Material performance and energy consumption collaborative optimization method and system for sheet steel member machining

Abstract

The invention relates to the technical field of industrial production management and optimization, and particularly discloses a material performance and energy consumption collaborative optimization method and system for sheet steel member processing, and the method comprises the steps: constructing a material performance prediction model and an energy consumption prediction model which are coupled with each other, and introducing a collaborative interaction layer to quantify the dynamic influence of the two models, thereby achieving the collaborative optimization of the material performance and energy consumption. An accurate collaborative prediction result is generated; meanwhile, on the basis of real-time machining data, optimal machining parameters are calculated through a collaborative objective function containing dynamic weights, and staged intelligent optimization is achieved; and finally, through real-time monitoring and feedback closed loop, the system can actively identify process deviation and trigger adaptive updating of the model, and long-term stable operation is ensured. According to the method, the energy utilization efficiency and the product quality consistency are remarkably improved, the problem that dynamic adjustment strategies are lacked in the prior art is solved, man-machine interaction and strategy-oriented hierarchical decision making are supported, and the method is suitable for the high-precision and low-energy-consumption machining requirements in the fields of automobile manufacturing, aerospace and the like.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of industrial production management and optimization, and relates to a material performance and energy consumption collaborative optimization method and system for sheet steel component processing. BACKGROUND

[0002] Sheet steel components are widely used in modern industrial fields such as automobile manufacturing, aerospace, household appliances, etc. due to their lightweight and high-strength characteristics. In the processing of these components, how to accurately control the final performance of the material, such as strength, hardness and forming accuracy, through processes such as hot forming and cold stamping, while effectively managing and reducing energy consumption in the processing, is a key indicator of the core competitiveness of manufacturing and a focus of continuous attention in the field of industrial production management.

[0003] Currently, in the processing practice of sheet steel components, the management of material performance and energy consumption is usually in a relatively separate manner. On the one hand, process engineers rely on experience or offline finite element simulation to set a fixed set of processing parameters in order to achieve the preset material performance target. On the other hand, energy management is achieved by independent monitoring of equipment or the use of energy-saving equipment. In some more advanced schemes, although optimization algorithms are introduced, they are mostly for single targets, such as establishing an energy consumption prediction model to minimize energy consumption, or establishing a performance prediction model to maximize performance, with little effective correlation and collaboration between the two.

[0004] Therefore, the existing technical solutions have some inherent defects. First, since material performance and energy consumption are managed as two isolated variables, the strong coupling relationship between the two in the physical process is ignored, resulting in one-sided optimization decisions and difficulty in finding the true global optimal solution. Second, relying on fixed process parameters or static models for production control makes the system unable to cope with real-time dynamic disturbances such as raw material batch differences, equipment wear, and environmental temperature changes, resulting in poor product quality consistency and unnecessary waste of energy. Third, the existing optimization methods lack the ability to dynamically adjust strategies in the entire processing flow, and cannot focus on optimization direction according to the core target of different processing stages. SUMMARY

[0005] In view of this, in order to solve the problems raised in the background art, a material performance and energy consumption collaborative optimization method and system for sheet steel component processing are proposed.

[0006] The purpose of the present application can be achieved by the following technical solutions: The first aspect of the present application provides a material performance and energy consumption collaborative optimization method for sheet steel component processing, comprising: S1, obtaining real-time processing parameters and material initial state parameters in the processing of sheet steel components, and preprocessing the real-time processing parameters and material initial state parameters to generate preprocessed data.

[0007] S2. Input the preprocessed data into the mutually coupled material performance prediction model and energy consumption prediction model. Through the collaborative interaction layer used to characterize the interaction between the material performance prediction model and the energy consumption prediction model, generate a collaborative prediction result containing the predicted values ​​of material performance and energy consumption.

[0008] S3. Based on the collaborative prediction results, the processing parameters are optimized by calculating the collaborative objective function, which includes material performance weights and energy consumption weights that are dynamically adjusted according to the processing stage.

[0009] S4. Based on the optimized processing parameters, generate and send control commands to the processing equipment to execute the processing operation.

[0010] The second aspect of the present invention provides a material performance and energy consumption synergistic optimization system for the processing of thin plate steel components, comprising: a data acquisition and preprocessing module, which acquires real-time processing parameters and initial material state parameters during the processing of thin plate steel components, and preprocesses the real-time processing parameters and initial material state parameters to generate preprocessed data.

[0011] The dynamic model building module inputs preprocessed data into the mutually coupled material performance prediction model and energy consumption prediction model. Through a collaborative interaction layer used to characterize the interaction between the material performance prediction model and the energy consumption prediction model, it generates a collaborative prediction result containing material performance prediction values ​​and energy consumption prediction values.

[0012] The processing parameter collaborative optimization module calculates and optimizes processing parameters based on collaborative prediction results through a collaborative objective function, which includes material performance weights and energy consumption weights that are dynamically adjusted according to the processing stage.

[0013] The machining control and execution module generates and sends control commands to the machining equipment to execute machining operations based on optimized machining parameters.

[0014] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention realizes the comprehensive management and optimization of production resources and product quality by constructing a coupled prediction model of performance and energy consumption and a dynamic collaborative optimization mechanism. It no longer treats material performance and energy consumption as two independent management objectives and separates them, but reveals the intrinsic relationship between the two through a collaborative interaction layer, so that the optimization decision can be based on a holistic perspective that is closer to physical reality, thereby significantly improving energy utilization efficiency while ensuring product performance, and realizing the improvement of overall operational efficiency.

[0015] This invention establishes a closed loop of real-time monitoring and feedback, enabling the system to proactively identify process deviations caused by factors such as raw material fluctuations and equipment status changes during production, and triggering a model self-learning and update mechanism. This continuous self-improvement capability allows the optimization strategy to dynamically adapt to complex industrial environments, ensuring long-term operational stability and reliability, and reducing production risks and quality problems caused by uncertainties.

[0016] This invention effectively combines intelligent decision-making with human experience, enhancing the flexibility of production management. This method not only automatically switches optimization priorities according to processing stages to achieve refined process management, but also provides a human-machine interface, allowing managers to set strategic optimization directions based on macro-level production plans or market demands. This hierarchical decision-making mechanism enables the system to autonomously execute optimal tactical adjustments while adhering to high-level management intentions, thereby making the production process more agile and efficient. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, 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.

[0018] Figure 1 This is a schematic diagram of the method steps of the present invention.

[0019] Figure 2 This is a schematic diagram of the system structure connection of the present invention. Detailed Implementation

[0020] The technical solutions of 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 Figure 1 The first aspect of the present invention provides a method for synergistic optimization of material properties and energy consumption in the processing of thin plate steel components, comprising: S1, acquiring real-time processing parameters and initial material state parameters during the processing of thin plate steel components, and preprocessing the real-time processing parameters and initial material state parameters to generate preprocessed data.

[0022] In a specific embodiment of the present invention, the specific steps of preprocessing real-time processing parameters and initial material state parameters to generate preprocessed data include: acquiring real-time processing parameters through a sensor group deployed on the processing equipment.

[0023] Retrieve the initial state parameters of the material from the material information system.

[0024] It should be noted that, in order to generate standardized preprocessed data that can be used in subsequent models, this method first requires the systematic collection and processing of raw data from different sources and in various formats. This process begins in the data acquisition stage, by deploying sensor groups at key locations on the thin-plate steel component processing equipment. These sensor groups include, for example, temperature sensors, current and voltage sensors, laser power meters, and displacement sensors, to continuously capture dynamic information during the processing. This information constitutes real-time processing parameters, such as welding speed, laser energy, and cooling rate. Simultaneously, for each thin-plate steel component to be processed, its inherent physical and chemical properties, i.e., the initial state parameters of the material, such as the steel plate grade, thickness, chemical composition, and initial yield strength, are queried and retrieved from the factory's material information system. The material information system is a database that stores and manages the specifications, batches, and performance indicators of production materials.

[0025] The acquired real-time processing parameters and initial material state parameters are subjected to data filtering and normalization to generate preprocessed data.

[0026] It should be noted that after acquiring the raw data, data filtering and normalization are required to eliminate interference from industrial environmental noise and address the issue of inconsistent dimensions in multi-source data. Data filtering aims to improve the signal-to-noise ratio, remove abnormal data points caused by equipment vibration or electromagnetic interference, and retain effective information that truly reflects the processing status. Subsequently, the filtered real-time processing parameters and initial material state parameters are normalized. This step is to eliminate the influence of different physical dimensions on the model training process and prevent parameters with large numerical ranges from dominating the model. Normalization can be achieved using the min-max normalization method, mapping all parameter values ​​to a unified interval, typically [0,1]. This process can be implemented using the following formula: ,in, This represents the normalized data value, which is a component of the final preprocessed data. This represents the original measured value after filtering for a specific parameter. and These represent the minimum and maximum values ​​of the parameter in historical datasets or preset ranges, respectively. These two values ​​are derived from statistical analysis of a large amount of historical processing data and serve as a normalization benchmark. Through the above steps, the scattered and heterogeneous raw data are transformed into structured and standardized preprocessed data, providing high-quality data input for the subsequent accurate and coordinated prediction of material properties and energy consumption.

[0027] This method significantly improves the quality and consistency of input data by integrating real-time sensor data with static material information and employing data filtering and normalization techniques. Data filtering ensures the authenticity and reliability of the data used in the model, effectively avoiding prediction bias caused by noise or outliers and enhancing the robustness of the entire optimization system. Normalization solves the problem of difficult model training or slow convergence speed caused by different dimensions of multi-source heterogeneous data, ensuring that features of different dimensions contribute equally to the model, thereby improving the training efficiency and prediction accuracy of the prediction model. Ultimately, this systematic data preprocessing workflow lays a solid data foundation for achieving accurate prediction and collaborative optimization of material properties and energy consumption during the processing of thin-plate steel components, and is a key prerequisite for ensuring the effective operation of the entire optimization method.

[0028] S2. Input the preprocessed data into the mutually coupled material performance prediction model and energy consumption prediction model. Through the collaborative interaction layer used to characterize the interaction between the material performance prediction model and the energy consumption prediction model, generate a collaborative prediction result containing the predicted values ​​of material performance and energy consumption.

[0029] In a specific embodiment of the present invention, the specific steps for generating a collaborative prediction result containing material performance prediction values ​​and energy consumption prediction values ​​include: inputting preprocessed data into the material performance prediction model and the energy consumption prediction model respectively to generate initial material performance prediction values ​​and initial energy consumption prediction values.

[0030] It should be noted that, to generate a collaborative prediction result that includes both material performance predictions and energy consumption predictions, this method first splits the preprocessed data into two paths, feeding them separately into two parallel, pre-trained prediction models: a material performance prediction model and an energy consumption prediction model. The material performance prediction model, for example, is a deep learning-based neural network that outputs initial material performance predictions based on the input processing parameters and initial material state parameters. These initial material performance predictions may be quantitative predictions of key mechanical properties such as yield strength and elongation. Similarly, the energy consumption prediction model may be a gradient boosting regression tree model that outputs initial energy consumption predictions, which represent the estimated electrical energy consumed to complete the current process.

[0031] Based on the real-time processing parameters in the preprocessed data, a dynamic influence factor is calculated to quantify the degree of interaction between energy consumption and performance.

[0032] It should be noted that the initial predicted material properties and initial predicted energy consumption were generated under the assumption that performance and energy consumption were independent of each other, failing to reflect the strong coupling relationship between the two in the physical process. Therefore, the core of this method lies in introducing a dynamic influence factor to quantify this coupling relationship. This dynamic influence factor is not a constant value, but is calculated in real-time based on the real-time processing status contained in the preprocessed data, such as the instantaneous values ​​of the current workpiece temperature, processing speed, and laser power. These real-time processing parameters directly determine the efficiency of the energy input's influence on the evolution of the material's microstructure, thus determining the strength of the interaction between energy consumption and performance. The calculation of the dynamic influence factor aims to quantify the marginal effect of small changes in energy consumption on performance, and conversely, the marginal increase in energy demand for pursuing higher performance targets.

[0033] It should also be noted that the formula for calculating the dynamic impact factor used to quantify the degree of interaction between energy consumption and performance is as follows: , ,in, and These are two dynamic influencing factors calculated based on real-time processing status parameters. This represents the change in performance caused by a change in unit energy consumption under current operating conditions. This represents the change in additional energy consumption required to achieve a unit performance improvement. , and These represent the instantaneous values ​​of the current workpiece temperature, laser power, and processing speed, respectively. , and These represent the reference values ​​for workpiece temperature, instantaneous laser power, and processing speed, as set in the process specifications. , , , , and These are weighting coefficients, determined through the experience of process experts. and All of these adjustments involve setting a baseline value to ensure that the predictive performance remains within a reasonable range. Defined as a relative proportion of the initial performance predictions, typically with a value of 0.1. Also defined as a relative proportion of the initial energy consumption prediction, with a typical value of 0.05.

[0034] Through the collaborative interaction layer, the initial material performance prediction values ​​and initial energy consumption prediction values ​​are corrected using dynamic influence factors, thereby generating collaborative prediction results.

[0035] It should be noted that after calculating the dynamic influence factor, the initial prediction value is corrected through a collaborative interaction layer. This collaborative interaction layer is an algorithm module, which essentially couples and fuses information between two prediction models. It uses the dynamic influence factor to adjust the isolated initial prediction, generating the final collaborative prediction result. This correction process can be described by the following set of coupling equations: ,in, and These represent the corrected predicted values ​​for synergistic material performance and synergistic energy consumption, respectively, which together constitute the synergistic prediction result. and These are the initial predicted material properties and initial predicted energy consumption values ​​output by the model; and These represent the initial predicted material properties and the initial predicted energy consumption of the current processing stage, respectively, as output by the model. and These represent the reference energy consumption baseline value and the reference performance baseline value for the current processing stage, respectively. These baseline values ​​can be obtained directly from the process specifications. Through this set of equations, the collaborative interaction layer quantifies the prediction deviation of energy consumption into its impact on performance, and quantifies the prediction deviation of performance into its impact on energy consumption, thereby outputting a collaborative prediction result that is closer to the physical reality.

[0036] This method, by constructing a collaborative interaction layer and introducing dynamic influencing factors, achieves a leap from two independent prediction models to a coupled prediction system. Its technical advantage lies in its ability to accurately capture the inherent correlation and dynamic balance between material properties and energy consumption—two objectives—in isolation, rather than viewing them in isolation. This mechanism allows the prediction results to reflect the combined impact of changes in processing parameters on both objectives. For example, while increasing laser power improves welding strength, it inevitably leads to increased energy consumption, and this gain-cost relationship dynamically changes at different processing stages. Therefore, the collaborative prediction results generated by this method are more accurate and comprehensive than single, independent predictions, providing high-quality, high-fidelity input for subsequent collaborative optimization decisions, greatly improving the reliability and practicality of the optimization results.

[0037] S3. Based on the collaborative prediction results, the processing parameters are optimized by calculating the collaborative objective function, which includes material performance weights and energy consumption weights that are dynamically adjusted according to the processing stage.

[0038] In a specific embodiment of the present invention, the current processing stage is identified based on the processing progress information in the preprocessed data.

[0039] It should be noted that, in order to calculate the optimal combination of processing parameters, this method first requires accurate identification of the current processing stage based on real-time data. This process relies on continuous analysis of processing progress information in the preprocessed data, which may include processing timestamps, the position of the welded joint in the workpiece coordinate system, and the length of the completed processing path. The system compares this real-time processing progress information with a preset process flow chart, which divides the entire processing of the thin-plate steel component into several stages with different process requirements: the initial processing stage, the critical processing stage, and the final processing stage. When the real-time progress information enters a preset segment, the system automatically identifies the current processing stage.

[0040] Based on the identified current processing stage, material performance weights and energy consumption weights are selected and set in the collaborative objective function from the weight strategy library.

[0041] In a specific embodiment of the present invention, the specific steps of selecting and setting the material performance weight and energy consumption weight in the collaborative objective function from the weight strategy library include: if the current processing stage is the initial processing stage, then the energy consumption weight is set higher than the material performance weight.

[0042] It should be noted that this method employs a refined adjustment mechanism combining a phased strategy and real-time fine-tuning when selecting and setting weights in the collaborative objective function from the weight strategy library. The weight strategy library is a set of rules or data pre-defined by process experts, defining a set of optimal material performance weights and energy consumption weights for each processing stage. First, based on the identification results of the current processing stage, the system performs differentiated weight settings. If the system determines that it is currently in the early stages of processing, such as preheating or connecting non-load-bearing areas, the main objective of this stage is to quickly and stably establish processing conditions, with minimal impact on the ultimate performance of the final component. Therefore, the system will retrieve a weight configuration from the weight strategy library that emphasizes economy, setting the energy consumption weight at a higher value and the material performance weight at a lower value, to guide the optimization algorithm to prioritize finding the processing parameter scheme with the lowest energy consumption while meeting basic process requirements.

[0043] If the current processing stage is a critical stage, then the weight of material performance is set higher than the weight of energy consumption.

[0044] It should be noted that, conversely, when the processing progresses to critical stages, such as welding the core load-bearing welds or precisely controlling the heat-affected zone, the final service performance of the component is determined at this stage. At this point, ensuring material properties, such as yield strength and fatigue resistance, becomes the primary task. The system will accordingly select another set of weights from the weight strategy library, setting the material performance weights significantly higher than the energy consumption weights. This ensures that the collaborative objective function, during iterative optimization, prioritizes meeting or even exceeding performance design targets, even if it means incurring higher energy costs.

[0045] Based on the deviation between real-time monitoring data and collaborative prediction results, the currently set weights are fine-tuned in real time to dynamically respond to process fluctuations.

[0046] It should be noted that relying solely on preset stage weights is insufficient to handle real-time fluctuations during processing, such as batch differences in materials or changes in ambient temperature. Therefore, this method introduces a closed-loop real-time fine-tuning mechanism. This mechanism continuously compares real-time monitoring data—the actual processing status and energy consumption values ​​collected by the sensor array—with the collaborative prediction results output by the dynamic model building module, calculating the deviation between the two. Based on this deviation, the system dynamically and incrementally adjusts the currently set weights. This adjustment process can be achieved through the following formula: , ,in, This represents the amount of adjustment to the weight. It is a preset adjustment coefficient that determines the system's response sensitivity to deviations. This value is set through experimental calibration or expert experience, and a typical value can be 0.5. These are the actual values ​​of real-time material performance indicators obtained through indirect measurement or online testing methods. These are the predicted material properties from the collaborative prediction results. It is a reference performance benchmark value used for normalizing deviations. and These are the material property weights and energy consumption weights, after fine-tuning, ultimately used for calculating the collaborative objective function. ,and and This refers to the baseline weight for the current stage, obtained from the weighting strategy library. This formula ensures that when actual performance falls short of expectations, the system automatically increases the material performance weight, and vice versa, thereby dynamically responding to process fluctuations.

[0047] A weighted collaborative objective function is used to generate optimized processing parameters through iterative calculation.

[0048] It should be noted that the material performance weights and energy consumption weights ultimately used in the collaborative objective function are employed to construct a dynamic collaborative objective function, which is the core of achieving a performance-energy consumption balance decision. This collaborative objective function aims to find a set of processing parameters that optimizes the function's evaluation value. Its mathematical expression can be constructed as follows: In this function, The evaluation value represents the collaborative objective function. Finally, iterative calculations are used to search for and generate optimized processing parameters. This process can employ intelligent optimization algorithms, such as genetic algorithms or particle swarm optimization algorithms. These algorithms use the processing parameters to be optimized as variables, iteratively searching by repeatedly generating parameter combinations, evaluating them using predictive models and objective functions, and continuously adjusting the parameter combinations based on the evaluation results, until a set of parameters that satisfies the objective function is found. The parameter combination that minimizes the value. This set of parameters is the final output optimized processing parameters, which represent the best balance between material properties and energy consumption targets at the current processing stage.

[0049] This method achieves intelligent and contextualized optimization objectives by introducing a weighting strategy that dynamically adjusts according to the processing stage. Its technical advantage lies in making the optimization process no longer static and unchanging, but rather flexibly adjusting the optimization focus based on the specific needs of different processing stages. In stages with low performance requirements, the system can focus on energy conservation, while in critical structural parts, it prioritizes ensuring that material performance meets standards. This adaptive optimization mechanism ensures more rational resource allocation throughout the entire processing process, avoiding unnecessary energy waste in pursuit of excessive performance in non-critical stages, or sacrificing necessary performance for excessive energy conservation in critical stages. Thus, it achieves the optimal balance between overall energy efficiency and processing quality while ensuring the quality of the final product.

[0050] S4. Based on the optimized processing parameters, generate and send control commands to the processing equipment to execute the processing operation.

[0051] In a specific embodiment of the present invention, after the step of generating and sending control commands to the processing equipment to perform processing operations, the method further includes: during the execution of processing operations, collecting actual operating data of the processing equipment through a sensor group to form real-time monitoring data.

[0052] It should be noted that after the processing control and execution module generates and sends control commands to the processing equipment based on optimized processing parameters, this method does not terminate its workflow. Instead, it initiates a continuous closed-loop monitoring and data accumulation process. Throughout the entire processing operation, the sensor array deployed on the processing equipment continuously works, collecting the actual operating data of the processing equipment without interruption. This data constitutes real-time monitoring data, the content of which is similar to the real-time processing parameters collected in step S1, including but not limited to the actual laser power output, the moving speed of the welding head, the real-time temperature of the workpiece surface, and the instantaneous power consumption of the equipment. This data truly reflects the execution of control commands on the physical equipment.

[0053] By comparing real-time monitoring data with collaborative prediction results, deviation analysis data is generated.

[0054] It should be noted that after obtaining real-time monitoring data, the system compares it point-by-point or time-by-time with the collaborative prediction results previously generated by the dynamic model building module. The collaborative prediction results are the expected performance and energy consumption values ​​predicted based on the processing parameters at the time of optimization, while the real-time monitoring data represents the actual effect produced by the optimized parameters. By comparing these two sets of data, the system can quantify the deviation between prediction and reality. This comparison process generates a series of deviation analysis data, such as the difference between actual energy consumption and predicted energy consumption. This process can be represented by the following formula: , ,in, and These represent deviation analysis data for material performance-related indicators and energy consumption, respectively. and These are actual performance index values ​​and actual energy consumption values ​​extracted from real-time monitoring data. This deviation analysis data not only includes the magnitude of the deviation but also includes timestamps and corresponding workpiece location information, constituting a complete evaluation of the optimization effect.

[0055] The generated deviation analysis data is stored and used for subsequent system adaptive adjustments.

[0056] It should be noted that, finally, the system will store these generated deviation analysis data, along with related real-time monitoring data, the collaborative prediction results at the time, and the optimized processing parameters used, as a complete data record unit in a structured manner. This data is archived in a historical database and tagged for subsequent system adaptive adjustments. This database continues to grow as processing tasks are carried out, forming a valuable knowledge base that records the interaction relationships and deviation patterns between the prediction model, optimization algorithm, and actual physical process under different operating conditions.

[0057] This method constructs a complete "prediction-optimization-execution-feedback" cycle by adding a closed-loop feedback mechanism for real-time monitoring, comparative analysis, and data storage after each processing operation. Its technical advantage lies in enabling the entire collaborative optimization system to learn and evolve. By continuously collecting deviation data between predictions and actual results, the system provides crucial nourishment for its iterative upgrades. This deviation analysis data serves as a direct basis for evaluating and improving the accuracy of the prediction model and the effectiveness of the optimization algorithm, providing a quantitative foundation for subsequent model retraining, collaborative interaction layer algorithm correction, and updates to the weight strategy library. This not only enhances the system's ability to monitor and trace the current processing process, but more importantly, it lays the data foundation for achieving long-term, adaptive performance improvements, enabling the system to become increasingly "intelligent" and accurate with the accumulation of production experience.

[0058] In a specific embodiment of the present invention, the following step is also included: acquiring real-time monitoring data during the processing.

[0059] Based on real-time monitoring data, it is determined whether the actual processing state deviates from the target state range. If it does, a model update trigger signal is generated.

[0060] It should be noted that this method integrates a model adaptive update mechanism to ensure the long-term effectiveness and accuracy of the prediction model. This mechanism begins with continuous monitoring of the processing. The system continuously acquires real-time monitoring data during the processing, collected by sensor arrays deployed on the equipment, which accurately reflects the actual state of material response and equipment energy consumption. Next, the system compares this real-time monitoring data with a preset target state range to determine if there is a significant deviation from the actual processing state. The preset target state range is not a fixed value, but a dynamic interval defined by process experts based on design requirements and historical experience. It specifies the reasonable limits for fluctuations in key performance indicators (such as temperature and melt pool size) and energy consumption indicators under normal operating conditions. When the system detects that a key real-time monitoring data point continuously or significantly exceeds its corresponding preset target state range—for example, if the actual cooling rate is far below the set lower limit, potentially leading to unexpected metallographic structures—the system determines that a state deviation has occurred. Once a deviation is determined, the system immediately generates a model update trigger signal. The above judgment logic can be expressed as: In this judgment logic, This represents the state of the model update trigger signal. If this expression is true, a trigger signal is generated. These are the actual values ​​of key status parameters extracted from real-time monitoring data. It is the target center value of this key state parameter, and This refers to the allowable deviation threshold around the target center value. and Together, they defined the target state range. This allowable deviation threshold... It is set according to the requirements of process stability, and the typical value can be ±3%.

[0061] In response to the model update trigger signal, the model update process is initiated to retrain the material property prediction model and the energy consumption prediction model.

[0062] In a specific embodiment of the present invention, the specific steps of initiating the model update process include: integrating real-time monitoring data and historical data that cause deviations to form an updated training dataset; It's important to note that initiating the model update process after responding to the model update trigger signal involves a series of sophisticated steps. First, the system needs to prepare the dataset for model retraining. It locates and extracts the segment of real-time monitoring data that triggered the update, detailing the entire process of the processing state deviating from the target state range. Subsequently, the system integrates this new data reflecting the process changes with a large amount of historical data stored in the historical database. This integration is not a simple patchwork, but rather a strategic construction of an updated training dataset that includes both historical data representing normal operating conditions and deviation data characterizing new operating conditions, ensuring the comprehensiveness of the model update.

[0063] The material property prediction model and energy consumption prediction model were fine-tuned using the updated training dataset.

[0064] It's important to note that, next, using this updated training dataset, the system fine-tunes the existing material performance prediction model and energy consumption prediction model. Fine-tuning is an efficient model update technique; it doesn't retrain the entire model from scratch, but rather makes small adjustments to the model's parameters using new data while preserving the model's existing knowledge. For deep learning-based models, this typically means freezing most of the lower network layers and retraining only a few top layers. This allows for rapid adaptation to new data while preventing catastrophic forgetting—the model forgetting old knowledge while learning new knowledge.

[0065] Based on the discrepancy between real-time monitoring data and collaborative prediction results, the algorithm used to calculate the dynamic influence factor in the collaborative interaction layer is modified to enhance the synergistic effect between models.

[0066] It's important to note that the final and crucial step in the model update process is revising the algorithm for the collaborative interaction layer. This layer acts as a bridge between the material performance prediction model and the energy consumption prediction model, and its core function is calculating the dynamic impact factor. The system utilizes previously stored deviation analysis data—the discrepancy between real-time monitoring data and the collaborative prediction results—to reflect on and revise the algorithm used to calculate the dynamic impact factor. For example, if the system finds that under a certain operating condition, the actual energy consumption's impact on performance is always systematically greater than or less than the model's prediction, this indicates a deviation in the function or rule used to calculate the dynamic impact factor. The system will then revise the function's parameters or rule based on the statistical regularity of this deviation. This revision process can be expressed as iterative optimization of the parameters in the dynamic impact factor calculation formula: ,in, It is the first of the modified dynamic impact factor calculation functions. One parameter; It is the first The original values ​​of each parameter; It is the learning rate, which controls the step size of the correction. Based on expert experience, a typical value of 0.1 can be obtained. It is a loss function used to quantify the overall deviation between the collaborative prediction results and the real-time monitoring data. This is the loss function with respect to the parameters. The gradient indicates the direction of parameter adjustment that can reduce prediction bias. Through this gradient-based optimization, the system can automatically adjust the internal logic of the collaborative interaction layer, thereby enhancing the synergistic effect between models and enabling the two corrected models to work together more closely and accurately.

[0067] It should also be further explained that the loss function The formula is: ,in, and These are weighting coefficients that reflect the relative importance of material properties and energy consumption deviations. They can be set at critical stages of processing through process target setting. =0.7、 =0.3, which can be set in the early stages of processing. =0.4、 =0.6.

[0068] This method significantly enhances the robustness and long-term adaptability of the entire collaborative optimization system by introducing an adaptive model update mechanism based on state deviation. Its technical advantage lies in endowing the system with a self-correcting and evolving capability. Equipment aging, material batch changes, or alterations in environmental conditions can all cause existing prediction models to gradually become ineffective. This method can proactively identify this "model drift" phenomenon through real-time monitoring. Once identified, the system triggers a self-update, absorbing new data and learning new process patterns, thereby ensuring that the material performance prediction model and energy consumption prediction model always maintain a high-fidelity description of the actual processing process. This avoids inaccurate optimization decisions due to model obsolescence, guaranteeing the continuous effectiveness and reliability of the collaborative optimization method during long-term operation.

[0069] Reference Figure 2 The second aspect of the present invention provides a material performance and energy consumption synergistic optimization system for the processing of thin plate steel components, comprising: a data acquisition and preprocessing module, a dynamic model construction module, a processing parameter synergistic optimization module, and a processing control and execution module.

[0070] The data acquisition and preprocessing module is connected to the dynamic model building module, the dynamic model building module is connected to the processing parameter collaborative optimization module, and the processing parameter collaborative optimization module is connected to the processing control and execution module.

[0071] The data acquisition and preprocessing module acquires real-time processing parameters and initial material state parameters during the processing of thin plate steel components, and preprocesses the real-time processing parameters and initial material state parameters to generate preprocessed data.

[0072] The dynamic model construction module inputs preprocessed data into the mutually coupled material performance prediction model and energy consumption prediction model, and generates a collaborative prediction result containing material performance prediction values ​​and energy consumption prediction values ​​through a collaborative interaction layer used to characterize the interaction between the material performance prediction model and the energy consumption prediction model.

[0073] The processing parameter collaborative optimization module calculates and optimizes processing parameters based on collaborative prediction results through a collaborative objective function, wherein the collaborative objective function includes material performance weights and energy consumption weights that are dynamically adjusted according to the processing stage.

[0074] The processing control and execution module generates and sends control commands to the processing equipment to execute processing operations based on optimized processing parameters.

[0075] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for synergistic optimization of material properties and energy consumption in the processing of thin-plate steel components, characterized in that, include: S1. Obtain real-time processing parameters and initial material state parameters during the processing of thin plate steel components, and preprocess the real-time processing parameters and initial material state parameters to generate preprocessed data; S2. Input the preprocessed data into the mutually coupled material performance prediction model and energy consumption prediction model. Through the collaborative interaction layer used to characterize the interaction between the material performance prediction model and the energy consumption prediction model, generate a collaborative prediction result containing material performance prediction values ​​and energy consumption prediction values. S3. Based on the collaborative prediction results, the processing parameters are optimized by calculating the collaborative objective function, which includes material performance weights and energy consumption weights that are dynamically adjusted according to the processing stage. S4. Based on the optimized processing parameters, generate and send control commands to the processing equipment to execute the processing operation.

2. The method for synergistic optimization of material properties and energy consumption in the processing of thin-plate steel components according to claim 1, characterized in that, The specific steps for preprocessing real-time processing parameters and initial material state parameters to generate preprocessed data include: Real-time processing parameters are acquired through a sensor array deployed on the processing equipment; Retrieve the initial state parameters of the material from the material information system; The acquired real-time processing parameters and initial material state parameters are subjected to data filtering and normalization to generate preprocessed data.

3. The method for synergistic optimization of material properties and energy consumption in the processing of thin-plate steel components according to claim 2, characterized in that, The specific steps for generating a collaborative prediction result that includes predicted material properties and predicted energy consumption include: The preprocessed data is input into the material performance prediction model and the energy consumption prediction model respectively to generate initial material performance prediction values ​​and initial energy consumption prediction values. Based on the real-time processing parameters in the preprocessed data, a dynamic influence factor is calculated to quantify the degree of interaction between energy consumption and performance. Through the collaborative interaction layer, the initial material performance prediction values ​​and initial energy consumption prediction values ​​are corrected using dynamic influence factors, thereby generating collaborative prediction results.

4. The method for synergistic optimization of material properties and energy consumption in the processing of thin-plate steel components according to claim 2, characterized in that, The specific steps for calculating and optimizing processing parameters using a collaborative objective function include: The current processing stage is identified based on the processing progress information in the preprocessed data; Based on the identified current processing stage, select and set the material performance weights and energy consumption weights in the collaborative objective function from the weight strategy library; A weighted collaborative objective function is used to generate optimized processing parameters through iterative calculation.

5. The method for synergistic optimization of material properties and energy consumption in the processing of thin-plate steel components according to claim 4, characterized in that, The specific steps for selecting and setting the material performance weights and energy consumption weights in the collaborative objective function from the weight strategy library include: If the current processing stage is the initial processing stage, then the energy consumption weight is set higher than the material performance weight. If the current processing stage is a critical processing stage, then the weight of material performance is set higher than that of energy consumption. Based on the deviation between real-time monitoring data and collaborative prediction results, the currently set weights are fine-tuned in real time to dynamically respond to process fluctuations.

6. The method for synergistic optimization of material properties and energy consumption in the processing of thin-plate steel components according to claim 3, characterized in that, After the step of generating and sending control commands to the processing equipment to execute the processing operation, the following steps are also included: During the processing operation, the actual operating data of the processing equipment is collected by the sensor group to form real-time monitoring data; Real-time monitoring data is compared with collaborative prediction results to generate deviation analysis data; The generated deviation analysis data is stored and used for subsequent system adaptive adjustments.

7. The method for synergistic optimization of material properties and energy consumption in the processing of thin-plate steel components according to claim 3, characterized in that, It also includes the following steps: Acquire real-time monitoring data during the processing; Based on real-time monitoring data, determine whether the actual processing state deviates from the target state range. If it does, generate a model update trigger signal. In response to the model update trigger signal, the model update process is initiated to retrain the material property prediction model and the energy consumption prediction model.

8. The method for synergistic optimization of material properties and energy consumption in the processing of thin-plate steel components according to claim 7, characterized in that, The specific steps of the initiation model update process include: Integrate real-time monitoring data and historical data that cause deviations to form an updated training dataset; The material property prediction model and energy consumption prediction model were fine-tuned using the updated training dataset; Based on the discrepancy between real-time monitoring data and collaborative prediction results, the algorithm used to calculate the dynamic influence factor in the collaborative interaction layer is modified to enhance the synergistic effect between models.

9. A system for synergistic optimization of material properties and energy consumption in the processing of thin-plate steel components, characterized in that, include: The data acquisition and preprocessing module acquires real-time processing parameters and initial material state parameters during the processing of thin plate steel components, and preprocesses the real-time processing parameters and initial material state parameters to generate preprocessed data; The dynamic model building module inputs preprocessed data into the mutually coupled material performance prediction model and energy consumption prediction model. Through the collaborative interaction layer used to characterize the interaction between the material performance prediction model and the energy consumption prediction model, it generates a collaborative prediction result containing material performance prediction values ​​and energy consumption prediction values. The processing parameter collaborative optimization module calculates and optimizes processing parameters based on collaborative prediction results and through a collaborative objective function, which includes material performance weights and energy consumption weights that are dynamically adjusted according to the processing stage. The machining control and execution module generates and sends control commands to the machining equipment to execute machining operations based on optimized machining parameters.

Images