includes the following modules: A simulation model building module is used to establish a simulation thermodynamic mathematical model of the laser soldering heating process based on the lumped parameter method; The model reconstruction module is used to introduce parameter error learning values, adjust the simulated thermodynamic mathematical model, and obtain the actual thermodynamic mathematical model. A nonlinear system is established with the parameter error learning value as input and the difference between the actual solder point temperature and the model simulation temperature as output. An online learning algorithm for adjusting the parameter error learning value is designed based on the principle of P-type iterative learning control. The parameter error learning expression is obtained to reconstruct the thermodynamic mathematical model based on learning support. The control quantity calculation formula design module is used to calculate the compensation control quantity based on learning support according to the target temperature and the simulation thermodynamic mathematical model after the parameter error learning value is introduced; Based on the actual solder point temperature, ambient temperature, and the residual parameter error between the parameter error learning value and its ideal value, the residual modeling error of the laser soldering process is defined. Based on the actual thermodynamic mathematical model and the calculation formula of the compensation control quantity based on learning support, the calculation formula of the adaptive feedback control quantity is designed; According to the compensation control quantity based on learning support and the adaptive feedback control quantity, the calculation formula of the total temperature control quantity is obtained; The control quantity calculation module is used to calculate the temperature control input value of the current cycle in each sampling cycle based on the target temperature and the solder point temperature measured in the current cycle through the calculation formula of the total temperature control quantity to control the error between the solder point temperature and the target temperature.

8. The adaptive feedback temperature control system according to claim 7, characterized in that: The reconstruction process of the model reconstruction module includes: Rewrite the time-varying parameter form in the simulation thermodynamic mathematical model into a summed value consisting of an initial value and a parameter error learning value, define the summed value as an estimate of the true value of the time-varying parameter in the actual project, and adjust the simulation thermodynamic mathematical model; The actual solder point temperature during the welding process is subtracted from the model simulation temperature. Based on the principle of P-type iterative learning control, an online learning algorithm that can periodically learn the parameter error value is designed to obtain the parameter error learning expression and realize the reconstruction of the thermodynamic mathematical model based on learning support.

9. The adaptive feedback temperature control system according to claim 8, characterized in that: The simulation thermodynamic mathematical model constructed by the simulation model construction module is: Where t is time, P is laser power, α is the absorption rate of solder paste to laser, m is the mass of solder paste, c is the specific heat capacity of solder paste, A is the surface area of ​​solder paste, h is the surface heat transfer coefficient, T0 is the ambient temperature, T m Simulate temperature for the model; definition The simulation thermodynamic mathematical model after adjustment by the model reconstruction module is: where b * and a * are the initial values ​​of b and a respectively, and the corresponding calculation formulas are and Δb and Δa are the parameter error learning values ​​of b and a respectively, and b is defined as * +Δb and a * +Δa are the estimates of the true values ​​of b and a in actual engineering; When the main components of solder paste are determined, parameter b * The value of is equivalent to the actual value of this parameter in actual engineering, Δb=0, and the simulation thermodynamic mathematical model is further adjusted to: The actual thermodynamic mathematical model is expressed as: Where T p is the actual solder point temperature, e a is the ideal value of the parameter error learning value Δa, a * +e a That is the actual value of parameter a; The model reconstruction module also defines T e =T p -T m , where T e The difference between the actual solder point temperature and the model simulation temperature is expressed as follows: In the nonlinear system, the difference T is adjusted by adjusting the parameter error learning value Δa. e tends to 0; The parameter error learning expression is designed as: Δa (k) =Δa (k-1) +K p T e(k-1) where K p is the learning gain of the P-type iterative learning algorithm. The subscript k represents that the corresponding parameter is the parameter calculated in the k-th unit cycle.

10. The adaptive feedback temperature control system according to claim 9, characterized in that: In the control quantity calculation design module, the target temperature is defined as T d According to the simulation thermodynamic mathematical model established by the simulation model building module, the compensation control quantity P based on learning support c The calculation formula is recursively as follows: Assume e a =Δa+ω, where ω represents the residual parameter error, and the following relationship is derived: Define E = T p -T d , where E is the difference between the actual solder point temperature and the target temperature, and Φ is defined as -ω(T p -T0), Φ is the residual modeling error of the laser soldering process, then: The adaptive feedback control quantity P a The calculation formula is designed as: Where β is the gain of the adaptive feedback control algorithm, and λ is a positive constant.

Images