Hydraulic forming control method and device, computer equipment and storage medium

Abstract

The invention provides a hydraulic forming control method and device, computer equipment and a storage medium, and belongs to the field of equipment control. The method comprises the steps that geometric parameters, product quality indexes and corresponding process strategies serve as a data set to be subjected to machine learning, and a pre-trained machine learning classification model is obtained; geometric parameters and expected product quality indexes of the target product are obtained, and a corresponding process strategy is determined through a pre-trained machine learning classification model; according to the process strategy and the global process parameters corresponding to the process strategy, multi-section process parameters are generated through a corresponding deterministic generation algorithm; the certainty generation algorithm is a preset calculation rule corresponding to the process strategy; and according to the multi-section process parameters, a control instruction for the hydraulic forming equipment to execute the forming process is generated through the programmable logic controller. In this way, an intelligent process parameter generation method and an open type control system architecture are organically fused, and an effective path is provided for upgrading of a complex industrial forming process.

Description

Technical Field

[0001] This invention belongs to the field of equipment control, and specifically relates to a control method, device, computer equipment, and storage medium for hydraulic forming. Background Technology

[0002] In modern industrial manufacturing, hydroforming (especially high-pressure forming of pipe fittings) technology has been widely used in the automotive, aerospace, and other fields due to its significant advantages in achieving lightweighting and structural integration. This process uses the synergistic effect of internal high-pressure fluid and axial feed to cause plastic deformation of the tube blank, allowing it to conform to a complex mold cavity. However, the process is extremely complex, and the final forming quality, such as wall thickness uniformity and thinning rate, is highly dependent on the precise setting of the loading path (i.e., the synergistic control curve of internal pressure and axial feed displacement).

[0003] Currently, this technology relies heavily on a "forward trial-and-error" approach at the process design and decision-making levels. Existing methods typically involve engineers pre-setting a set of loading paths based on their personal experience, then verifying their effectiveness through numerical simulations or physical experiments, and repeatedly adjusting them to obtain a feasible solution. This approach is inefficient, has a long development cycle, high trial-and-error costs, and makes it difficult to guarantee the consistency of process quality. Although academia has conducted extensive "forward" research on loading path optimization, industrial production still lacks an effective method that can automatically and intelligently generate optimal process parameters from the desired product quality indicators.

[0004] Furthermore, the system is closed, making it difficult to flexibly integrate advanced intelligent algorithms (such as machine learning models) developed using modern general-purpose programming languages ​​(such as Python); software development and runtime licensing costs are high; and integration with third-party systems (such as MES, cloud platforms) is difficult, resulting in poor scalability. For example, the host computer program is "hard-coded" and tightly coupled with the PLC's brand, model, and internal memory address. Once the PLC is replaced or the program address is changed, the host computer program must be modified and redeployed accordingly, leading to extremely poor maintainability. At the same time, the PLC's limited communication processing capabilities cannot support concurrent access from multiple clients, resulting in connection bottlenecks and real-time risks, and directly exposing the core control device, the PLC, to the network environment brings high security risks.

[0005] In summary, existing technologies are insufficient to support the development of hydroforming processes towards intelligence and flexibility. There is an urgent need for a new solution that can simultaneously address the dependence on "forward trial and error" in process decision-making and the problem of open and scalable control system architecture. Summary of the Invention

[0006] To address the aforementioned problems, the present invention provides a control method, apparatus, computer equipment, and storage medium for hydraulic forming.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A control method for hydroforming is applied to a three-layer decoupled architecture control system, including an upper-computer application layer node, a lower-computer execution layer node, and a middleware gateway layer node. The upper-computer application layer node is configured to run an application program developed in a general-purpose programming language to generate multi-segment process parameters. The lower-computer execution layer node includes a programmable logic controller (PLC) for executing specific equipment control logic. The middleware gateway layer node is used to establish communication between the upper-computer application layer node and the lower-computer execution layer node. The method includes: The host computer application layer node is used to acquire the geometric parameters and expected product quality indicators of the target product; the geometric parameters and expected product quality indicators are used to determine the corresponding process strategy through a pre-trained machine learning classification model; based on the process strategy and its corresponding global process parameters, multi-segment process parameters are generated through a corresponding deterministic generation algorithm; wherein, the deterministic generation algorithm is a calculation rule preset according to the process strategy; the dataset for pre-training the machine learning classification model includes: acquiring the product quality indicators of workpieces with different geometric parameters under different process strategies, and using the geometric parameters, product quality indicators and corresponding process strategies as the dataset; The upper computer application layer node sends multi-segment process parameters to the lower computer execution layer node through the middleware gateway layer node, and generates control instructions for the hydraulic forming equipment to execute the forming process through the programmable logic controller.

[0008] Optionally, the deterministic generation algorithm is a piecewise function of displacement and pressure, with the following paradigm: S=F(Dtotal, Ptotal, Ctype); Where S is the output multi-segment displacement and pressure numerical sequence, D_total is the total displacement, P_total is the total pressure, and C_type is the process strategy; The algorithm performs precise calculations by invoking a preset mathematical paradigm based on the process strategy.

[0009] Optionally, the process strategy includes one or more of the following: linear loading, fast-then-slow loading, and plateau loading, with each process strategy corresponding to a different piecewise displacement and pressure function.

[0010] Optionally, the middleware gateway layer node is an intermediate server, and the upper computer application layer node sends multi-segment process parameters to the lower computer execution layer node through the middleware gateway layer node. The control instructions for the hydraulic forming equipment to execute the forming process are generated by the programmable logic controller, including: The host computer application layer node encapsulates the multi-segment process parameters into a hardware-independent data format and sends them to the intermediate server; The intermediate server converts the hardware-independent data format into an instruction format that the programmable logic controller (PLC) can recognize, and then sends it to the target PLC. The target programmable logic controller executes the instructions to control the hydraulic forming equipment to complete the forming process.

[0011] Optionally, the hardware-independent data format is JSON, which is transmitted to an intermediate server via TCP / IP protocol.

[0012] Optionally, converting the hardware-independent data format into an instruction format recognizable by the programmable logic controller (PLC) via the intermediate server and then sending it to the target PLC includes: The intermediate server parses the JSON format data to extract process parameters and control commands. According to the preset address mapping table, the process parameters are converted into the target PLC register address and written. The converted instructions are sent to the PLC using a dedicated PLC communication protocol.

[0013] Optionally, when the target programmable logic controller executes the instruction, the intermediate server reads the status register data of the PLC in a polling manner, encapsulates the read real-time data into JSON format and feeds it back to the host computer. The host computer updates the monitoring information in the graphical interface according to the feedback data to realize closed-loop control.

[0014] A control device for hydraulic forming, applied to the control method for hydraulic forming, the device comprising: The training module is used to obtain product quality indicators of workpieces with different geometric parameters under different process strategies; the geometric parameters, product quality indicators and corresponding process strategies are used as a dataset for machine learning to obtain a pre-trained machine learning classification model. The acquisition module is used to acquire the expected product quality indicators of the target product. The classification module is used to determine the corresponding process strategy by using the geometric parameters and expected product quality indicators through a pre-trained machine learning classification model. The generation module is used to generate multi-segment process parameters based on the process strategy and its corresponding global process parameters using a corresponding deterministic generation algorithm; the deterministic generation algorithm is a calculation rule preset according to the process strategy. The control module is used to generate control instructions for the hydraulic forming equipment to execute the forming process based on multi-segment process parameters via a programmable logic controller.

[0015] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned control method for hydraulic forming.

[0016] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the aforementioned control method for hydroforming.

[0017] The hydraulic forming control method provided by this invention has the following beneficial effects: This invention employs a three-layer decoupled control system architecture and uses a general-purpose programming language to implement host computer configuration. This allows the upper layer to generate parameters, the middle layer to transmit instructions, and the lower layer to execute operations, resulting in a clear division of labor and efficient collaboration. The entire process logic is clear, easily expandable, and can flexibly adapt to different production needs, achieving system openness and decoupling. Furthermore, by collecting product quality indicators of workpieces with different geometric parameters under different process strategies and constructing a dataset for machine learning, it breaks away from the reliance on traditional "forward trial and error" in process decision-making. It utilizes data-driven approaches to quickly and accurately determine the process strategy corresponding to the target product, improving decision-making efficiency and accuracy. Moreover, based on the process strategy and global process parameters, it generates multi-segment process parameters, ensuring the standardization and stability of parameter generation. This provides a feasible and efficient new solution for the intelligent and flexible development of hydroforming processes. Attached Figure Description

[0018] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic flowchart of a control method for hydraulic forming according to an exemplary embodiment of the present invention.

[0020] Figure 2 This is a block diagram of a control device for hydraulic forming according to an exemplary embodiment of the present invention. Detailed Implementation

[0021] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0022] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] First, this invention provides an open three-layer decoupled architecture, including: The host computer application layer node is configured to run an application developed in a general-purpose programming language. This application integrates a process intelligence decision-making module to receive user-inputted desired product quality indicators and generate hardware-independent multi-segment process parameter instructions based on these indicators using a built-in algorithm model. The slave execution layer node includes at least one programmable logic controller (PLC) for executing specific equipment control logic. The middleware gateway layer node runs as an independent server program and communicates with both the host computer application layer node and the slave execution layer node. The host computer application layer node and the middleware gateway layer node communicate via a first standard network protocol and exchange instructions and data using a hardware-independent data format. The middleware gateway layer node and the slave execution layer node communicate via a second communication protocol, which is compatible with the brand and model of the PLC. The middleware gateway layer node is configured to receive the process parameter instructions from the host computer application layer node, translate them into low-level control instructions that the slave execution layer node can recognize, and forward them, thereby decoupling the host computer application layer node and the slave execution layer node.

[0024] Based on the above system, the present invention provides a control method for hydraulic forming, specifically as follows: Figure 1 As shown, it includes the following steps: S101. Obtain product quality indicators for workpieces with different geometric parameters under different process strategies; use the geometric parameters, product quality indicators, and corresponding process strategies as a dataset for machine learning to obtain a pre-trained machine learning classification model.

[0025] In this step, the first step is to build a machine learning classifier that can intelligently recommend the optimal process strategy (loading path category) based on the desired product quality.

[0026] For example, simulations can be combined based on pipe fittings with various geometric parameters and various process strategies to construct a simulation dataset containing multiple geometric features and process path categories. Based on this simulation dataset, forming quality indicators are extracted as features, and process path categories are extracted as labels. A classification model is trained using machine learning algorithms, and the model is evaluated. The classification model that meets the preset requirements is saved as a pre-trained machine learning classification model.

[0027] In one embodiment, the most representative geometric feature parameters in pipe hydroforming are selected, such as expansion ratio (110-160), expansion ratio (1-6), and fillet radius (5-20). To cover a wide range of working conditions, a reasonable range of values ​​is set for each parameter and discretized sampling is performed to design 1000 sets of molds with different geometric features.

[0028] N typical loading paths (N=6 in this embodiment) that are most representative in engineering practice are selected, such as "linear loading", "fast-then-slow" loading, and "platform loading". These 6 paths are numbered 1 to 6 respectively, serving as the category labels that the model needs to predict.

[0029] Using the finite element analysis software Dynaform, large-scale numerical simulations were performed on all combinations of the aforementioned "mold" and "loading path categories". For each simulation case, after the calculation was completed, key performance indicators that can measure the final forming quality, such as thinning rate, wall thickness distribution uniformity, and forming pressure, were automatically extracted from the results.

[0030] The performance metrics of each case are used as input features of the model, and the loading path number used in that case is used as the output label. All case data are aggregated to form a structured dataset. Each row of data represents a complete process experiment, in the format: [expansion rate, expansion ratio, fillet radius, maximum thinning rate, ..., loading path category].

[0031] The constructed dataset is divided into training, validation, and test sets. Input features are standardized (e.g., normalized) to eliminate the influence of different physical units, improving the efficiency and stability of model training.

[0032] Three machine learning classification algorithms (XGBoost, RF, and SVM) were selected and trained on the training set. The goal of the training was to minimize the model's classification prediction error, that is, to enable the model to learn to accurately determine which type of loading path is most likely to produce such a result based on the quality and geometric features of the input.

[0033] The model's performance metrics, such as accuracy, precision, and recall, are evaluated on the test set, and the best-performing model is selected. Finally, the trained, high-performance model is serialized and saved as a .pkl file so that it can be quickly loaded and called in the host computer application.

[0034] S102. The host computer application layer node is used to obtain the geometric parameters and expected product quality indicators of the target product; and to determine the corresponding process strategy by using the geometric parameters and expected product quality indicators through a pre-trained machine learning classification model.

[0035] After obtaining the trained machine learning classification model, the geometric parameters and expected product quality indicators of the target product to be produced are classified based on the model, and the corresponding process strategy is determined.

[0036] For example, operators can operate through a GUI interface developed using Python and the Tkinter graphics library. Instead of directly setting cumbersome five-segment displacement-pressure values, operators input a set of workpiece geometric parameters and the desired final product target based on the new product's process requirements, such as: maximum thickness: 1.53mm, minimum thickness: 1.23mm. Simultaneously, global parameters required to complete the process are determined and input, such as material replenishment displacement D_total=12mm, shaping pressure P_total=60MPa. When the operator clicks the "Generate Process" button, the Python program first calls the machine learning classification model trained in the first stage (i.e., loading the .pkl file). It takes the user-input target feature data as input, and the model performs inference calculations, outputting an optimal process path category (i.e., process strategy), for example, 3, representing a "linear loading" path.

[0037] S103. Based on the process strategy and its corresponding global process parameters, multi-segment process parameters are generated using the corresponding deterministic generation algorithm.

[0038] In this step, after generating the process strategy, the host computer application layer node generates multi-segment process parameters based on its corresponding global process parameters and a preset deterministic generation algorithm, so that the lower-level programmable logic controller can execute them. That is, this step requires obtaining the necessary global process parameters according to the process strategy determined in the preceding steps, and then calculating the specific multi-segment process parameters using a deterministic generation algorithm. Here, the deterministic generation algorithm is a preset calculation rule corresponding to the process strategy.

[0039] In one embodiment, after obtaining the specific loading path category (i.e., process strategy), the program automatically determines the path category and calls the corresponding deterministic generation algorithm: S = F(D_total, P_total, C_type). Here, S is the output multi-segment displacement and pressure numerical sequence, D_total is the total displacement, P_total is the total pressure, and C_type is the process strategy. The algorithm calls a preset corresponding mathematical paradigm for precise calculation based on the process strategy. This mathematical paradigm includes at least one of linear loading, fast-then-slow loading, and plateau loading, with each paradigm corresponding to a different piecewise displacement and pressure function.

[0040] For example, function F receives three input parameters: D_total=12, P_total=60, and C_type=3. The internal logic of function F acts as a scheduler or dispatcher. Based on the value of input C_type, it selects and activates the paradigm f3 corresponding to C_type=3 from its internally encapsulated set of mathematical processing paradigms (f1, f2, ... f6).

[0041] The activated paradigm f3 (paradigm 3 linear loading mathematical model: , Let i = {1, 2, 3, 4, 5}. The mathematical model of Paradigm 1 is: β*Ptotal*(i+1) , , Let i = {1, 2, 3, 4, 5}. The formulas for normal forms 2, 5, and 6 are the same, but... Different coefficients, in Normal Form 2 Normal form 5 is the opposite of normal form 2. Normal form 6 is the opposite of normal form 1. The mathematical model of paradigm 4 can be described as follows: , , The calculation begins with i = {1, 2, 3, 4}. Based on the input D_total = 12 and P_total = 60, it accurately calculates the specific displacement and pressure target values ​​for the five stages according to its internally preset mathematical model used to describe the linear loading curve (the first stage is when the pressure rises to 20% of the total pressure, the second stage is when the pressure is maintained while the displacement reaches 40% of the total displacement, the third stage is when the pressure rises to 60% at the same time, etc.).

[0042] Finally, the function F returns a dataset S containing these precise values, such as Displacements: [2.4,4.8, 7.2,9.6, 12.0, Pressures: [18.0, 36.0, 36.0, 45.0, 60.0].

[0043] S104. The upper computer application layer node sends multi-segment process parameters to the lower computer execution layer node through the middleware gateway layer node, and generates control instructions for the hydraulic forming equipment to execute the forming process through the programmable logic controller.

[0044] The middleware gateway layer node is an intermediate server. The host computer application layer node encapsulates the multi-segment process parameters into a hardware-independent data format and sends it to the intermediate server. The intermediate server converts the hardware-independent data format into an instruction format that can be recognized by the programmable logic controller (PLC) and sends it to the target PLC. The target PLC executes the instruction to control the hydraulic forming equipment to complete the forming process.

[0045] The hardware-independent data is in JSON format and is transmitted to an intermediate server via TCP / IP protocol.

[0046] After determining the multi-stage process parameters in the aforementioned steps, this step involves issuing instructions and converting protocols.

[0047] In one embodiment, the intermediate server parses JSON format data to extract process parameters and control instructions; according to a preset address mapping table, the process parameters are converted into write operations to the target PLC register address; and the converted instructions are sent to the PLC using a PLC-specific communication protocol.

[0048] For example, a Python program can encapsulate the precise numerical dataset S generated in the previous step, along with the control commands ("sent to the PLC"), into a hardware-independent, standardized JSON object. This JSON object is then sent via TCP / IP to a C# intermediate server listening on the network, where protocol conversion transforms the instructions into a language understood by the PLC.

[0049] The C# intermediate server receives and parses the JSON object, recognizing it as a "process" instruction, and identifies the relevant PLC register addresses already defined in the server. The C# intermediate server then begins the "compilation" task: it iterates through each floating-point value in the JSON (e.g., 2.4) and performs data conversion operations according to the target PLC's (Mitsubishi FX3GA) requirements (e.g., multiplying by a scaling factor of 10 and rounding to obtain the integer 24). Subsequently, the C# intermediate server uses a Mitsubishi PLC-specific communication library (e.g., ActUtlTypeLib) and MXComponent to establish a connection with the PLC and writes the converted integer values ​​sequentially into the PLC's designated data registers according to a preset address mapping table (e.g., writing 24 to D100, 48 to D101, etc.).

[0050] After all data has been confirmed to be correct, the C# intermediate server sends a set signal to a relay (M101) in the PLC, triggering the PLC to start execution. Upon receiving the trigger signal, the PLC begins running its internal control program, precisely controlling the servo motor and hydraulic system according to the value in the D register to complete the pipe forming process.

[0051] During execution, the C# intermediate server reads the status registers in the PLC that represent the actual pressure and actual displacement in a polling mode (every 1 second), packages this real-time data into JSON format, and sends it back to the Python host computer. The host computer updates the monitoring values ​​on the interface in real time, thereby realizing a complete closed-loop control of "target input" - "intelligent decision-making and process generation" - "server conversion" - "lower-level execution" - "real-time feedback".

[0052] The above method employs a three-layer decoupled control system architecture and uses a general-purpose programming language to implement the host computer configuration. This allows the upper layer to generate parameters, the middle layer to transmit instructions, and the lower layer to execute operations, resulting in a clear division of labor and efficient collaboration. The entire process logic is clear, easily expandable, and can flexibly adapt to different production needs, achieving system openness and decoupling. Furthermore, by collecting product quality indicators of workpieces with different geometric parameters under different process strategies and constructing a dataset for machine learning, it breaks away from the reliance on traditional "forward trial and error" in process decision-making. Data-driven approaches quickly and accurately determine the process strategy corresponding to the target product, improving decision-making efficiency and accuracy. Moreover, based on the process strategy and global process parameters, multi-segment process parameters are generated, ensuring the standardization and stability of parameter generation. This provides a feasible and efficient new solution for the intelligent and flexible development of hydroforming processes.

[0053] Based on the above steps, this invention also provides an open, three-layer decoupled industrial control system. The system generally includes: one or more upper-level application layer nodes (such as Python upper-level nodes) responsible for user interaction and intelligent decision-making; a middleware gateway layer node (such as a C# middleware server) serving as the core hub for protocol translation and connection management; and one or more lower-level execution layer nodes (such as standard PLC lower-level nodes) responsible for interacting with field devices and performing physical control. These three layers of nodes are connected via standard network protocols.

[0054] The host computer application layer node can include one or more graphical user interface (GUI) applications running on a general-purpose computer. Unlike traditional HMIs, this application layer is the intelligent decision-making core of the entire system, capable of human-computer interaction, algorithm execution and invocation, real-time monitoring, data transmission and communication requests, and I / O reading. It internally integrates and executes an innovative two-stage hybrid process parameter generation method, which includes a machine learning classification model and a deterministic generation algorithm. This application layer is responsible for receiving the user's input of the desired product target (rather than specific process parameters) and automatically generating precise, multi-stage numerical process instructions using this method. The host computer can also monitor the displacement and pressure during the forming process in real time and integrates the PLC's input / output points, allowing for intuitive display of errors.

[0055] The middleware gateway layer node can include one or more independently running server applications, such as a TCP server, which is used for protocol conversion, polling, data interaction, PLC driving, and data feedback. This middleware is key to implementing the decoupled architecture of this system. Logically, it plays a dual role: upwards, as a server, it receives hardware-independent process instruction sets encapsulated in high-level data formats (such as JSON) from the host application layer; downwards, as a client, it communicates with the lower-level execution layer using a proprietary communication protocol specific to the PLC brand (such as Mitsubishi MX Component or Siemens S7 protocol). The core function of this middleware is "compilation" and "translation," that is, accurately converting the abstract process instruction set from the upper layer into low-level register read / write operations that the lower-level PLC can recognize and execute.

[0056] The lower-level execution layer node can include any brand of standard programmable logic controller (PLC), which can be used for electrical system control, hydraulic system control, cylinder booster system control, and servo motor control. This PLC is responsible for the final physical I / O operations and real-time control logic. In this invention's architecture, the PLC does not communicate directly with the upper-level application layer; its only communication object is the middleware gateway layer node. This allows the PLC to be viewed as a standard, replaceable "black box" execution unit; changes in its brand and model will not affect the structure and code of the upper-level application.

[0057] The systems of this invention employ a layered and hardware-independent communication approach. Communication between the host computer application layer and the middleware gateway layer uses standard, open network protocols (such as TCP / IP) and data formats (such as JSON). Instruction and data addressing is business logic-oriented, rather than specific hardware physical addresses. For example, the host computer only sends a JSON object containing complete process parameters, without needing to know which specific D register or DB block address of the PLC these parameters will ultimately be written to. All hardware-related address mappings and protocol details are centrally processed and encapsulated by the middleware server.

[0058] The advantage of this communication method is that it allows the development of the host computer application layer to be completely independent of the underlying hardware. Developers can use modern high-level languages ​​such as Python and their rich algorithm libraries to focus on implementing complex business logic and intelligent algorithms, without having to worry about the communication details of the underlying PLC. This provides unprecedented convenience for rapid system iteration, functional expansion, and cross-platform integration.

[0059] Furthermore, the two-stage hybrid process parameter generation method of this invention is a core innovation deployed at the upper computer application layer. This method includes: First stage, a machine learning classifier: Based on a large amount of simulation or historical data, a classification model is trained to predict the optimal process based on the user-inputted desired product quality. This solves the complex, nonlinear strategic decision-making problem in process selection. Second stage, a deterministic generation algorithm: Receiving the "path category" output by the first-stage classifier, and combining it with the user-inputted global process objective (such as total displacement, total pressure), the algorithm accurately calculates the specific, multi-stage numerical instructions executable by the PLC using a preset deterministic mathematical paradigm corresponding to each category. This solves the tactical calculation problem of transforming abstract strategies into precise engineering values.

[0060] Based on the above system, the present invention provides a feasible implementation, comprising: a host computer application layer including a graphical user interface, a machine learning classification model, and a deterministic generation algorithm; this node is used to receive the desired product characteristics input by the user, determine a process strategy using the classification model, and then generate a set of numerical process instructions based on the process strategy using the deterministic generation algorithm; a lower-level execution layer node, which is a standard PLC, used to perform physical control; and a middleware gateway layer node, which is connected to the host computer via a first network protocol (TCP) and to the lower-level machine via a second network protocol (MX Component); wherein, the middleware server is used to receive the numerical process instructions from the host computer, encapsulated in a hardware-independent format (such as JSON), translate the instructions into low-level write operations for specific memory addresses that the lower-level machine can execute, and then send them to the lower-level machine via the second network protocol.

[0061] Secondly, the present invention also provides a control device for hydraulic forming, applied to a control method for hydraulic forming, such as... Figure 2 As shown, it includes: Training module 201 is used to obtain product quality indicators of pipe fittings with different geometric parameters under different process strategies; the geometric parameters, product quality indicators and corresponding process strategies are used as a dataset for machine learning to obtain a pre-trained machine learning classification model.

[0062] The acquisition module 202 is used to acquire the expected product quality indicators of the target product.

[0063] The classification module 203 is used to determine the corresponding process strategy by using the geometric parameters and the expected product quality indicators through a pre-trained machine learning classification model.

[0064] The generation module 204 is used to generate multi-segment process parameters based on the process strategy and its corresponding global process parameters using a corresponding deterministic generation algorithm; the deterministic generation algorithm is a calculation rule preset for the process strategy.

[0065] The control module 205 is used to generate control instructions for the hydraulic forming equipment to execute the forming process based on the multi-segment process parameters through a programmable logic controller.

[0066] The present invention also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described... Figure 1 The steps of the provided hydraulic forming control method.

[0067] This invention also provides a computer device. At the hardware level, the computer device includes a processor, an internal bus, a network interface, memory, and non-volatile memory, and may also include other hardware required for various operations. The processor reads the corresponding computer program from the non-volatile memory into the memory and then executes it to achieve the above-mentioned functions. Figure 1 The steps of the provided hydraulic forming control method.

[0068] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0069] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0070] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0071] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0072] It should be noted that the specific embodiments described above enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present invention has been described in detail in this specification, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are covered within the protection scope of the patent of the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A control method for hydraulic forming, characterized in that, A control system with a three-layer decoupled architecture is provided, comprising an upper-level application layer node, a lower-level execution layer node, and a middleware gateway layer node; the upper-level application layer node is configured to run an application program developed in a general-purpose programming language to generate multi-segment process parameters, and the lower-level execution layer node includes a programmable logic controller to execute specific equipment control logic; The middleware gateway layer node is used to implement the communication connection between the host computer application layer node and the lower-level machine execution layer node; the method includes: The host computer application layer node is used to acquire the geometric parameters and expected product quality indicators of the target product; the geometric parameters and expected product quality indicators are used to determine the corresponding process strategy through a pre-trained machine learning classification model; based on the process strategy and its corresponding global process parameters, multi-segment process parameters are generated through a corresponding deterministic generation algorithm; wherein, the deterministic generation algorithm is a calculation rule preset according to the process strategy; the dataset for pre-training the machine learning classification model includes: acquiring the product quality indicators of workpieces with different geometric parameters under different process strategies, and using the geometric parameters, product quality indicators and corresponding process strategies as the dataset; The upper computer application layer node sends multi-segment process parameters to the lower computer execution layer node through the middleware gateway layer node, and generates control instructions for the hydraulic forming equipment to execute the forming process through the programmable logic controller.

2. The method according to claim 1, characterized in that, The deterministic generation algorithm is a piecewise function of displacement and pressure, and its mathematical paradigm is as follows: S=F(Dtotal, Ptotal, Ctype); Where S is the output multi-segment displacement and pressure numerical sequence, D_total is the total displacement, P_total is the total pressure, and C_type is the process strategy; The algorithm performs precise calculations by invoking a preset mathematical paradigm based on the process strategy.

3. The method according to claim 2, characterized in that, The process strategy includes one or more of the following: linear loading, fast-then-slow loading, and plateau loading. Each process strategy corresponds to a different piecewise displacement and pressure function.

4. The method according to claim 1, characterized in that, The middleware gateway layer node is an intermediate server. The upper computer application layer node sends multi-segment process parameters to the lower computer execution layer node through the middleware gateway layer node. The control instructions for the hydraulic forming equipment to execute the forming process are generated by the programmable logic controller, including: The host computer application layer node encapsulates the multi-segment process parameters into a hardware-independent data format and sends them to the intermediate server; The intermediate server converts the hardware-independent data format into an instruction format that the programmable logic controller (PLC) can recognize, and then sends it to the target PLC. The target programmable logic controller executes the instructions to control the hydraulic forming equipment to complete the forming process.

5. The method according to claim 4, characterized in that, The hardware-independent data format is JSON, and it is transmitted to the intermediate server via the TCP / IP protocol.

6. The method according to claim 5, characterized in that, The intermediate server converts hardware-independent data formats into instruction formats recognizable by the programmable logic controller (PLC) and sends them to the target PLC, including: The intermediate server parses the JSON format data to extract process parameters and control commands. According to the preset address mapping table, the process parameters are converted into the target PLC register address and written. The converted instructions are sent to the PLC using a dedicated PLC communication protocol.

7. The method according to claim 6, characterized in that, When the target programmable logic controller executes the instruction, the intermediate server reads the status register data of the PLC in a polling manner, encapsulates the read real-time data into JSON format and feeds it back to the host computer. The host computer updates the monitoring information in the graphical interface according to the feedback data, thereby realizing closed-loop control.

8. A control device for hydraulic forming, characterized in that, The control method for hydroforming according to claim 1, the apparatus comprising: The training module is used to obtain product quality indicators of workpieces with different geometric parameters under different process strategies; the geometric parameters, product quality indicators and corresponding process strategies are used as a dataset for machine learning to obtain a pre-trained machine learning classification model. The acquisition module is used to acquire the expected product quality indicators of the target product. The classification module is used to determine the corresponding process strategy by using the geometric parameters and expected product quality indicators through a pre-trained machine learning classification model. The generation module is used to generate multi-segment process parameters based on the process strategy and its corresponding global process parameters using a corresponding deterministic generation algorithm; the deterministic generation algorithm is a calculation rule preset according to the process strategy. The control module is used to generate control instructions for the hydraulic forming equipment to execute the forming process based on multi-segment process parameters via a programmable logic controller.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method described in any one of claims 1 to 7.

10. A computer device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in any one of claims 1 to 7.

Images