Titanium-based alloy component optimization and performance prediction intelligent management and control system

Abstract

The application discloses a kind of titanium base alloy component optimization and performance prediction intelligent management and control system, it is related to performance prediction management and control technical field.The system includes cross-scale structure-effect mapping module, intelligent optimization and prediction module and standard adaptation and interaction module.The application calibrates digital twin model in real time through cross-scale structure-effect mapping module, generates feedback control signal to realize process dynamic adjustment through intelligent optimization and prediction module, and with the aid of standard adaptation and interaction module, the high consistency of two-way transmission of whole-process data is guaranteed, finally, titanium base alloy full closed-loop intelligent management and control from component to performance prediction is built, the problem that titanium base alloy component optimization and performance prediction intelligent management and control efficiency is low due to the fuzzy multi-scale mechanism of structure-effect mapping and insufficient modeling precision in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of performance prediction and control technology, and in particular to an intelligent control system for titanium-based alloy composition optimization and performance prediction. Background Technology

[0002] At the structure-property mapping level, the cross-scale coupling mechanism has not been fully revealed. There are gaps in our understanding of the quantitative evolution laws between the atomic scale (such as the interaction between solid-solution atoms and dislocations), the mesoscale, and macroscopic properties (strength, plasticity, and toughness). Materials science lacks a unified theoretical framework for the complex interactions between multiple elements (such as Al, V, and Mo) and interstitial elements (such as O and N) in titanium-based alloys, as well as the atomic-level physicochemical mechanisms of defect initiation and evolution (such as dislocations and twins). For example, the embrittlement mechanism that leads to a sharp drop in plasticity when the oxygen content exceeds a critical value (such as >0.2 wt.%) and its suppression principle for deformation twinning are still unclear, making it difficult to predict phase transformation and defect evolution. Traditionally, physical descriptors designed manually based on expert experience (such as mole fraction and equivalence ratio) have low information dimensionality and limited characterization capabilities, and cannot fully capture the above-mentioned complex nonlinear, high-dimensional, multi-scale structure-property relationships. Under the constraint of scarce high-quality experimental data (small sample size), machine learning models are prone to overfitting or a sharp drop in prediction accuracy.

[0003] At the intelligent model level, the generalization ability of data-driven models is severely insufficient. Due to the significant difference in data distribution between training data and target scenarios, model prediction errors increase dramatically. At the intelligent control level, material mechanism models, process control algorithms, and online sensing technologies have not achieved deep collaboration. Furthermore, the development of real-time, high-precision in-situ microstructure and defect detection technologies (such as online LIBS for compositional uniformity analysis and in-situ ultrasound for grain size assessment) is lagging behind, failing to provide high-quality, low-latency feedback signals. The integration and fidelity of digital twin technology are also insufficient. Virtual models designed to map physical production processes often suffer from simplification, computational delays, or poor data interfaces, making it difficult to achieve high-fidelity interaction and dynamic control with physical systems. This results in low efficiency in intelligent control of titanium-based alloy composition optimization and performance prediction due to the ambiguity of the multi-scale mechanism of structure-property mapping and insufficient modeling accuracy. Summary of the Invention

[0004] To address the technical problem of low efficiency in intelligent control of titanium-based alloy composition optimization and performance prediction due to the fuzziness of the multi-scale mechanism of structure-property mapping and insufficient modeling accuracy in existing technologies, this invention provides an intelligent control system for titanium-based alloy composition optimization and performance prediction. The technical solution is as follows:

[0005] On the one hand, an intelligent control system for titanium-based alloy composition optimization and performance prediction is provided. This system includes: a cross-scale structure-property mapping module, an intelligent optimization and prediction module, and a standard adaptation and interaction module.

[0006] The cross-scale structure-property mapping module receives the preprocessed standardized dataset of titanium-based alloy composition, compares and fuses it with the prediction output of the digital twin model at preset corresponding spatiotemporal points, dynamically inverts and calibrates the implicit state variables and key parameters in the digital twin model, thereby continuously correcting the prediction trajectory of the digital twin model to ensure real-time synchronization and state consistency between the virtual and physical spaces. The intelligent optimization and prediction module, based on the calibrated digital twin model after comparison and fusion with the cross-scale structure-property mapping module, collects real-time physical space state data of the entire titanium-based alloy production process, and compares the state data with the virtual prediction data of the digital twin model at preset corresponding spatiotemporal points. The system performs comparative analysis on empty points, extracts deviation information from the two, and generates feedback control signals. Based on the feedback control signals, it dynamically controls the actual production process parameters of titanium-based alloys in real time. The deviation information is then transmitted back to the cross-scale structure-property mapping module to assist in continuously optimizing the implicit state variables and key parameters of the digital twin model. This achieves dynamic matching between the physical and virtual spaces of the entire titanium-based alloy production process and full closed-loop control of composition, process, and performance. The standard adaptation and interaction module is used to achieve highly consistent bidirectional transmission of the pre-processed standardized dataset of titanium-based alloy composition, the predicted output of the digital twin model at preset time and space points, deviation information, and feedback control signals to ensure the uniformity and real-time nature of data interaction.

[0007] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0008] 1. By employing precise modeling through a cross-scale structure-property mapping module and dynamic calibration of the digital twin model, efficient synchronization and state consistency between virtual and physical spaces were achieved, significantly improving the accuracy of composition-microstructure-property correlation prediction for titanium-based alloys. By dimensional alignment, deviation calculation, and fusion of the standardized titanium-based alloy composition dataset and the preset spatiotemporal point prediction output of the digital twin model, reliable corrected benchmark data was generated, providing accurate data support for model calibration. Simultaneously, based on the dynamic mapping relationship between transient phase transition driving force and interface mobility, combined with critical range control of the driving force, problems such as prediction of non-physical ultrafast interface motion and premature simulation termination were effectively avoided, ensuring the realism and continuity of the phase transition process simulation. Furthermore, through adaptive linkage between the virtual time step and the virtual real-time spatiotemporal point matching error threshold, the matching accuracy requirements and simulation calculation resolution were synergistically optimized. This avoided the erroneous rejection of effective data when the simulation step size was large, and prevented the excessive introduction of noisy data when the step size was small. While ensuring the reliability of data assimilation, this significantly improved data utilization efficiency and the overall fidelity of the digital twin model, laying a solid model foundation for titanium-based alloy composition optimization and performance prediction.

[0009] 2. Through the real-time control and closed-loop feedback mechanism of the intelligent optimization and prediction module, precise dynamic control of process parameters throughout the entire titanium-based alloy production process is achieved, effectively ensuring the stability and consistency of product quality. Based on the calibrated digital twin model, physical space state data is collected in real time and compared with virtual prediction data to accurately extract deviation information and generate feedback control signals. By configuring dedicated weight attenuation coefficients for adjustable process parameters, a monotonically decreasing relationship between instantaneous feedback weight and absolute value of control increment is constructed, achieving precise matching between process control commands and deviation information feedback, avoiding control fluctuations caused by sudden changes in process parameters. Simultaneously, through the dynamic mapping of model calibration trigger threshold and feedback signal transmission frequency, the transmission frequency is adaptively adjusted according to the virtual and real deviation levels. When the deviation is small, the system maintains stable operation, and when the deviation reaches or exceeds the trigger threshold, the transmission real-time performance is rapidly improved to support emergency model calibration, ensuring immediate response of process control and model optimization. This effectively solves problems such as link breakage and response lag in traditional control, achieving full closed-loop control of composition-process-performance, and significantly reducing the risk of performance fluctuations during production.

[0010] 3. Through standard adaptation and multi-dimensional transmission optimization and exception handling mechanisms of interactive modules, the uniformity, real-time performance and reliability of data interaction throughout the entire process are ensured, providing strong support for the collaborative operation of various modules of the system and the improvement of management and control efficiency. By optimizing data transmission timing and spatiotemporal synchronization, the system dynamically adjusts the real-time transmission trigger threshold for deviation information based on the predicted synchronization time offset. Precise control of the trigger threshold gain and reduction coefficients suppresses unnecessary transmission when the offset is small and ensures rapid response when the offset is large, guaranteeing accurate adaptation between deviation information transmission and spatiotemporal synchronization. Simultaneously, through data transmission consistency verification and anomaly handling optimization, the system monitors the threshold reset delay time in real time and dynamically adjusts the preset adjustment step size based on the reset duration. When the reset delay exceeds the limit, the target step size is optimized by adjusting the step size reduction coefficient. This effectively ensures the timeliness and effectiveness of real-time deviation information transmission trigger threshold adjustment, avoiding the aggravation of spatiotemporal synchronization deviation and the decline in control effectiveness caused by threshold reset delay. Furthermore, the standardization of data formats, unification of protocols, and multi-stage consistency verification throughout the entire process effectively avoids problems such as format incompatibility, transmission loss, and tampering in data interaction. This ensures highly consistent bidirectional transmission of component-standardized datasets, model prediction outputs, deviation information, and feedback control signals, providing a stable data link for the collaborative operation of the cross-scale structure-property mapping module, intelligent optimization, and prediction module, comprehensively improving the overall robustness and control efficiency of the system. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0012] Figure 1 This is a schematic diagram of the structure of an intelligent control system for optimizing the composition and predicting the performance of titanium-based alloys, provided in an embodiment of this application.

[0013] Figure 2 A flowchart illustrating the dynamic correction of interface migration rate in an intelligent control system for composition optimization and performance prediction of titanium-based alloys, provided in this application embodiment;

[0014] Figure 3 The flowchart illustrates the adaptive control of the feedback signal transmission frequency of an intelligent control system for optimizing the composition and predicting the performance of titanium-based alloys, as provided in this application embodiment. Detailed Implementation

[0015] The technical solution provided in this application will now be described in conjunction with the accompanying drawings.

[0016] To facilitate understanding of the embodiments of this application, the following points will be explained first:

[0017] First, in this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates an "or" relationship between the preceding and following related objects, but it does not exclude the possibility of indicating an "and" relationship; the specific meaning can be understood in context. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c; a and b; a and c; b and c; or a and b and c. Here, a, b, and c can be single or multiple.

[0018] Second, the use of prefixes such as "first" and "second" in this application is solely for the purpose of distinguishing and describing different things belonging to the same category, and does not constrain the order, size, or quantity of things. For example, "first message" and "second message" are simply different messages, and there is no chronological, size, or priority relationship between them.

[0019] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0020] like Figure 1 The diagram shown is a structural schematic of an intelligent control system for optimizing the composition and predicting the performance of titanium-based alloys provided in an embodiment of this application. The intelligent control system for optimizing the composition and predicting the performance of titanium-based alloys provided in an embodiment of this application includes: a cross-scale structure-property mapping module, an intelligent optimization and prediction module, and a standard adaptation and interaction module.

[0021] The cross-scale structure-property mapping module receives the preprocessed standardized dataset of titanium-based alloy composition, compares and fuses the standardized dataset of titanium-based alloy composition with the prediction output of the digital twin model at the preset corresponding spatiotemporal points, dynamically inverts and calibrates the implicit state variables (such as transient temperature field distribution and stress triaxiality) and key parameters (such as oxygen diffusion coefficient and nucleation barrier) in the digital twin model, thereby continuously correcting the prediction trajectory of the digital twin model to ensure real-time synchronization and state consistency between the virtual space and the physical space.

[0022] It is important to understand that the specific steps for comparing and fusing the predicted outputs at the preset corresponding spatiotemporal points are as follows:

[0023] For each component data in the standardized dataset of titanium-based alloy composition, a unique corresponding spatiotemporal point preset by the digital twin model is matched to form standardized titanium-based alloy composition data with spatiotemporal labels. The predicted output data of the digital twin model at each preset spatiotemporal point corresponding to the titanium-based alloy composition is obtained. The predicted output data is the digital twin model based on the initial hidden state variables and key parameters.

[0024] The standardized data of titanium-based alloy composition with spatiotemporal labels is dimensionally aligned with the predicted output data of the digital twin model at the corresponding preset spatiotemporal points to construct a corresponding spatiotemporal point data comparison matrix. The deviation of each component data matching in the spatiotemporal point data comparison matrix is ​​calculated, and the results of the deviation calculation are fused to generate fused corrected benchmark data. This completes the comparison and fusion of the standardized dataset of titanium-based alloy composition and the predicted output of the digital twin model at the preset corresponding spatiotemporal points.

[0025] In this embodiment, firstly, a single component data point in the standardized titanium-based alloy composition dataset is precisely matched and assigned a unique corresponding spatiotemporal point identifier preset by the digital twin model, forming standardized titanium-based alloy composition data with spatiotemporal labels. This step enables precise binding of the component data with the spatiotemporal dimension of the digital twin model, allowing the component data to possess spatiotemporal attributes recognizable by the digital twin model, laying the foundation for spatiotemporal dimension matching for subsequent cross-data type comparison and fusion. Then, the digital twin model is retrieved, and based on the initial implicit state variables and key parameters, it outputs predicted data for the corresponding titanium-based alloy composition at each preset spatiotemporal point. This step relies on the digital twin model's... The simulation prediction capability can quickly obtain theoretical prediction results of alloy composition at various spatiotemporal points, providing a digital comparison reference for actual composition data. Next, dimensional alignment processing is performed on the standardized titanium-based alloy composition data with spatiotemporal tags and the prediction output data of the digital twin model at corresponding preset spatiotemporal points. This ensures that the two types of data are completely consistent in the dimensions of composition indicators and spatiotemporal dimensions. Then, based on the aligned two types of data, a one-to-one spatiotemporal point data comparison matrix is ​​constructed. This dimensional alignment and matrix construction step can eliminate the comparison error caused by the dimensional differences between the two types of data, ensuring that each set of data in the matrix represents both actual composition data and model predictions at the same spatiotemporal and indicator levels. To ensure the accuracy and effectiveness of subsequent deviation calculations, precise deviation calculations are then performed on each component data matching item within the spatiotemporal point data comparison matrix. Based on the deviation calculation results, weighted fusion and other fusion operations are performed on the actual component data and the model prediction data to generate fused corrected baseline data. This deviation calculation and fusion step quantifies the degree of deviation between the actual component data and the model prediction data. Furthermore, the fusion combines the authenticity of the actual data with the theoretical guidance of the model prediction data, effectively correcting the limitations of a single data type. Finally, the entire process of predicting outputs from the standardized titanium-based alloy composition dataset and the digital twin model at preset spatiotemporal points is completed. Through comparison and fusion, the entire process achieves deep integration of actual standardized component data and digital twin model prediction data by connecting each step and empowering it with technology. This not only preserves the measured authenticity of the actual component data but also incorporates the predictive logic of the digital twin model based on core variables and parameters. The resulting corrected benchmark data has higher accuracy and reliability compared to single actual component data or model prediction data. It can serve as the core data benchmark for titanium-based alloy-related R&D and production processes. At the same time, it completes the actual data verification of the digital twin model prediction results, providing a real data analysis basis for subsequent model parameter optimization and implicit state variable adjustment.

[0026] It should be understood that, such as Figure 2The diagram shows a flowchart of dynamic correction of interface migration rate in an intelligent control system for titanium-based alloy composition optimization and performance prediction provided in this application embodiment. The specific process is as follows: starting from "start correcting interface migration rate", the transient phase transition driving force is first acquired in real time, and then the range of the driving force is determined. If it is higher than the upper limit, the mapping relationship is input and the saturated migration rate is output to avoid non-physical ultrafast prediction; if it is lower than the lower limit, the mapping relationship is input and the low migration rate is output to avoid simulation termination; if it is within the critical range, the current interface migration rate is maintained to maintain stability; finally, the interface migration rate in the model is updated uniformly to drive the tissue evolution model to complete the correction of the predicted trajectory.

[0027] It should be further explained that the specific steps for correcting the predicted trajectory of the digital twin model are as follows:

[0028] S1, real-time acquisition of transient phase transformation driving force at each preset time and space point during the phase transformation of titanium-based alloy, retrieve the initial value of interface mobility at the corresponding preset time and space point, and construct the mapping relationship between transient phase transformation driving force and interface mobility. Specifically, the interface mobility increases linearly with the increase of transient phase transformation driving force, and decreases with the decrease of transient phase transformation driving force, and the decay rate matches the critical state of phase transformation cessation.

[0029] S2, if the transient phase transition driving force is greater than the critical upper limit of the driving force, the current driving force exceeding the critical upper limit of the driving force is input into the mapping relationship between the transient phase transition driving force and the interface mobility, and the corresponding interface mobility is output, thus avoiding the non-physical ultrafast interface motion prediction caused by unrestricted linear extrapolation.

[0030] S3, if the transient phase transition driving force is within the driving force critical range, then maintain the current interface mobility and adjust. The driving force critical range represents the closed interval formed by the driving force critical lower limit and the driving force critical upper limit.

[0031] S4. If the transient phase transition driving force is less than the critical lower limit of the driving force, the current driving force below the critical lower limit of the driving force is input into the mapping relationship between the transient phase transition driving force and the interface mobility, and the corresponding interface mobility is output, thus avoiding premature termination of the simulation due to the mobility being incorrectly set to zero.

[0032] In this embodiment, transient phase transformation driving force data at each preset spatiotemporal point during the phase transformation of the titanium-based alloy are first collected in real time. This ensures that the acquired driving force data accurately reflects the dynamic changes in the phase transformation process at different spatiotemporal locations, providing a real and real-time input basis for the subsequent dynamic control of the interface mobility. Simultaneously, the initial value of the interface mobility corresponding to each preset spatiotemporal point is retrieved. This initial value is based on preset fundamental parameters of the phase transformation characteristics of the titanium-based alloy, ensuring the rationality of the initial benchmark for control. Based on this, a specific mapping relationship between the transient phase transformation driving force and the interface mobility is constructed, specifically configured such that the interface mobility varies with the transient phase transformation driving force. The phase transition driving force increases linearly with increasing dynamic phase transition driving force, and decreases with decreasing transient phase transition driving force. Moreover, the decay rate strictly matches the phase transition stopping critical state. The construction of this mapping relationship breaks the limitation of the traditional fixed interface mobility and realizes the dynamic linkage between interface mobility and phase transition driving force. It can adapt to the physical law of rapid migration of the phase transition interface when the driving force increases through linear increase characteristics, and avoid the phase transition simulation distortion caused by the mobility decaying too fast or too slow when the driving force decreases by matching the decay rate of the phase transition stopping critical state. This lays the core control foundation for the accurate simulation of the entire phase transition process. After completing the mapping relationship construction, a threshold judgment is performed on the real-time acquired transient phase transition driving force. If the current transient phase transition driving force is detected to be greater than the preset driving force critical upper limit, the driving force value exceeding the critical upper limit is immediately input into the transient phase transition driving force and interface mobility mapping relationship constructed by S1. The corresponding interface mobility is calculated and output through the mapping relationship. This step effectively avoids the non-physical ultrafast interface motion prediction problem that may be caused by unlimited linear extrapolation in the traditional linear control mode by setting the driving force critical upper limit and triggering dedicated control logic. It also prevents the interface mobility from the extreme case that does not conform to the physical law of phase transition of titanium-based alloys due to the unlimited increase of driving force, and ensures the physical rationality and prediction accuracy of phase transition simulation in high driving force scenarios. If the transient phase transition driving force is detected in real time to be within the preset critical range of driving force (this critical range is a closed interval formed by the lower critical limit and the upper critical limit of driving force), the current interface mobility value is maintained unchanged and dynamic monitoring and adjustment are performed. This control logic is designed for the relatively stable intermediate stage of the phase transition process. By maintaining the stability of the interface mobility, it avoids simulation fluctuations caused by unnecessary frequent adjustments. At the same time, the dynamic monitoring setting can ensure a rapid response when the driving force is about to exceed the critical range, taking into account both the stability of the simulation and the timeliness of the control, and ensuring the smoothness and efficiency of the simulation process in the intermediate stage of the phase transition.If the transient phase transition driving force is detected to be less than the preset critical lower limit of the driving force in real time, the driving force value below the critical lower limit is input into the transient phase transition driving force and interface mobility mapping relationship constructed by S1. The corresponding interface mobility is calculated and output through the mapping relationship. This step effectively solves the problem that the mobility may be incorrectly set to zero due to the driving force being lower than the lower limit in traditional control, and avoids the situation that the phase transition simulation will terminate prematurely due to the abnormal zeroing of the mobility. It ensures the integrity of the entire process simulation of the phase transition process of titanium-based alloys from start-up, stable progress to near the critical state of stopping. It can accurately capture the slow migration characteristics of the phase transition interface under low driving force scenarios, and provides a guarantee for the accurate simulation of the entire phase transition process.

[0033] It should be further added that revising the predicted trajectory of the digital twin model also includes:

[0034] Establish a mapping relationship between the virtual time step and the matching error threshold of the virtual real-time space point. Specifically, the larger the virtual simulation time step, the larger the corresponding matching error threshold of the virtual real-time space point; conversely, the shorter the virtual simulation time step, the smaller the corresponding matching error threshold of the virtual real-time space point.

[0035] If the time step of the virtual simulation is within the time step reference interval, the matching error threshold of the current virtual real-time point is maintained. The time step reference interval represents the closed interval formed by the lower limit of the time step reference and the upper limit of the time step reference.

[0036] If the time step of the virtual simulation is less than the lower limit of the time step reference, the current time step of the virtual simulation is input into the mapping relationship between the virtual time step and the matching error threshold of the virtual real space-time point, and the corresponding matching error threshold of the virtual real space-time point is output.

[0037] If the time step of the virtual simulation is greater than the lower limit of the time step reference, then the current time step of the virtual simulation is input into the mapping relationship between the virtual time step and the matching error threshold of the virtual real-time point, and the corresponding matching error threshold of the virtual real-time point is output.

[0038] In this embodiment, an adaptive coordination between matching accuracy requirements and simulation calculation resolution is achieved. This avoids the erroneous rejection of a large amount of valid data due to overly strict fixed thresholds when the simulation step size is large, or the introduction of excessive noise due to overly lenient thresholds when the step size is small. Thus, while ensuring the reliability of data assimilation, data utilization efficiency and overall system robustness are optimized. By intelligently linking matching fault tolerance capability with simulation calculation resolution, a large amount of still valuable observation information is avoided from being discarded due to overly stringent fault tolerance standards when the simulation is coarse. When the simulation is fine, higher standards are raised to select higher-quality data to drive accurate model updates, thereby optimizing data assimilation efficiency and the overall fidelity of the digital twin model under all operating conditions.

[0039] The intelligent optimization and prediction module is used to compare and fuse the digital twin model with the cross-scale structure-property mapping module and then calibrate it. It collects physical space state data of the entire titanium-based alloy production process in real time, compares and analyzes the state data with the virtual prediction data of the digital twin model at preset corresponding spatiotemporal points, extracts the deviation information between the two and generates feedback control signals. Based on the feedback control signals, it performs real-time dynamic control of the actual production process parameters of titanium-based alloys, and transmits the deviation information back to the cross-scale structure-property mapping module to help it continuously optimize the implicit state variables and key parameters of the digital twin model. This realizes the dynamic matching of physical space and virtual space of the entire titanium-based alloy production process and the closed-loop control of composition-process-performance.

[0040] It should be noted that, as Figure 3 The diagram shown is a flowchart of the adaptive control of the feedback signal transmission frequency of an intelligent control system for titanium-based alloy composition optimization and performance prediction provided in this application embodiment. The specific process is as follows: starting from "start frequency control", the current virtual-real deviation value is collected and calculated in real time, and then the deviation value is compared with the model calibration trigger threshold. If the deviation does not reach the threshold, the basic value of the feedback signal transmission frequency is maintained; if the deviation reaches the threshold, the transmission frequency is increased by a preset increase; if the deviation exceeds the threshold preset ratio, the frequency is increased to the upper limit by the maximum increase. Subsequently, the current frequency is uniformly used for data transmission, ultimately supporting model calibration.

[0041] It should be understood that the specific steps for real-time dynamic control of actual production process parameters of titanium-based alloys based on feedback control signals are as follows:

[0042] For each adjustable process parameter, a weight decay coefficient is predefined. Within a preset control period, when the change in the real-time control command value applied to a certain process parameter relative to the previous period, i.e., the absolute value of the control increment, increases, the instantaneous feedback weight of the deviation information collected by the monitoring point directly associated with that process parameter should be decayed accordingly. There is a monotonically decreasing relationship between the instantaneous feedback weight and the absolute value of the control increment.

[0043] After each generation and issuance of process parameter control instructions, the adjustment increments are immediately determined based on the parameters contained in the instructions. For each input deviation, an instantaneous feedback weight is obtained, specifically:

[0044] Extract the associated monitoring point corresponding to each input deviation information, match the target controllable process parameter of the associated monitoring point, and retrieve the predefined weight attenuation coefficient of the target controllable process parameter.

[0045] Obtain the control increment of the target adjustable process parameter in the process parameter control command, and get the absolute value of the control increment. If the absolute value of the control increment is greater than or equal to the increment reference value, the corresponding instantaneous feedback weight is reduced.

[0046] If the absolute value of the control increment is less than the increment reference value, the corresponding instantaneous feedback weight is increased to serve as the weight for generating process parameter control commands in the next preset control cycle.

[0047] Real-time dynamic control of actual production process parameters for titanium-based alloys based on feedback control signals also includes:

[0048] A set of model calibration trigger thresholds is preset, and a set of feedback signal transmission frequencies is defined accordingly. A mapping relationship between the model calibration trigger thresholds and the feedback signal transmission frequencies is established. The model calibration trigger thresholds are compared with the multi-level calibration trigger threshold range in real time, and the preset feedback signal transmission frequency corresponding to the range is dynamically selected and switched.

[0049] The preset model calibration trigger threshold and feedback signal transmission frequency are the base values ​​and adjustment ranges. The model calibration trigger threshold is the critical deviation value for determining whether to start the cross-scale structure-property mapping module to calibrate the parameters of the digital twin model. The feedback signal transmission frequency is the number of times the deviation information is transmitted from the intelligent optimization and prediction module to the cross-scale structure-property mapping module per unit time. The adjustment range includes the frequency increase range and the frequency decrease range.

[0050] The current virtual-real deviation value of the core parameters of the digital twin model is collected and calculated in real time. The current virtual-real deviation value is the comparison deviation between the physical space state data and the virtual space prediction data at each preset time and space point in the entire titanium-based alloy production process. The current virtual-real deviation value is compared with the corresponding model calibration trigger threshold to obtain the deviation comparison result. The deviation comparison result is divided into deviation not reaching the trigger threshold, deviation reaching the trigger threshold, and deviation exceeding the trigger threshold by a preset proportion.

[0051] In this embodiment, during the real-time dynamic control of actual production process parameters for titanium-based alloys, a specific weight attenuation coefficient is first predefined for each controllable process parameter (such as melting temperature, forging pressure, cooling rate, etc.). This coefficient can be configured based on the priority and control sensitivity of the parameter's impact on the performance of titanium-based alloy products (such as strength, toughness, and corrosion resistance), providing a core benchmark for subsequent dynamic adjustment of feedback weights. The technical effect is to proactively avoid misjudgment of feedback signals caused by excessive control of a single parameter, laying the foundation for accurate generation of control commands. Within a preset control period (set according to the characteristics of the titanium-based alloy production process, such as…),… (The melting process is measured in minutes, and the forging process in seconds). When the absolute value of the real-time control command value applied to a certain process parameter increases relative to the change in the previous cycle (i.e., the control increment), the instantaneous feedback weight of the deviation information (the difference between the actual monitored value and the preset standard value) collected by the monitoring points directly related to that process parameter (such as temperature sensors, pressure sensors, displacement sensors, etc.) will decrease accordingly. This monotonically decreasing relationship between the instantaneous feedback weight and the absolute value of the control increment is designed to effectively suppress the interference of feedback signals when parameters fluctuate greatly, avoid erroneous responses of the control system due to sudden parameter changes, and ensure the stability of process control. And continuity; after each generation and issuance of process parameter control instructions, the system will immediately calculate the instantaneous feedback weight for each input deviation information according to the adjustment increment of each parameter contained in the control instruction. The specific implementation process is as follows: First, extract the associated monitoring point corresponding to each input deviation information. Through the preset monitoring point-process parameter mapping relationship, accurately match the target controllable process parameter corresponding to the associated monitoring point, and retrieve the predefined weight attenuation coefficient of the target controllable process parameter from the system parameter library. The technical effect of this step is to achieve accurate association between deviation information and control parameters, ensure the pertinence of weight adjustment, and avoid inconsistencies. The first step is to prevent cross-interference of parameter feedback signals. The second step is to obtain the control increment of the target adjustable process parameter in this control command. The absolute value of the control increment is calculated by absolute value. Then, the absolute value is compared with the preset incremental benchmark value (set according to the stable range of titanium-based alloy process and parameter control experience). If the absolute value of the control increment is greater than or equal to the incremental benchmark value, it means that the parameter control amplitude is large. At this time, the system will reduce the corresponding instantaneous feedback weight according to the weight attenuation coefficient. The technical effect is to reduce the excessive influence of feedback signals after large-scale control, prevent the system from overshooting due to feedback fluctuations, and ensure that the process parameters smoothly transition to the stable range.If the absolute value of the adjustment increment is less than the increment benchmark value, it indicates that the adjustment range of the parameter is within a reasonable range. In this case, the system will appropriately increase the corresponding instantaneous feedback weight. The technical effect is to strengthen the guiding role of the feedback signal during small-amplitude adjustments, improve the response accuracy of the control command, and ensure that the process parameters remain stable within the optimal range. The instantaneous feedback weight calculated above will serve as the core weight basis for generating process parameter control commands in the next preset control cycle, achieving closed-loop optimization of the control process.

[0052] It should be further explained that the specific steps for dynamically selecting and switching to the preset feedback signal transmission frequency corresponding to this interval are as follows:

[0053] Based on the deviation comparison results, a dynamic adjustment rule for the feedback signal transmission frequency is constructed. The specific dynamic adjustment rule for the feedback signal transmission frequency is as follows: if the current virtual-real deviation value does not reach the corresponding model calibration trigger threshold, the basic value of the feedback signal transmission frequency is maintained.

[0054] If the current virtual-to-real deviation value reaches the corresponding model calibration trigger threshold, the feedback signal transmission frequency will be adjusted to the corresponding value within the frequency increase range according to the preset increase range, so as to improve the real-time performance of deviation information transmission.

[0055] If the current virtual-to-real deviation value exceeds the preset ratio of the corresponding model calibration trigger threshold, the feedback signal transmission frequency will be adjusted to the upper limit of the frequency increase range according to the maximum allowable increase, so as to ensure that the deviation information is transmitted quickly to support the emergency calibration of the model.

[0056] In this embodiment, in the calibration and feedback control process of the digital twin model for titanium-based alloy production, the system dynamically selects and switches to the preset feedback signal transmission frequency corresponding to the current deviation range. This is based on the deviation comparison results and is executed step by step according to preset rules. The specific steps and technical effects are as follows: First, preparation and rule implementation: Based on the deviation comparison results obtained in the early stage (deviation not reaching the trigger threshold, deviation reaching the trigger threshold, deviation exceeding the trigger threshold by a preset proportion), the system constructs a complete dynamic adjustment rule system for the feedback signal transmission frequency. This rule system is precisely correlated with the previously preset model calibration trigger threshold, the basic value of the feedback signal transmission frequency, the frequency increase range (including the upper limit of the range), the preset increase range, and the maximum allowable increase range. This ensures that each deviation scenario has a clear basis for frequency adjustment, achieving standardization and normalization of frequency adjustment, avoiding adjustment delays or misoperations caused by manual intervention, and ensuring the automated operation of the process. Second, deviation scenario judgment and basic frequency maintenance: The system retrieves the current virtual-real deviation value calculated in the early stage (the comparison deviation between the physical space state data and the virtual space prediction data at each preset time and space point in the entire titanium-based alloy production process) in real time and compares it again with the corresponding model calibration trigger threshold. Yes, if the current virtual-to-real deviation value is determined not to have reached the corresponding model calibration trigger threshold, it indicates that the current prediction accuracy of the digital twin model is good and its fit with the actual production process is high. Therefore, there is no need to initiate the model calibration process. At this time, the system will maintain the basic value of the feedback signal transmission frequency to avoid unnecessary frequency increases during the model stabilization phase, reduce data transmission redundancy, lower system operating energy consumption, and ensure the stable transmission of basic deviation information, providing basic data support for subsequent process control. The third step involves deviation compliance scenarios and frequency tiered increases. If, through comparison, the current virtual-to-real deviation value is determined to have reached the corresponding model calibration trigger threshold, it indicates that the digital twin model... If the prediction accuracy has deviated, the regular model calibration process needs to be initiated. At this time, the system will adjust the transmission frequency of the feedback signal from the base value to the corresponding level value within the frequency increase range according to the preset increase range (set according to the deviation compliance level and production process response requirements, such as increasing the transmission frequency by 20% for each level). By increasing the frequency, the real-time performance of deviation information transmission is significantly enhanced, ensuring that the deviation data collected by the intelligent optimization and prediction module can be quickly and accurately transmitted to the cross-scale structure-property mapping module. This provides timely data support for the accurate calibration of model parameters and avoids untimely calibration due to data transmission lag, which in turn affects the accuracy of process control.The fourth step involves switching between the deviation exceeding the limit scenario and the frequency upper limit. If the comparison result shows that the current virtual-real deviation value exceeds the preset proportion of the corresponding model calibration trigger threshold (e.g., exceeding 30% or more), it indicates that the digital twin model has a large deviation, which has affected the accuracy of the generation of process control instructions. An emergency calibration process needs to be initiated immediately. At this time, the system will adjust the feedback signal transmission frequency directly to the upper limit of the frequency increase range according to the maximum allowable increase (the system's preset frequency increase limit value, ensuring maximum transmission efficiency while avoiding system overload). This achieves extremely fast transmission of deviation information, minimizes data transmission delay, ensures that the cross-scale structure-property mapping module can obtain complete deviation data in real time, quickly complete the emergency calibration of model parameters, curb further expansion of deviation, and avoid problems such as fluctuations in titanium-based alloy production processes and product quality defects caused by model inaccuracies. The system monitors the stability of the transmission link and data transmission efficiency in real time to ensure that the adjusted frequency is suitable for the current production scenario and model calibration requirements. Simultaneously, it feeds back the frequency adjustment results to the intelligent optimization and prediction module, providing a reference for process parameter control and deviation comparison in the next cycle. This forms a closed-loop optimization of "deviation comparison - frequency adjustment - model calibration - process control," further improving the stability of the titanium-based alloy production process and the consistency of product quality.

[0057] The standard adaptation and interaction module is used to achieve highly consistent bidirectional transmission of the preprocessed titanium-based alloy composition standardized dataset, the prediction output of the digital twin model at preset spatiotemporal points, deviation information and feedback control signals, so as to ensure the uniformity and real-time performance of data interaction.

[0058] It should be noted that the standard adaptation and interaction module includes optimization of data transmission timing and spatiotemporal synchronization, and optimization of data transmission consistency verification and exception handling.

[0059] It should be further explained that the specific steps for optimizing data transmission timing and spatiotemporal synchronization are as follows:

[0060] Based on the established mapping relationship between the predicted synchronization time offset and the real-time transmission trigger threshold coefficient of deviation information, the real-time transmission trigger threshold of deviation information in the current control cycle is dynamically adjusted according to the predicted synchronization time offset.

[0061] If the predicted synchronization offset is within the offset critical interval, then the current deviation information real-time transmission trigger threshold is maintained. The offset critical interval represents the closed interval formed by the offset critical lower limit and the offset critical upper limit.

[0062] If the predicted synchronization time offset is less than the critical lower limit of the offset, the current predicted synchronization time offset is input into the mapping relationship between the predicted synchronization time offset and the real-time transmission trigger threshold coefficient of the deviation information, and the trigger threshold gain coefficient is output. The trigger threshold gain coefficient is combined with the current real-time transmission trigger threshold of the deviation information to obtain the target real-time transmission trigger threshold of the deviation information in order to suppress unnecessary transmission.

[0063] Dynamically adjusting the real-time transmission trigger threshold of deviation information within the current control cycle also includes:

[0064] If the predicted synchronization time offset is greater than the critical upper limit of the offset, the current predicted synchronization time offset is input into the mapping relationship between the predicted synchronization time offset and the real-time transmission trigger threshold coefficient of the deviation information, and the trigger threshold reduction coefficient is output. The trigger threshold reduction coefficient is combined with the current real-time transmission trigger threshold of the deviation information to obtain the target real-time transmission trigger threshold of the deviation information.

[0065] Within the current control cycle, the instantaneous actual deviation from the latest synchronization state is obtained, and the instantaneous actual deviation is compared with the adjusted deviation information real-time transmission trigger threshold. If the instantaneous actual deviation is greater than or equal to the adjusted deviation information real-time transmission trigger threshold, the deviation information real-time transmission is immediately triggered; otherwise, it remains silent and no data transmission is performed.

[0066] In this embodiment, firstly, a precise mapping relationship needs to be established between the predicted synchronization time offset and the real-time transmission trigger threshold coefficient of deviation information. This mapping relationship is the core basis for realizing the dynamic adjustment of the trigger threshold. Its establishment can be based on historical operating data, control scenario requirements, and transmission efficiency optimization goals, and can be determined through algorithm fitting or simulation verification. This ensures the scientific and reasonable nature of subsequent threshold adjustment based on the offset, laying the foundation for accurate allocation of transmission resources and stable improvement of control performance. After establishing this mapping relationship, in each control cycle, the predicted synchronization time offset within the current control cycle is first obtained. Based on the predicted offset, the dynamic adjustment process of the real-time transmission trigger threshold of deviation information within the current control cycle is initiated. This breaks the limitations of the traditional fixed threshold mode, realizes real-time linkage between threshold adjustment and system synchronization status, and avoids the transmission redundancy or untimely triggering problems that occur when the system status changes dynamically with a fixed threshold. During the dynamic adjustment process, it is first determined whether the predicted synchronization time offset is within the preset critical offset range (this range is a closed interval formed by the lower critical offset limit and the upper critical offset limit; the setting of the upper and lower critical limits needs to be determined in conjunction with the system synchronization accuracy requirements, transmission delay tolerance, and other core indicators). If the predicted synchronization time offset is within this critical range, it indicates that the current system synchronization state is stable and the probability of large fluctuations in deviation information is low. At this time, the current real-time transmission trigger threshold of deviation information is kept unchanged. This operation can reduce unnecessary threshold adjustment operations and reduce system computational overhead while ensuring system control accuracy. At the same time, it avoids transmission trigger logic disorder caused by frequent threshold adjustments, ensuring the stability and continuity of system operation. If the predicted synchronization time offset is determined to be less than the critical lower limit of the offset, it indicates that the current system synchronization state is better than expected, and the deviation information is likely within a small range. At this time, the current predicted synchronization time offset is input into the pre-established mapping relationship between the predicted synchronization time offset and the real-time transmission trigger threshold coefficient of the deviation information. The corresponding trigger threshold gain coefficient is output through this mapping relationship. Then, the trigger threshold gain coefficient is combined with the current real-time transmission trigger threshold of the deviation information (such as multiplication or other preset calculation methods) to obtain the target real-time transmission trigger threshold of the deviation information. By increasing the trigger threshold, small deviation information transmission requests can be effectively filtered out, significantly reducing unnecessary data transmission volume, reducing network bandwidth occupation and equipment energy consumption, while not affecting the normal control effect of the system, because the system synchronization state is good at this time, and small deviations will not affect the control accuracy.If the predicted synchronization time offset is determined to be greater than the critical upper limit of the offset, it indicates that the current system synchronization state deviation is large, which poses a risk to the control accuracy. At this time, the current predicted synchronization time offset is also input into the above mapping relationship, and the corresponding trigger threshold reduction coefficient is output. Then, the trigger threshold reduction coefficient is combined with the current deviation information real-time transmission trigger threshold to obtain the target deviation information real-time transmission trigger threshold. By reducing the trigger threshold, the sensitivity of deviation information transmission can be improved, ensuring that once deviation information exceeds the tolerance range, it can be detected and triggered for transmission in a timely manner. This ensures that the control center can quickly obtain the system synchronization deviation and take timely adjustment measures to avoid further expansion of the deviation and decline in system control performance, effectively improving the system's anti-interference ability and control stability. After dynamically adjusting the real-time transmission trigger threshold for deviation information within the current control cycle, the system continuously acquires the instantaneous actual deviation between the system and the latest synchronization state within that control cycle. This instantaneous actual deviation is then compared in real-time with the adjusted real-time transmission trigger threshold. If the instantaneous actual deviation is greater than or equal to the adjusted trigger threshold, it indicates that the deviation information has reached a level requiring feedback, and the real-time transmission of the deviation information is immediately triggered. This ensures that the control center can promptly grasp any abnormalities in the system's synchronization state, quickly respond to and execute control strategies, and guarantee that the system's control accuracy always meets preset requirements. If the instantaneous actual deviation is less than the adjusted trigger threshold, it indicates that the current deviation is within a tolerable range. In this case, the system remains silent and does not transmit data. This design further optimizes transmission resource allocation, avoids invalid data transmission consuming network resources and equipment processing capacity, and balances the dual requirements of system control accuracy and transmission efficiency. Overall, it achieves on-demand triggering and precise control of deviation information transmission, significantly improving the system's operational stability, control reliability, and resource utilization efficiency.

[0067] It should be further explained that the specific steps for data transmission consistency verification and anomaly handling optimization are as follows:

[0068] Real-time monitoring and calculation of threshold reset delay time: After the deviation information is transmitted in real time and the threshold adjustment is completed, the reset timer is started, the state change of the trigger threshold is monitored in real time, and the start and end times of the trigger threshold resetting from the adjusted value to its preset base value are recorded to obtain the threshold reset delay time.

[0069] If the threshold reset delay time is less than or equal to the reset reference time, the current preset adjustment step size will be maintained to ensure the stability of subsequent threshold adjustments.

[0070] If the threshold reset delay time is greater than the reset reference time, it indicates that the threshold reset delay has affected the synchronization time offset correction and virtual-real synchronization control effect. The current threshold reset delay time is input into the preset threshold reset delay time-preset adjustment step size mapping relationship, and the adjustment step size reduction coefficient is output. The current preset adjustment step size and the adjustment step size reduction coefficient are combined to obtain the target preset adjustment step size, so as to ensure the timeliness and effectiveness of subsequent deviation information real-time transmission triggering threshold adjustment.

[0071] In this embodiment, on the one hand, by accurately capturing the time node when the trigger threshold is reset to a preset base value after adjustment, the delay time calculation is completed. This allows for real-time perception of the dynamic characteristics of threshold reset and accurate identification of whether the reset delay negatively impacts the correction of the synchronization time offset and the effectiveness of virtual-real synchronization control. This provides a precise and real-time basis for step size adjustment, ensuring the scientific and targeted nature of the control decision from the data source. On the other hand, for scenarios where the delay time meets the benchmark requirements, the original adjustment step size is maintained, effectively avoiding system fluctuations caused by meaningless step size adjustments. This stabilizes the operation of subsequent trigger threshold adjustments and ensures the stability and consistency of threshold adjustment during virtual-real synchronization control. For scenarios where the delay time exceeds the benchmark requirements, the actual delay time is input into a preset mapping relationship to output a reduction coefficient, and the original... By combining and optimizing the adjustable step size to obtain the target step size, this approach can specifically address the issues of untimely correction of synchronization offset caused by reset delay and the attenuation of virtual-real synchronization control effect. Through quantized reduction of the step size, subsequent deviation information is transmitted in real time to trigger threshold adjustment step size adaptation, significantly improving the timeliness and effectiveness of threshold adjustment. This fundamentally ensures the accuracy of synchronization offset correction and the reliability of control effect in the virtual-real synchronization control system. Overall, this solution constructs a closed-loop dynamic optimization mechanism for threshold adjustment step size, achieving dual guarantees of stability and timeliness during threshold adjustment. It effectively solves the problem of control effect attenuation caused by reset delay, significantly improving the overall accuracy, response efficiency, and operational reliability of virtual-real synchronization control, and ensuring the stable and efficient operation of the virtual-real synchronization system under complex dynamic conditions.

[0072] The various features and processes described above can be used independently of each other or can be combined in various ways. All possible combinations and sub-combinations are intended to fall within the scope of this disclosure. Furthermore, certain method or process blocks may be omitted in some embodiments. The methods and processes described herein are not limited to any particular order, and the blocks or states associated with them may be performed in other suitable orders. For example, the described blocks or states may be performed in an order different from the order specifically disclosed, or multiple blocks or states may be combined in a single block or state. Example blocks or states may be performed serially, in parallel, or in some other manner. Blocks or states may be added to or removed from the disclosed example embodiments. The exemplary systems and components described herein may be configured differently from those described. For example, elements may be added to, removed from, or rearranged compared to the disclosed example embodiments.

[0073] The various operations of the example methods described herein can be performed at least in part by an algorithm. This algorithm can be contained in program code or instructions stored in memory (e.g., the aforementioned non-transitory computer-readable storage medium). Such an algorithm may include a machine learning algorithm. In some embodiments, the machine learning algorithm may not be explicitly programmed into the computer to perform the function, but can learn from training data to create a predictive model that performs the function.

[0074] The various operations of the example methods described herein can be performed, at least in part, by one or more processors that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors can constitute the engine of a processor implementation that operates to perform one or more of the operations or functions described herein.

[0075] Similarly, the methods described herein can be implemented at least in part by a processor, where one or more specific processors are examples of hardware. For example, at least some operations of a method can be performed by one or more processors or an engine implemented by a processor. Furthermore, one or more processors can also be operated to support the performance of related operations in a “cloud computing” environment or as “Software as a Service” (SaaS). For example, at least some operations can be performed by a set of computers (as an example of a machine including processors), where these operations are accessible via a network (e.g., the Internet) and via one or more suitable interfaces (e.g., application programming interfaces (APIs)).

[0076] The performance of certain operations can be distributed across processors, residing not only within a single machine but also deployed across multiple machines. In some example embodiments, the processor or processor-implemented engine may reside in a single geographic location (e.g., within a home environment, office environment, or server cluster). In other example embodiments, the processor or processor-implemented engine may be distributed across multiple geographic locations.

[0077] In this specification, multiple instances may implement components, operations, or structures described as single instances. Although individual operations of one or more methods are shown and described as separate operations, one or more of the separate operations may be performed simultaneously and do not need to be performed in the order shown. Structures and functions presented as separate components in the example configuration may be implemented as composite structures or components. Similarly, structures and functions presented as single components may be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of this document.

[0078] While an overview of the subject matter has been described with reference to specific example embodiments, various modifications and changes can be made to these embodiments without departing from the broader scope of embodiments of this disclosure. Such embodiments of the subject matter are referred to herein, individually or collectively, by the term "invention," and are used for convenience only and are not intended to limit the scope of this application to any single disclosure or concept, should more than one disclosure or concept be disclosed in fact.

[0079] The embodiments described herein have been described in sufficient detail to enable those skilled in the art to practice the disclosed teachings. Other embodiments may be used and derived therefrom, such that structural and logical substitutions and changes may be made without departing from the scope of this disclosure. Therefore, the detailed description should not be construed as limiting, and the scope of the various embodiments is defined only by the appended claims and the full scope of their equivalents.

Claims

1. A smart control system for optimizing the composition and predicting the performance of titanium-based alloys, characterized in that, It includes a cross-scale structure-property mapping module, an intelligent optimization and prediction module, and a standard adaptation and interaction module: The cross-scale structure-property mapping module is used to receive the preprocessed titanium-based alloy composition standardized dataset, compare and fuse the titanium-based alloy composition standardized dataset with the prediction output of the digital twin model at the preset corresponding spatiotemporal points, dynamically invert and calibrate the hidden state variables and key parameters in the digital twin model, thereby continuously correcting the prediction trajectory of the digital twin model to ensure real-time synchronization and state consistency between the virtual space and the physical space. The intelligent optimization and prediction module is used to collect physical space state data of the entire titanium-based alloy production process in real time based on the digital twin model calibrated after comparison and fusion with the cross-scale structure-property mapping module. The state data is compared and analyzed with the virtual prediction data of the digital twin model at preset corresponding spatiotemporal points. The deviation information between the two is extracted and feedback control signals are generated. Based on the feedback control signals, the actual production process parameters of titanium-based alloy are dynamically controlled in real time. The deviation information is transmitted back to the cross-scale structure-property mapping module to assist it in continuously optimizing the implicit state variables and key parameters of the digital twin model, so as to realize the dynamic matching of physical space and virtual space of the entire titanium-based alloy production process and the closed-loop control of composition-process-performance. The standard adaptation and interaction module is used to realize highly consistent bidirectional transmission of the preprocessed titanium-based alloy composition standardized dataset, the prediction output of the digital twin model at preset time and space points, deviation information and feedback control signals, so as to ensure the uniformity and real-time performance of data interaction.

2. The intelligent control system for titanium-based alloy composition optimization and performance prediction as described in claim 1, characterized in that, The specific steps for comparing and fusing the predicted outputs at preset corresponding spatiotemporal points are as follows: For each component data in the standardized dataset of titanium-based alloy composition, a unique corresponding spatiotemporal point preset by the digital twin model is matched to form standardized titanium-based alloy composition data with spatiotemporal labels. The predicted output data of the digital twin model at each preset corresponding spatiotemporal point of titanium-based alloy composition is obtained. The predicted output data is the digital twin model based on the initial hidden state variables and key parameters. The standardized data of titanium-based alloy composition with spatiotemporal labels is dimensionally aligned with the predicted output data of the digital twin model at the corresponding preset spatiotemporal points to construct a corresponding spatiotemporal point data comparison matrix. The deviation of each component data matching in the spatiotemporal point data comparison matrix is ​​calculated, and the results of the deviation calculation are fused to generate fused corrected benchmark data. This completes the comparison and fusion of the standardized dataset of titanium-based alloy composition and the predicted output of the digital twin model at the preset corresponding spatiotemporal points.

3. The intelligent control system for titanium-based alloy composition optimization and performance prediction as described in claim 1, characterized in that, The specific steps for correcting the predicted trajectory using the digital twin model are as follows: S1, real-time acquisition of transient phase transformation driving force at each preset time and space point during the phase transformation of titanium-based alloy, retrieve the initial value of interface mobility at the corresponding preset time and space point, and construct the mapping relationship between transient phase transformation driving force and interface mobility. Specifically, the interface mobility increases linearly with the increase of transient phase transformation driving force, and decreases with the decrease of transient phase transformation driving force, and the decay rate matches the critical state of phase transformation cessation. S2, if the transient phase transition driving force is greater than the critical upper limit of the driving force, the current driving force exceeding the critical upper limit of the driving force is input into the mapping relationship between the transient phase transition driving force and the interface mobility, and the corresponding interface mobility is output, thus avoiding the non-physical ultrafast interface motion prediction caused by unrestricted linear extrapolation. S3, if the transient phase change driving force is within the driving force critical range, then maintain the current interface mobility and adjust it. The driving force critical range refers to the closed range formed by the driving force critical lower limit and the driving force critical upper limit. S4. If the transient phase transition driving force is less than the critical lower limit of the driving force, the current driving force below the critical lower limit of the driving force is input into the mapping relationship between the transient phase transition driving force and the interface mobility, and the corresponding interface mobility is output, thus avoiding premature termination of the simulation due to the mobility being incorrectly set to zero.

4. The intelligent control system for titanium-based alloy composition optimization and performance prediction as described in claim 3, characterized in that, The corrected digital twin model predicts the trajectory, which also includes: Establish a mapping relationship between the virtual time step and the matching error threshold of the virtual real-time space point. Specifically, the larger the virtual simulation time step, the larger the corresponding matching error threshold of the virtual real-time space point; conversely, the shorter the virtual simulation time step, the smaller the corresponding matching error threshold of the virtual real-time space point. If the time step of the virtual simulation is within the time step reference interval, the current matching error threshold of the virtual real-time point is maintained. The time step reference interval represents the closed interval formed by the lower limit of the time step reference and the upper limit of the time step reference. If the time step of the virtual simulation is less than the lower limit of the time step reference, the current time step of the virtual simulation is input into the mapping relationship between the virtual time step and the matching error threshold of the virtual real space-time point, and the corresponding matching error threshold of the virtual real space-time point is output. If the time step of the virtual simulation is greater than the lower limit of the time step reference, then the current time step of the virtual simulation is input into the mapping relationship between the virtual time step and the matching error threshold of the virtual real-time point, and the corresponding matching error threshold of the virtual real-time point is output.

5. The intelligent control system for titanium-based alloy composition optimization and performance prediction as described in claim 1, characterized in that, The specific steps for real-time dynamic control of actual production process parameters of titanium-based alloys based on feedback control signals are as follows: For each adjustable process parameter, a weight decay coefficient is predefined. Within a preset control period, when the change in the real-time control command value applied to a certain process parameter relative to the previous period, i.e., the absolute value of the control increment, increases, the instantaneous feedback weight of the deviation information collected by the monitoring point directly associated with that process parameter should be decayed accordingly. The instantaneous feedback weight has a monotonically decreasing relationship with the absolute value of the control increment. After each generation and issuance of process parameter control instructions, the adjustment increments are immediately determined based on the parameters contained in the instructions. For each input deviation, an instantaneous feedback weight is obtained, specifically: Extract the associated monitoring point corresponding to each input deviation information, match the target adjustable process parameter of the associated monitoring point, and retrieve the predefined weight attenuation coefficient of the target adjustable process parameter; The adjustment increment of the target adjustable process parameter in the process parameter adjustment command is obtained, and the absolute value of the adjustment increment is obtained. If the absolute value of the adjustment increment is greater than or equal to the increment reference value, the corresponding instantaneous feedback weight is reduced. If the absolute value of the control increment is less than the increment reference value, the corresponding instantaneous feedback weight is increased to serve as the weight for generating process parameter control commands in the next preset control cycle.

6. The intelligent control system for titanium-based alloy composition optimization and performance prediction as described in claim 5, characterized in that, The real-time dynamic control of actual production process parameters of titanium-based alloys based on feedback control signals also includes: A set of model calibration trigger thresholds is preset, and a set of feedback signal transmission frequencies is defined accordingly. A mapping relationship between the model calibration trigger thresholds and the feedback signal transmission frequencies is established. The model calibration trigger thresholds are compared with the multi-level calibration trigger threshold range in real time, and the preset feedback signal transmission frequency corresponding to the range is dynamically selected and switched. The preset model calibration trigger threshold and feedback signal transmission frequency are defined as the base values ​​and adjustment ranges. The model calibration trigger threshold is the critical deviation value for determining whether to activate the cross-scale structure-property mapping module to calibrate the parameters of the digital twin model. The feedback signal transmission frequency is the number of times the deviation information is transmitted from the intelligent optimization and prediction module to the cross-scale structure-property mapping module per unit time. The adjustment range includes a frequency increase range and a frequency decrease range. The current virtual-real deviation value of the core parameters of the digital twin model is collected and calculated in real time. The current virtual-real deviation value is the comparison deviation between the physical space state data and the virtual space prediction data at each preset time and space point in the entire titanium-based alloy production process. The current virtual-real deviation value is compared with the corresponding model calibration trigger threshold to obtain the deviation comparison result. The deviation comparison result is divided into deviation not reaching the trigger threshold, deviation reaching the trigger threshold, and deviation exceeding the trigger threshold by a preset proportion.

7. The intelligent control system for titanium-based alloy composition optimization and performance prediction as described in claim 6, characterized in that, The specific steps for dynamically selecting and switching to the preset feedback signal transmission frequency corresponding to the interval are as follows: Based on the deviation comparison results, a dynamic adjustment rule for the feedback signal transmission frequency is constructed. Specifically, if the current virtual-real deviation value does not reach the corresponding model calibration trigger threshold, the basic value of the feedback signal transmission frequency is maintained. If the current virtual-to-real deviation value reaches the corresponding model calibration trigger threshold, the feedback signal transmission frequency will be adjusted to the corresponding value within the frequency increase range according to the preset increase range, so as to improve the real-time performance of deviation information transmission. If the current virtual-to-real deviation value exceeds the preset ratio of the corresponding model calibration trigger threshold, the feedback signal transmission frequency will be adjusted to the upper limit of the frequency increase range according to the maximum allowable increase, so as to ensure that the deviation information is transmitted quickly to support the emergency calibration of the model.

8. The intelligent control system for titanium-based alloy composition optimization and performance prediction as described in claim 1, characterized in that, The standard adaptation and interaction module includes data transmission timing and spatiotemporal synchronization optimization, and data transmission consistency verification and exception handling optimization. The specific steps for optimizing data transmission timing and spatiotemporal synchronization are as follows: Based on the established mapping relationship between the predicted synchronization time offset and the real-time transmission trigger threshold coefficient of deviation information, the real-time transmission trigger threshold of deviation information in the current control cycle is dynamically adjusted according to the predicted synchronization time offset. If the predicted synchronization time offset is within the offset critical interval, then the current deviation information real-time transmission trigger threshold is maintained. The offset critical interval represents the closed interval formed by the offset critical lower limit and the offset critical upper limit. If the predicted synchronization time offset is less than the critical lower limit of the offset, the current predicted synchronization time offset is input into the mapping relationship between the predicted synchronization time offset and the real-time transmission trigger threshold coefficient of the deviation information, and the trigger threshold gain coefficient is output. The trigger threshold gain coefficient is combined with the current real-time transmission trigger threshold of the deviation information to obtain the target real-time transmission trigger threshold of the deviation information in order to suppress unnecessary transmission.

9. The intelligent control system for titanium-based alloy composition optimization and performance prediction as described in claim 8, characterized in that, The method of dynamically adjusting the real-time transmission trigger threshold of deviation information within the current control cycle also includes: If the predicted synchronization time offset is greater than the critical upper limit of the offset, the current predicted synchronization time offset is input into the mapping relationship between the predicted synchronization time offset and the real-time transmission trigger threshold coefficient of the deviation information, and the trigger threshold reduction coefficient is output. The trigger threshold reduction coefficient is combined with the current real-time transmission trigger threshold of the deviation information to obtain the target real-time transmission trigger threshold of the deviation information. Within the current control cycle, the instantaneous actual deviation from the latest synchronization state is obtained, and the instantaneous actual deviation is compared with the adjusted deviation information real-time transmission trigger threshold. If the instantaneous actual deviation is greater than or equal to the adjusted deviation information real-time transmission trigger threshold, the deviation information real-time transmission is immediately triggered; otherwise, it remains silent and no data transmission is performed.

10. The intelligent control system for titanium-based alloy composition optimization and performance prediction as described in claim 1, characterized in that, The specific steps for data transmission consistency verification and anomaly handling optimization are as follows: Real-time monitoring and calculation of threshold reset delay time: After the deviation information is transmitted in real time and the threshold adjustment is completed, the reset timer is started, the state change of the trigger threshold is monitored in real time, and the start and end times of the trigger threshold resetting from the adjusted value to its preset base value are recorded to obtain the threshold reset delay time. If the threshold reset delay time is less than or equal to the reset reference time, the current preset adjustment step size will be maintained to ensure the stability of subsequent threshold adjustments. If the threshold reset delay time is greater than the reset reference time, it indicates that the threshold reset delay has affected the synchronization time offset correction and virtual-real synchronization control effect. The current threshold reset delay time is input into the preset threshold reset delay time-preset adjustment step size mapping relationship, and the adjustment step size reduction coefficient is output. The current preset adjustment step size and the adjustment step size reduction coefficient are combined to obtain the target preset adjustment step size, so as to ensure the timeliness and effectiveness of subsequent deviation information real-time transmission triggering threshold adjustment.

Images