Spring horizontal processing and single heat treatment coordinated intelligent control system

Abstract

The application relates to the field of spring manufacturing intelligent control, and discloses a spring horizontal processing and single heat treatment coordinated intelligent regulation and control system, which comprises the following steps: a disturbance characteristic online coding module, which is used for online acquisition and coding of physical disturbance generated by each spring in horizontal processing procedures such as coiling and welding, and generates a unique disturbance characteristic spectrum; an online control law dynamic calibration module, which receives the disturbance characteristic spectrum and inversely solves a compensatory individualized heat treatment instruction sequence for the spring; an individualized heat treatment execution module, which executes accurate single heat treatment on a single spring according to the instruction sequence; and a quality feedback and model self-optimization module, which optimizes a control model in a long period according to the final measured performance of the spring. Through construction of a direct mapping and feedback optimization closed loop from horizontal processing disturbance to single heat treatment control, the application realizes deep coordination between the two, effectively compensates for process fluctuation, and significantly improves the consistency of spring product quality.

Description

Technical Field

[0001] This invention relates to the field of intelligent control in spring manufacturing, specifically to an intelligent control system that coordinates horizontal spring processing with single heat treatment. Background Technology

[0002] As a key basic component, the consistency of the mechanical properties of springs is crucial to ensuring the overall performance and service life of equipment. The production process of springs usually includes horizontal processing steps such as rolling and welding, as well as individual heat treatment steps to give them the final properties.

[0003] Currently, due to the lack of an effective coordination mechanism between the two key stages of horizontal processing and single heat treatment, it is difficult to fundamentally improve the quality consistency of springs. In the rolling and welding process, random factors such as the micro-inhomogeneity of materials, instantaneous vibration of equipment, and progressive wear of tools will cause each semi-finished spring to have unique differences in geometric contours, internal residual stress, etc. Existing technologies cannot capture these disturbances in real time, let alone establish a digital disturbance file for each spring that can be used by downstream processes.

[0004] This lack of individual status information directly leads to the downstream heat treatment process only being able to adopt a crude control strategy that treats all springs with individual differences as if they were all subjected to the same pre-set process parameters. This control method cannot develop targeted compensation schemes to offset specific defects introduced from upstream, and existing heat treatment equipment (such as traditional tunnel furnaces) is physically unable to perform differentiated heat treatment procedures on individual springs on the production line.

[0005] Furthermore, the entire production process is a broken closed loop at the quality feedback level. The performance test data of the final product is usually limited to product sorting and scrapping, and cannot be effectively used to back-optimize the upstream control model. This makes the system lack the ability to learn and adapt from production experience, and it is difficult to cope with dynamic changes such as changes in raw material batches or long-term drift of equipment status. Therefore, the existing technology lacks the ability in three aspects: disturbance perception, personalized execution and adaptive learning, resulting in the failure of the coordination between horizontal processing and single heat treatment. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an intelligent control system that coordinates horizontal spring processing with single heat treatment. This system solves the problem that in existing spring manufacturing processes, processes such as horizontal rolling and laser welding introduce individualized and unpredictable complex physical disturbances, which prevent subsequent single heat treatment with uniform parameters from effectively eliminating the differentiated residual stress and microstructure inhomogeneity within each spring, thus leading to a decrease in the consistency of the final product's performance.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an intelligent control system for the coordinated horizontal machining and single heat treatment of springs, comprising: The online perturbation feature encoding module is configured to collect physical perturbation data in real time during the horizontal processing of springs, and generate a unique perturbation feature spectrum for each independent spring through signal processing and data fusion, which characterizes its composite perturbation history. The online control law dynamic calibration module is configured to generate an initial heat treatment command sequence by inversely solving based on the disturbance feature spectrum, and to perform online verification and calibration of the initial heat treatment command sequence through a built-in virtual spring object, and output a final heat treatment command sequence. A personalized heat treatment execution module is configured to receive the final heat treatment instruction sequence and control multiple independently addressable induction heating units within it to apply a non-uniform, personalized heat treatment in time and space to the passing spring. The quality feedback and model self-optimization module is configured to perform final performance testing on the spring after it has been processed by the personalized heat treatment execution module, and associate the test results with the corresponding disturbance feature spectrum to iteratively optimize at least one model built into the online control law dynamic calibration module.

[0008] Preferably, the online perturbation feature encoding module includes: The coiling disturbance acquisition unit is configured to acquire the three-dimensional contour data of the spring during the horizontal coiling process and extract the vertical displacement function caused by gravity. ,in The spring's unfolded length along its neutral layer; The welding disturbance acquisition unit is configured to acquire transient temperature field data at the welding point during the laser welding process of the spring. Harmony emission signal data ; Feature spectrum generation processor, configured to perform frequency domain analysis on the vertical displacement function to extract gravity disturbance feature vectors. The transient temperature field data and acoustic emission signal data are analyzed to extract welding disturbance feature vectors. Then, the gravity disturbance feature vector and the welding disturbance feature vector are fused to generate the disturbance feature spectrum.

[0009] In one specific embodiment, the feature spectrum generation processor processes the vertical displacement function. Methods for performing frequency domain analysis include: Applying a Fourier transform to the function yields its complex spectrum.

[0010] ; in, It is the total unfolded length of the spring; It is spatial frequency; It is an imaginary unit; and the frequency of the low-frequency main peak corresponding to the periodic effect of gravity is extracted from the spectrum. Amplitude and phase This constitutes the gravity disturbance feature vector. .

[0011] In one specific embodiment, the feature spectrum generation processor analyzes the transient temperature field data and acoustic emission signal data in the following ways: Calculate the maximum spatial temperature gradient norm in the transient temperature field data. and effective heat-affected zone area And calculate the signal energy in the acoustic emission signal data. and signal kurtosis These four elements together constitute the welding disturbance feature vector: ; Preferably, the operation of the online control law dynamic calibration module includes the following steps: First, based on the input perturbation feature spectrum and the preset performance target... The initial heat treatment command sequence is generated by solving a built-in heat treatment energy reverse compensation algorithm. ; Secondly, the initial heat treatment command sequence is simulated and executed on the virtual spring object to obtain a predicted final performance result. ; Then, the predicted final performance result is compared with the preset performance target to obtain a prediction error. ; Finally, when the norm of the prediction error Greater than the preset threshold Then, the initial heat treatment command sequence is corrected based on the prediction error to generate the final heat treatment command sequence. ; When the norm of the prediction error is not greater than a preset threshold In this case, the initial heat treatment instruction sequence is directly used as the final heat treatment instruction sequence.

[0012] In one specific embodiment, the virtual spring object is an object that incorporates the perturbation feature spectrum data and the material thermodynamic response model. The software object, the thermal response model Used to calculate in a given heat treatment instruction sequence The final residual stress state of the spring is as follows: ; in, Indicates the first The perturbation characteristic spectrum of a spring; The quality feedback and model self-optimization module iteratively optimizes at least one model built into the online control law dynamic calibration module, specifically the material thermodynamic response model. The parameters are updated.

[0013] In one specific embodiment, the heat treatment energy reverse compensation algorithm is implemented as a constraint optimization solver based on a physical model, the optimization objective being to find a solution that makes the objective function... Minimize heat treatment instruction sequence The quality feedback and model self-optimization module iteratively optimizes at least one model built into the online control law dynamic calibration module, specifically by adjusting the physical model or constraint conditions in the constraint optimization solver.

[0014] Preferably, the quality feedback and model self-optimization module includes: The final performance testing unit is configured for non-destructive testing of the final residual stress distribution or geometric accuracy of the spring. The data association and optimization processor is configured to receive the detection results of the final performance detection unit, establish an association database with the disturbance feature spectrum recorded by the spring during the production process and the final heat treatment instruction sequence, and then update the model parameters in the online control law dynamic calibration module through machine learning or statistical regression methods.

[0015] Preferably, the personalized heat treatment execution module is an online tempering channel composed of a linear arrangement of the plurality of independently addressable induction heating units.

[0016] This invention provides an intelligent control system that coordinates horizontal spring machining with single heat treatment. It has the following beneficial effects: 1. This invention, by setting up an online perturbation feature encoding module, can generate a unique perturbation feature spectrum for each individual spring. The spectrum quantifies the complex physical perturbations experienced during the rolling and welding processes. Subsequently, the online control law dynamic calibration module and the personalized heat treatment execution module determine and apply a non-uniform heat treatment scheme in time and space for that spring based on the unique perturbation feature spectrum. By using the individual differences of the spring as control input, it directly solves the problem of performance inconsistency caused by using uniform heat treatment parameters due to different processing histories, thereby improving the consistency of residual stress distribution and geometric accuracy within the final product batch.

[0017] 2. This invention establishes a feedforward correction closed loop by embedding a virtual spring object in the online control law dynamic calibration module and performing online verification and calibration of the generated initial heat treatment command sequence before physical execution. This allows for the early detection and correction of control errors caused by deviations between the model and reality just before the heat treatment command is applied to the physical spring. Compared to open-loop feedforward control techniques that rely solely on the model, this pre-calibration mechanism improves the accuracy of single heat treatment control, thereby increasing the first-pass yield of products.

[0018] 3. By adding a quality feedback and model self-optimization module, this invention establishes a long-term learning and optimization closed loop. It correlates the measured performance data of the final product with the corresponding disturbance characteristic spectrum of the product, and based on this, continuously iteratively optimizes the core model (such as the material thermodynamic response model) in the online control law dynamic calibration module. This enables the system to learn autonomously and adapt to slowly changing factors such as batch changes in raw materials and equipment state drift, thereby enhancing the long-term stability of the system and its robustness to external disturbances, and reducing the need for manual intervention and recalibration. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the system flow of the present invention; Figure 2 This is a schematic diagram of the structural architecture of the online perturbation feature encoding module of the present invention; Figure 3 This is a flowchart illustrating the workflow of the online control law dynamic calibration module of the present invention. Figure 4 This is a schematic diagram of the structure of the personalized heat treatment execution module of the present invention; Figure 5 This is a flowchart of the quality feedback and model self-optimization module of the present invention. Detailed Implementation

[0020] The technical solutions in the embodiments 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, and 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.

[0021] Please see the appendix Figure 1 This invention provides an intelligent control system that coordinates horizontal spring processing and single heat treatment, including: an online disturbance feature encoding module 100, an online control law dynamic calibration module 200, a personalized heat treatment execution module 300, and a quality feedback and model self-optimization module 400.

[0022] In this embodiment, the physical layout of the system is arranged sequentially along a production line 10. A spring 1 to be processed first passes through multiple acquisition stations of the disturbance feature online encoding module 100, then enters the personalized heat treatment execution module 300, and finally passes through a final performance detection unit of the quality feedback and model self-optimization module 400. The processor parts of the online control law dynamic calibration module 200 and the quality feedback and model self-optimization module 400 can be integrated into a central controller 500, such as an industrial computer or an edge computing server.

[0023] The overall workflow of the system includes a spring Physical flow parallel information flow, when spring As the spring 1 moves along production line 10, the online disturbance feature encoding module 100 is configured to collect physical disturbance data generated by the spring 1 during processing at different stations. Specifically, the online disturbance feature encoding module 100 collects data at a coiling station and a welding station, and processes this data to generate a unique disturbance feature spectrum characterizing the individual disturbance history of the spring. The spectrum The data structure can be represented as: ; in, For the first Disturbance characteristic spectrum of the spring; This is the normalized eigenvector of gravity perturbation; This is the normalized welding disturbance feature vector.

[0024] Generated perturbation feature spectrum The disturbance characteristics are transmitted from the online encoding module 100 to the online control law dynamic calibration module 200 within the central controller 500. The online control law dynamic calibration module 200 receives... Subsequently, it is configured to inversely solve and calibrate based on the spectrum, outputting a final heat treatment command sequence for spring 1. This process establishes a control decision path based on individual perturbation characteristics.

[0025] Final heat treatment instruction sequence The data is transmitted from the online control law dynamic calibration module 200 to the personalized heat treatment execution module 300. When the spring 1 physically reaches and enters the personalized heat treatment execution module 300, the personalized heat treatment execution module 300 is configured for precise execution. The command controls multiple induction heating units inside the spring 1 to apply non-uniform energy injection in time and space to different sections of the spring 1, completing a personalized heat treatment process. This step constitutes a real-time, individual-specific control execution closed loop.

[0026] After heat treatment, spring 1 continues to move along production line 10, passing through a final performance testing unit 410 located downstream. This final performance testing unit 410 is part of the quality feedback and model self-optimization module 400 and is configured to acquire the final performance parameters of spring 1, such as its actual residual stress distribution. .

[0027] Measured performance parameters The data is transmitted to the processor of the quality feedback and model self-optimization module 400 within the central controller 500. The quality feedback and model self-optimization module 400 is configured to correlate three data items of the same spring. Initial perturbation feature spectrum The final heat treatment command sequence applied And the final measured performance By analyzing multiple accumulated data triples, the quality feedback and model self-optimization module 400 iteratively updates the physical model or algorithm parameters built into the online control law dynamic calibration module 200. This process establishes a long-cycle self-optimization learning closed loop for improving the accuracy of the system model.

[0028] See attached document Figure 2 In one specific embodiment, the disturbance feature online encoding module 100 is configured at the front end of the production line 10 to quantify and encode the physical disturbance generated by each passing spring 1 during the processing. The disturbance feature online encoding module 100 may include a rolling disturbance acquisition unit 110, a welding disturbance acquisition unit 120, and a feature spectrum generation processor 130.

[0029] The winding disturbance acquisition unit 110 is positioned after the horizontal winding station of the spring 1 to acquire the shape disturbance caused by gravity. The winding disturbance acquisition unit 110 may include a high-frequency laser profile scanner. The scanning axis of the high-frequency laser profile scanner is perpendicular to the conveying direction of the spring 1. When the spring 1 passes through the scanning area, the scanner acquires three-dimensional point cloud data of its surface. The feature spectrum generation processor 130 receives the point cloud data and extracts the displacement function of the spring neutral layer in the vertical direction. ,in This is the unfolded length along the neutral layer of the spring.

[0030] To extract the vertical displacement function The periodic perturbation characteristics in the function are analyzed by the feature spectrum generator 130, which applies a Fourier transform to the function to obtain its spatial spectrum.

[0031] ; in, It is the total unfolded length of the spring; It is spatial frequency; It is the imaginary unit; the characteristic spectrum generation processor 130 then analyzes the spectrum. And from the low-frequency band corresponding to the periodic effects of gravity, the frequency of the main peak is identified. Amplitude and phase These three parameters together constitute the gravitational perturbation eigenvector. .

[0032] A welding disturbance acquisition unit 120 is installed at the laser welding station of spring 1 to acquire thermal shock and acoustic disturbances generated during the welding process. The welding disturbance acquisition unit 120 may include an infrared thermal imager and at least one acoustic emission sensor. The infrared thermal imager is configured to focus on the welding area and acquire transient temperature field data of the welding point and its surroundings. The acoustic emission sensor is coupled to the spring clamp or spring body to collect acoustic emission signal data during the welding process. .

[0033] The feature spectrum generation processor 130 receives the above two sets of data, for transient temperature field data. The feature spectrum generator 130 calculates the welding time. Maximum spatial temperature gradient norm and the effective heat-affected zone area exceeding a specific threshold temperature For acoustic emission signal data The feature spectrum generator processor 130 calculates the total energy of the signal. and signal kurtosis These four calculation results together constitute the welding disturbance feature vector. .

[0034] Obtaining the gravitational perturbation feature vector and welding disturbance eigenvectors Then, the feature spectrum generation processor 130 normalizes each component of the two vectors separately, for example, by using minimum and maximum normalization methods to eliminate the influence of dimensions. The normalized vectors are then... and They are combined into a unified, structured data object, namely, the unique perturbation characteristic spectrum of this spring. , It is then transmitted via the data bus to the online control law dynamic calibration module 200 as input for subsequent control decisions.

[0035] See attached document Figure 1 and attached Figure 3 The online control law dynamic calibration module 200 is integrated into the central controller 500. The online control law dynamic calibration module 200 receives the disturbance feature spectrum generated by the disturbance feature online encoding module 100 for a specific spring 1 via the data bus. Its core function is to generate a final heat treatment instruction sequence for this spring 1. .

[0036] In one specific embodiment, a heat treatment energy reverse compensation algorithm is deployed within the online control law dynamic calibration module 200. The goal of the heat treatment energy reverse compensation algorithm is to solve for an initial heat treatment command sequence. The solution process is constructed as a constrained optimization problem. The objective function is to minimize the difference between the predicted final performance after heat treatment and the preset performance target. The optimization problem can be expressed as: ; in, It is the heat treatment command sequence to be solved, which can be a vector containing the power and operating frequency settings for multiple induction heating units; It is the perturbation characteristic spectrum of the input spring 1; These are preset performance targets; for example, an ideal residual stress distribution function or a set of stress values ​​at key locations. It is a built-in physical model describing the thermodynamic response of a spring under given perturbation and thermal input; the optimization problem can be solved using standard numerical methods, such as gradient descent or interior point methods, i.e., the initial heat treatment command sequence. .

[0037] Obtain the initial heat treatment instruction sequence Subsequently, the online control law dynamic calibration module 200 instantiates a virtual spring object for online verification and calibration of the initial command sequence. The virtual spring object is a software entity whose internal state is determined by the input perturbation characteristic spectrum. The virtual spring object is initialized and also encapsulates the same material thermodynamic response model as the inverse compensation algorithm. .

[0038] The virtual spring object receives the initial heat treatment instruction sequence. As input, and simulating the execution of this instruction, a predicted final performance result is calculated. : ; Subsequently, the online control law dynamic calibration module 200 calculates the prediction error between the prediction result and the preset performance target. .

[0039] The online control law dynamic calibration module 200 will calculate the norm of the prediction error. With a preset error threshold When comparing, This indicates that the initial instruction sequence is insufficient to meet performance requirements and feedforward calibration is necessary. The online control law dynamic calibration module 200 will adjust the calibration based on the prediction error. Calculate a correction command quantity The final heat treatment instruction sequence is generated as follows: ; Among them, the amount of correction instructions It can be determined by a pre-calibrated sensitivity matrix or a proportional controller. Confirmed, when This indicates that the prediction result of the initial instruction sequence is within an acceptable range, and the online control law dynamic calibration module 200 directly uses the initial instruction sequence as the final instruction sequence. ; The final generated heat treatment instruction sequence The data is sent via the central controller 500 to the personalized heat treatment execution module 300 to guide the precise heat treatment operation of the physical spring 1.

[0040] See attached document Figure 1 and attached Figure 4 The personalized heat treatment execution module 300 is deployed on the production line 10, located after the disturbance feature online encoding module 100, and communicates with the output of the central controller 500. Its function is to receive and execute the final heat treatment instruction sequence to apply precise and individualized heat treatment to the passing spring 1.

[0041] In one specific embodiment, the personalized heat treatment execution module 300 is constructed as an online tempering channel. Multiple independently addressable induction heating units 310-j are linearly arranged inside the channel along the conveying direction of the spring 1, where j = 1, 2, ..., N, and N is the total number of heating units. The system also includes a conveying device to ensure that the spring 1 maintains a preset, constant speed. Through this online tempering channel.

[0042] The personalized heat treatment execution module 300 receives the final heat treatment command sequence for a specific spring 1 from the central controller 500 through its control interface. The final heat treatment instruction sequence is a structured dataset, which can take the following form: ; in, and Each is assigned to the first The power and operating frequency values ​​of each induction heating unit 310-j are specified in this instruction sequence, which is specifically generated for the spring to compensate for its unique disturbance history.

[0043] When spring 1 passes through the channel, the internal controller of the personalized heat treatment execution module 300 parses the instruction sequence. The drive power supply for each induction heating unit 310-j is independently adjusted, with a specified power for each. and frequency In operation, different sections of spring 1 will pass through different induction heating units 310-j in sequence during the movement process, and the energy output of each unit is different according to the command sequence. Therefore, non-uniform energy injection is achieved in space along the entire length of spring 1, thus completing the personalized heat treatment process.

[0044] See attached document Figure 1 and attached Figure 5 The quality feedback and model self-optimization module 400 includes a final performance testing unit 410 located downstream of the production line 10, and a data association and optimization processor 420 integrated in the central controller 500. The function of the quality feedback and model self-optimization module 400 is to establish a long-term learning loop and continuously optimize the accuracy of the internal model of the system.

[0045] The final performance testing unit 410 is deployed after the personalized heat treatment execution module 300. In one specific embodiment, the final performance testing unit 410 may be an X-ray diffractometer configured for non-destructive testing of the final actual residual stress distribution of the heat-treated spring 1. After the test is completed, the measured data is assigned a unique identifier for the spring, forming a data pair. The data is then transmitted to the central controller 500.

[0046] After receiving the above data, the data association and optimization processor 420 first determines the spring identifier. Retrieve the disturbance characteristic spectrum of this spring from its internal database, which was recorded in the early stages of production. and the final heat treatment instruction sequence generated for it The data association and optimization processor 420 associates these three elements to form a complete data triple: ; The data triples are stored in a long-term learning database, which continues to accumulate a large number of such data records as production progresses.

[0047] The data association and optimization processor 420 is configured to periodically, or after a certain amount of data has accumulated, use a database to compare the material thermodynamic response model built into the online control law dynamic calibration module 200. Perform iterative optimization and set the model. From a set of parameters As defined, the goal of optimization is to find a new set of parameters. By minimizing the cumulative error between model predictions and a large number of actual measurements, the objective can be achieved by minimizing a loss function. To achieve: ; in, It represents the total number of triples in the database; Is the model with parameters? At the time of the first Predicted performance values ​​for each spring; These are the corresponding actual measured values.

[0048] The minimization process can be performed using standard optimization algorithms, such as gradient descent-based algorithms or statistical regression methods, to solve for the optimized parameters. It will be used to update the material thermodynamic response model in the online control law dynamic calibration module 200. This allows the system's subsequent control decisions to be based on a more accurate model calibrated with actual data.

[0049] See attached document Figure 1 -Appendix Figure 5First, a spring 1 with an identification mark enters the production line 10 and passes through each acquisition unit of the disturbance feature online encoding module 100 in sequence. When it passes through the winding disturbance acquisition unit 110, the system acquires its three-dimensional contour data, and the feature spectrum generation processor 130 calculates its vertical displacement function, thereby extracting the gravity disturbance feature vector. .

[0050] Subsequently, spring 1 moves to the welding station, and welding disturbance acquisition unit 120 acquires the transient temperature field and acoustic emission signal during the welding process. Feature spectrum generation processor 130 receives this data and extracts the welding disturbance feature vector. The feature map generation processor 130 will generate two feature vectors. and After normalization and combination, a unique, structured perturbation feature spectrum for spring 1 is finally generated. .

[0051] Perturbation feature spectrum The data is transmitted to the online control law dynamic calibration module 200 deployed within the central controller 500, and the online control law dynamic calibration module 200, according to... and preset performance targets Using its built-in heat treatment energy reverse compensation algorithm, an initial heat treatment command sequence can be solved. .

[0052] Following this, the online control law dynamic calibration module 200 utilizes... To initialize a virtual spring object and simulate the execution of initial instructions on this object. To obtain a predictive performance result The online control law dynamic calibration module 200 calculates the error between this prediction result and the target, and performs feedforward calibration on the initial command based on the error to generate a final heat treatment command sequence. .

[0053] When the physical form of spring 1 arrives at the entrance of the personalized heat treatment execution module 300 along the production line 10, the personalized heat treatment execution module 300 has received the final instruction sequence customized for it from the central controller 500. Spring 1 passes through the heat treatment channel of the module at a constant speed, and multiple induction heating units 310-j within the channel operate according to a command sequence. By applying precise and non-uniform energy to different sections of spring 1 according to specific parameters, personalized heat treatment is achieved.

[0054] After heat treatment, spring 1 continues to move to the downstream final performance testing unit 410, where it undergoes non-destructive testing to obtain its final actual performance parameters. The measurement result is transmitted to the data association and optimization processor 420 in the central controller 500.

[0055] The data association and optimization processor 420 will generate the perturbation feature spectrum of spring 1. Final heat treatment instruction sequence and actual performance parameters The data is correlated to form a complete data triple and stored in the database. Once a sufficient amount of data has accumulated in the database, the data correlation and optimization processor 420 will execute an optimization algorithm to optimize the material thermodynamic response model in the online control law dynamic calibration module 200. The parameters are updated to complete a long-cycle learning and self-optimization closed loop. This process is repeated for every spring that subsequently enters the production line.

[0056] The system in this embodiment can be used to execute the above algorithm embodiments, and its principle and technical effect are similar, so they will not be described again here.

Claims

1. An intelligent control system for the coordinated horizontal machining and single heat treatment of springs, characterized in that, include: The online perturbation feature encoding module is configured to collect physical perturbation data in real time during the horizontal processing of springs, and generate a unique perturbation feature spectrum for each independent spring by signal processing and data fusion, which characterizes its composite perturbation history. The online control law dynamic calibration module is configured to generate an initial heat treatment command sequence by inversely solving based on the disturbance feature spectrum, and to perform online verification and calibration of the initial heat treatment command sequence through a built-in virtual spring object, and output a final heat treatment command sequence. A personalized heat treatment execution module is configured to receive the final heat treatment instruction sequence and control multiple independently addressable induction heating units inside to apply a non-uniform, personalized heat treatment in time and space to the passing spring. The quality feedback and model self-optimization module is configured to perform final performance testing on the spring after it has been processed by the personalized heat treatment execution module, and associate the test results with the corresponding disturbance feature spectrum to iteratively optimize at least one model built into the online control law dynamic calibration module.

2. The intelligent control system for the coordinated horizontal machining and single heat treatment of a spring according to claim 1, characterized in that, The online encoding module for the perturbation features includes: The coiling disturbance acquisition unit is configured to acquire three-dimensional contour data during the horizontal coiling process of the spring and extract the vertical displacement function caused by gravity. The welding disturbance acquisition unit is configured to acquire transient temperature field data and acoustic emission signal data at the welding point during the laser welding process of the spring. The feature spectrum generation processor is configured to perform frequency domain analysis on the vertical displacement function to extract the gravity disturbance feature vector, and to analyze the transient temperature field data and acoustic emission signal data to extract the welding disturbance feature vector. Then, the gravity disturbance feature vector and the welding disturbance feature vector are fused to generate the disturbance feature spectrum.

3. The intelligent control system for the coordinated horizontal machining and single heat treatment of springs according to claim 2, characterized in that, The feature spectrum generation processor is further configured to: obtain the spectrum by applying Fourier transform to the vertical displacement function, and extract the frequency, amplitude and phase of the low-frequency main peak corresponding to the periodic influence of gravity from the spectrum to form the gravity disturbance feature vector.

4. The intelligent control system for the coordinated horizontal machining and single heat treatment of a spring according to claim 2, characterized in that, The feature spectrum generation processor is further configured to: jointly construct the welding disturbance feature vector by calculating the maximum spatial temperature gradient norm and effective heat-affected zone area in the transient temperature field data, and calculating the signal energy and signal kurtosis in the acoustic emission signal data.

5. The intelligent control system for the coordinated horizontal machining and single heat treatment of a spring according to claim 1, characterized in that, The online control law dynamic calibration module is configured to perform the following functions: Based on the input perturbation feature spectrum and the preset performance target, the initial heat treatment command sequence is generated by a built-in heat treatment energy reverse compensation algorithm. The initial heat treatment instruction sequence is simulated and executed on the virtual spring object to obtain a predicted final performance result; The predicted final performance result is compared with the preset performance target to obtain a prediction error; When the prediction error is greater than a preset threshold, the initial heat treatment instruction sequence is corrected according to the prediction error to generate the final heat treatment instruction sequence. When the prediction error is not greater than a preset threshold, the initial heat treatment instruction sequence is directly used as the final heat treatment instruction sequence.

6. The intelligent control system for the coordinated horizontal machining and single heat treatment of springs according to claim 5, characterized in that, The virtual spring object is a software object configured to have the perturbation feature spectrum data and the material thermodynamic response model built in. The thermodynamic response model is used to calculate the final residual stress state of the spring under a given heat treatment command sequence. Furthermore, the quality feedback and model self-optimization module iteratively optimizes at least one model built into the online control law dynamic calibration module, specifically updating the parameters of the material thermodynamic response model.

7. The intelligent control system for the coordinated horizontal machining and single heat treatment of a spring according to claim 5, characterized in that, The heat treatment energy reverse compensation algorithm is implemented as a constraint optimization solver based on a physical model; Furthermore, the quality feedback and model self-optimization module iteratively optimizes at least one model built into the online control law dynamic calibration module, that is, it specifically adjusts the physical model or constraints in the heat treatment energy reverse compensation algorithm.

8. The intelligent control system for the coordinated horizontal machining and single heat treatment of springs according to claim 1, characterized in that, The quality feedback and model self-optimization module includes: The final performance testing unit is configured for non-destructive testing of the final residual stress distribution and geometric accuracy of the spring. The data association and optimization processor is configured to receive the detection results of the final performance detection unit, establish an association database with the disturbance feature spectrum recorded by the spring during the production process and the final heat treatment instruction sequence, and then update the model parameters in the online control law dynamic calibration module through machine learning and statistical regression methods.

9. The intelligent control system for the coordinated horizontal machining and single heat treatment of a spring according to claim 1, characterized in that, The personalized heat treatment execution module is an online tempering channel composed of a linear arrangement of multiple independently addressable induction heating units, through which the springs sequentially complete the personalized heat treatment.

10. The intelligent control system for the coordinated horizontal machining and single heat treatment of a spring according to claim 1, characterized in that, The online control law dynamic calibration module and the quality feedback and model self-optimization module are both deployed on edge computing controllers or cloud servers.

Images