A dynamic coupling control method and system for a titanium-iron powder reduction process

Abstract

The application discloses a dynamic coupling control method and system for a titanium-iron powder reduction process, and belongs to the technical field of metallurgy. The method comprises the following steps: continuously acquiring multi-source process parameters and performing time sequence feature extraction to generate a dynamic process state data set, combining multi-dimensional phase change determination rules to perform real-time mode matching, and dynamically determining the running stage and real-time state information of the titanium-iron powder reduction process. Based on specific running stages and control target data, dynamic adjustment and parameter optimization of a segmented optimization control strategy are performed. In combination with historical process information, coupling lag effects and synergistic relationships between process links are analyzed, and synergistic effect data are dynamically determined. Based on these data, priority adjustment analysis of the process links is performed, adaptive control instructions are generated, and closed-loop regulation and control of the equipment are performed, so that the reduction process can be smoothly performed. The scheme can ensure stable and efficient titanium-iron powder reduction process, improve production precision and flexibility, and improve product quality and efficiency.

Description

Technical Field

[0001] This application belongs to the field of metallurgical technology, specifically relating to a dynamic coupling control method and system for the reduction process of ferrotitanium powder. Background Technology

[0002] The reduction process of ferrotitanium powder is a key link in the metallurgical industry with high energy consumption and high complexity. Its process control needs to respond in real time to the dynamic changes of multi-source heterogeneous data such as temperature, pressure and gas flow rate in order to ensure product quality and production efficiency.

[0003] The current method first determines the fixed process parameter thresholds for each reduction stage through offline calibration. Then, during the production process, data from a single sensor is collected in real time and compared with the preset thresholds to determine the current stage. Subsequently, the corresponding fixed control strategy, such as constant heating power or gas flow, is called according to the stage identifier. Finally, the actuator is adjusted by traditional controllers such as PID controllers, and the control logic of each link is independent of each other.

[0004] However, the fixed threshold of existing technologies cannot adapt to process drift caused by raw material fluctuations or equipment aging, which can easily lead to misjudgment of stages or control lag. Furthermore, the independent control of each stage breaks the dynamic coupling relationship between each link, resulting in the lack of synergistic effect, such as the time lag matching problem between heating rate and gas permeation. Summary of the Invention

[0005] To overcome the above-mentioned defects, this invention is proposed to provide solutions or at least partially solve the technical problems of existing technologies that are prone to stage misjudgment or control lag, and lack of synergistic effects such as the time lag matching between heating rate and gas permeation.

[0006] In a first aspect, the present invention provides a dynamic coupling control method for the reduction process of ferrotitanium powder, the method comprising: Continuously acquire multi-source process parameters during the reduction of ferro-titanium powder, and dynamically reconstruct hierarchically and extract time-series features based on the multi-source process parameters to obtain a dynamic process state dataset; Based on the preset multidimensional phase transformation determination rules, the dynamic process state dataset is mapped to the preset process state domain, and real-time pattern matching and classification are performed to dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information. Acquire preset control target data, perform dynamic correlation analysis based on specific operating stages, real-time status information and preset control target data, and dynamically determine segmented optimization control strategies that match specific operating stages; Acquire and dynamically update the current control parameters, perform real-time multivariate decoupling analysis and parameter optimization based on the piecewise optimization control strategy and the current control parameters, and dynamically determine the optimal control parameters; Historical process information is acquired, and dynamic analysis of coupling lag effects and synergistic relationships between various process steps is conducted based on optimized control parameters and historical process information. The synergistic effect data between each process step and optimized control parameters is dynamically determined. Based on the synergistic effect data, priority adjustment analysis of each process step and optimized control parameter is performed, adaptive control commands are dynamically generated, and the actuators of the reduction equipment are controlled and dynamically corrected in a closed loop based on the adaptive control commands until the titanium iron powder reduction process is completed.

[0007] In a second aspect, the present invention provides a dynamic coupling control system for the reduction process of ferrotitanium powder, the system comprising: The data acquisition module is used to continuously acquire multi-source process parameters during the reduction of ferro-titanium powder, and to dynamically reconstruct the hierarchical structure and extract time-series features based on the multi-source process parameters to obtain a dynamic process state dataset. The matching and classification module is used to map the dynamic process state dataset to the preset process state domain based on the preset multi-dimensional phase transformation determination rules, perform real-time pattern matching and classification, and dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information. The dynamic correlation analysis module is used to acquire preset control target data, perform dynamic correlation analysis based on specific operating stages, real-time status information and preset control target data, and dynamically determine the segmented optimization control strategy that matches the specific operating stage. The optimization module is used to acquire and dynamically update the current control parameters. Based on the segmented optimization control strategy and the current control parameters, it performs real-time multivariate decoupling analysis and parameter optimization to dynamically determine the optimized control parameters. The coupling analysis module is used to acquire historical process information, and to perform dynamic analysis of the coupling lag effect and synergistic relationship between various process links based on the optimized control parameters and historical process information, so as to dynamically determine the synergistic effect data between each process link and the optimized control parameters. The correction module is used to perform priority adjustment analysis of each process link and optimized control parameters based on the synergistic effect data, dynamically generate adaptive control commands, and perform closed-loop control and dynamic correction of the actuator of the reduction equipment based on the adaptive control commands until the titanium iron powder reduction process is completed.

[0008] In a third aspect, an electronic device is provided, comprising a processor, a memory, and a program or instructions stored in the memory and executable on the processor, the program or instructions being loaded and run by the processor to perform the steps of the aforementioned dynamic coupling control method for the reduction process of ferrotitanium powder.

[0009] In a fourth aspect, a computer-readable storage medium is provided, wherein a plurality of program codes are stored therein, the program codes being adapted to be loaded and run by a processor to perform the steps of the aforementioned dynamic coupling control method for the reduction process of ferro-titanium powder.

[0010] The above-described technical solutions of the present invention have at least one or more of the following beneficial effects: In implementing the technical solution of this invention, multi-source process parameters are dynamically acquired, control strategies are optimized in real time, and each process step is adjusted to ensure a stable and efficient reduction process of ferrotitanium powder. Adaptive control commands achieve closed-loop regulation, improving production precision and flexibility, and enhancing product quality and efficiency. Attached Figure Description

[0011] The disclosure of this invention will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Furthermore, similar numbers in the drawings are used to denote similar components, wherein: Figure 1 This is a schematic diagram of the first main steps of a dynamic coupling control method for the reduction process of ferrotitanium powder according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the second main step of a dynamic coupling control method for the reduction process of ferrotitanium powder according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the main structure of a dynamic coupling control system for a titanium-iron powder reduction process according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0012] Some embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0013] In the description of this invention, "module" and "processor" can include hardware, software, or a combination of both. A module can include hardware circuitry, various suitable sensors, communication ports, memory, and may also include software components, such as program code, or a combination of software and hardware. A processor can be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and / or signal processing capabilities. The processor can be implemented in software, in hardware, or a combination of both. Non-transitory computer-readable storage media includes any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" means all possible combinations of A and B, such as only A, only B, or A and B. The terms "at least one A or B" or "at least one of A and B" have a similar meaning to "A and / or B" and can include only A, only B, or A and B. The singular terms "a" or "this" can also include plural forms.

[0014] See appendix Figure 1 , Figure 1 This is a schematic flowchart of the first main steps of a dynamic coupling control method for the reduction process of ferrotitanium powder according to an embodiment of the present invention. Figure 1 As shown, the dynamic coupling control method for the reduction process of ferro-titanium powder in an embodiment of the present invention mainly includes the following steps S101-S106.

[0015] Step S101: Continuously acquire multi-source process parameters during the reduction of ferrotitanium powder, dynamically reconstruct hierarchically based on the multi-source process parameters and extract time-series features to obtain a dynamic process state dataset.

[0016] Multi-source process parameters are various data closely related to the process, collected from multiple different sources. These include real-time sensor data, production equipment records, and environmental monitoring feedback. Common data sources include temperature, pressure, flow rate, and atmospheric composition. These parameters comprehensively reflect the actual production environment, the real-time operating status of the process, and the impact of various external factors on the process.

[0017] The dynamic process status dataset is a structured data set built upon multi-source process parameters continuously collected during the process, using data processing techniques such as time-series feature extraction and hierarchical structure reconstruction. This dataset can map the characteristic performance of the process at different time points and under different operating states in real time, preserving the evolution patterns of each process parameter over time, as well as the inherent correlations between parameters.

[0018] In the reduction process of ferrotitanium powder, the acquisition of multi-source process parameters requires specialized detection equipment. This involves deploying double platinum-rhodium thermocouples, high-precision pressure transmitters, and non-dispersive infrared gas analyzers in the key temperature zones of the reduction furnace to capture raw signals of temperature, system pressure, and the composition of the reducing atmosphere in real time. The underlying sensors utilize industrial Ethernet for high-frequency polling, achieving millisecond-level sampling frequencies to rapidly convert analog signals to digital signals, ensuring timely and accurate data transmission. During data transmission to the gateway, a precise time synchronization protocol is employed to synchronize data from all acquisition points, ensuring that parameter data at different locations remain consistent over time and preventing data deviations. Simultaneously, a Kalman filter algorithm is applied to optimize the raw data, effectively eliminating interference signals and filtering out invalid information, ultimately generating a continuous raw digital sequence.

[0019] After data acquisition, the acquired parameter sequences need to be restructured hierarchically, transforming the originally scattered parameter data into structured information with clear logic. Using semantic tagging technology, combined with the equipment's physical topology, all process parameters are precisely associated with different functional sections of the reduction furnace, clarifying the corresponding process steps for each parameter. Simultaneously, using an object-oriented representation approach, each acquisition point is defined as a digital entity with complete attributes, constructing a three-level nested architecture at the equipment, location, and measurement point levels to achieve precise correspondence between parameters and process steps. Building upon this, the node association function of a graph database is used to bind static equipment parameters with dynamic operating parameters, forming a complete process logic system that reveals the inherent relationships between the parameters.

[0020] After data reconstruction, in-depth temporal feature extraction of process parameters is required. A sliding window technique is used to extract relevant process data segments, and the stability of the reduction process is quantified by calculating parameters such as mean and standard deviation. For fluctuations in the reduction atmosphere, discrete wavelet transform is used to decompose the signal at multiple scales, accurately capturing subtle changes. Simultaneously, first- and second-order difference operations are used to extract dynamic change features of the parameters, accurately identifying process state switching nodes. Furthermore, mutual information and Pearson correlation coefficients are calculated to quantify the correlation between parameters, comprehensively uncovering the dynamic change patterns in the process and ensuring that the extracted features fully reflect the process operating status. Finally, a dynamic process state dataset is constructed by fusing and mapping hierarchical physical topology information with the extracted temporal features. A vector space mapping mechanism is used to encapsulate the feature parameters of different process stages into a unified feature vector, and a dedicated time-series database is used to achieve efficient data storage and index optimization. Simultaneously, batch information and process indicators of ferrotitanium powder are linked through dynamic linking technology to form a self-describing process state dataset. This dataset can map the operating status of the reduction furnace in real time, clearly presenting subtle changes in the process.

[0021] Step S102: Based on the preset multidimensional phase transformation determination rules, the dynamic process state dataset is mapped to the preset process state domain, and real-time pattern matching and classification are performed to dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information.

[0022] Pre-defined multi-dimensional phase transition judgment rules are a set of process state judgment and identification criteria developed in advance, taking into account the actual characteristics of the production process. These rules are based on changes in multiple process parameters such as temperature, pressure, atmosphere composition, and flow rate, while also matching actual process operation requirements to clearly define the triggering conditions for process state changes. For example, specific atmosphere fluctuations or temperature changes can directly affect the reduction rate; pre-defined multi-dimensional phase transition judgment rules determine whether a process state switch is necessary by capturing the changing characteristics of these key parameters.

[0023] A preset process state domain is a collection of all possible process states during the process, including different stages and conditions of the production flow. Each process state domain corresponds to a specific set of process parameter ranges, which define the process characteristics of that stage. In the reduction process of ferrotitanium powder, typical process state domains include the initial heating stage, the stable reduction stage, and the over-reduction stage, each with corresponding parameter ranges such as temperature, pressure, and flow rate. Preset process state domains provide a clear decision framework for the process management system, enabling the system to accurately distinguish and map the current process stage during real-time processing.

[0024] The specific operational stages are the different process steps in the reduction of ferroilamide powder, each with unique process characteristics and operational requirements. The reduction of ferroilamide powder typically involves several specific operational stages, including heating, preheating, stabilization reduction, and cooling.

[0025] Real-time status information is the sum of process-related data acquired in real time during the production process through sensors, monitoring equipment, and data acquisition systems. It includes not only basic process parameters such as temperature, pressure, flow rate, and atmospheric composition, but also derived information such as the trends and fluctuations of these parameters.

[0026] In the mapping process, a Kalman filter algorithm is first used to smooth the multi-source parameters in the dynamic process state dataset in real time, effectively eliminating high-frequency noise generated by sensors. Simultaneously, a sliding window technique is employed to extract the time-domain statistical characteristics and first-order differential rate of change of each parameter. Under this premise, a preset multi-dimensional phase transition determination rule is introduced. Through logical conjunction and multi-criteria decision analysis, the extracted real-time change features, such as temperature gradient and atmosphere concentration fluctuations, are compared one by one with the state switching trigger conditions in the rule. This process utilizes a fuzzy inference mechanism to quantify the degree to which the current parameter combination deviates from the steady-state operating point, accurately capturing the key node where the reduction rate transitions from kinetic control to diffusion control, and then outputting the probability trigger signal for state transition at the logical level. Using Euclidean distance or Mahalanobis distance metric algorithms, the real-time multi-dimensional feature vectors are projected into a hyperspace composed of a preset process state domain. Through support vector machine or kernel clustering analysis methods, real-time pattern matching is performed between the current dynamic feature points and the cluster centers of the preset process state domain, which includes various process stages such as the initial heating domain and the stable reduction domain. By calculating the membership degree of feature points falling within a specific parameter range, the system dynamically identifies and classifies the specific operational stage of the reduction process of ferrotitanium powder. During the classification process, the system automatically matches the preset temperature, pressure, and flow rate benchmarks for that stage to ensure that the identification results accurately correspond to the actual process steps such as heating, preheating, or reduction of ferrotitanium powder in the physical space.

[0027] After determining the specific operational stage, the system further extracts the corresponding real-time state information through a state observer. The system uses the least squares method to linearly fit the real-time data stream of the current stage, deriving the evolution trend term and instantaneous fluctuation amplitude of the process parameters. Combining the determined operational stage labels, the system performs residual calculations between the original temperature, pressure, and composition data and the corresponding stage characteristic intervals, ultimately outputting a sum of real-time state information including the stability index, parameter offset, and process evolution vector. Through this mapping and classification mechanism, the system can not only make macroscopic qualitative judgments on the reduction stage of ferrotitanium powder but also quantify the parameter fluctuation amplitude and its deviation from a preset benchmark.

[0028] The process of constructing the preset process state domain is as follows: In the reduction process of ferrotitanium powder, the construction of the preset process state domain needs to be based on historical process big data. First, various process parameter data during the reduction process of ferrotitanium powder are retrieved, and cluster analysis algorithms are used to sort out these data, identify the parameter patterns corresponding to different operating stages, and then determine the core parameter range of the preset process state domain.

[0029] When constructing a preset process state domain, it is necessary to combine the actual process requirements of ferrotitanium powder reduction, integrate the standard ranges of key parameters such as temperature, pressure, and flow rate, clarify the parameter boundaries of different process stages, and form a set of process states covering the entire process. The construction of the preset process state domain requires first identifying the key parameters in the ferrotitanium powder reduction process, combining historical process data, determining the parameter standards for different process stages, and dividing the corresponding state intervals to ensure that each state domain accurately corresponds to the actual production stage, making subsequent pattern matching more targeted.

[0030] By integrating historical process data and actual production experience, the parameter range of the preset process state domain is clearly defined, so that each process stage has a clear parameter benchmark, ensuring that the preset process state domain can accurately reflect the actual situation of titanium iron powder reduction.

[0031] Based on the above technical solution, optionally, the dynamic process state dataset is mapped to a preset process state domain based on preset multidimensional phase transition determination rules, and real-time pattern matching and classification are performed to dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information, including: Based on the preset multidimensional phase transition determination rules, the system continuously performs real-time feature extraction and pattern recognition on dynamic process state datasets to dynamically determine the key determination conditions for process state changes. Based on key judgment conditions, the dynamic process state dataset is mapped to the preset process state domain in real time, and the process state is classified to obtain the preliminary matching label for process state matching. Based on the initial matching tags, time-series consistency comparison and rule reasoning are performed to dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information.

[0032] In this scheme, the key judgment condition is the extraction of process parameter change thresholds and related characteristics during the reduction of ferrotitanium powder, based on a preset multidimensional phase transformation judgment rule. These judgment conditions include the change patterns of process parameters such as temperature, pressure, atmosphere composition, and flow rate, which can directly trigger the switching of process states. Specific fluctuations in atmosphere composition or temperature changes will directly affect the reduction rate, and these change characteristics are the basis for determining whether to switch process states.

[0033] Preliminary matching labels are identifiers generated based on process state classification results. They are primarily used to characterize the degree of matching between the current dynamic process state dataset and a specific state within a preset process state domain. The generation process involves comparing and analyzing real-time process data with the preset process state domain to preliminarily identify the current process stage or specific state. For example, when the values ​​of temperature and atmosphere composition meet the threshold range of a certain process stage, the identifier corresponding to that stage becomes the preliminary matching label.

[0034] The system continuously employs a sliding window algorithm to extract features such as temperature gradient, atmospheric concentration change rate, and system pressure fluctuation from the dynamic process state dataset. Simultaneously, it captures the instantaneous evolution trend of process parameters through first- and second-order difference operations. Based on this, these feature variables are mapped in real-time to preset multi-dimensional phase transition judgment rules. Using Boolean logic comparison and dynamic threshold cross-validation, the deviation of each process dimension from the state transition critical point at the current moment is quantified. This comparison method can accurately identify the change patterns of process parameters, and then dynamically output key judgment conditions, including specific process index thresholds, providing a reliable logical trigger for the immediate transition of the process state. Specifically, in this comparison and verification process, Boolean logic comparison analyzes the real-time collected process parameters one by one, ensuring that each parameter is accurately matched with the preset critical point. When a parameter deviates from the set threshold, the system records this deviation and assesses its potential impact on the process state. By continuously quantifying the deviation of each process parameter, the system can obtain the degree of deviation from the state transition critical point, thereby more accurately determining whether the current process flow needs to be adjusted. The introduction of dynamic threshold cross-validation ensures that the comparison process can continuously adapt to changes in actual processes. Process parameters are constantly affected by external environments or operating conditions, so a single static threshold may lead to outdated or erroneous judgments. By dynamically adjusting the threshold range, the system can flexibly adjust the threshold based on real-time process data, making the judgment process more accurate and real-time. The implementation of dynamic threshold cross-validation ensures that the system can always make decisions that conform to the actual situation under different process requirements and operating environments at different stages. Finally, through Boolean logic comparison and dynamic threshold cross-validation, the system can quantify the degree of deviation of each process dimension from the state transition critical point at the current moment, and thus accurately output key judgment conditions, including specific process indicator thresholds.

[0035] After obtaining the key decision conditions, the system employs vector space projection technology to project the current dynamic process state dataset into a multi-dimensional hyperspace composed of preset process state domains in real time. This process uses Euclidean distance or Mahalanobis distance metric algorithms to calculate the geometric closeness between the current real-time feature vector and the center and boundary of each preset state domain. For example, it calculates the geometric closeness between the current real-time feature vector and the center and boundary of the initial heating domain and the stable reduction domain. Through fuzzy comprehensive evaluation or support vector machine classification algorithms, the current working condition is classified into the most matching parameter feature space, and the corresponding state identifier is automatically assigned, thereby generating preliminary matching labels that reflect the degree of coupling between the current dataset and the physical stage.

[0036] Based on the initial tag matching, the system initiates a timing consistency comparison algorithm. Within a specific time window, it applies de-jitter logic and state smoothing filtering to eliminate instantaneous tag jumps caused by random sensor noise. Subsequently, combining preset process timing constraints and IF-THEN expert rule reasoning, it verifies whether the tag evolution path conforms to the physicochemical timing law of ferrotitanium powder reduction. After confirming that the tag remains stable, the system finally determines the specific operational stage of the current reduction process. Simultaneously, it calls the state observer and extracts the parameter deviation, stability index, and trend vector of that stage using the least squares method, integrating and encapsulating them to form real-time state information including fluctuation details across all dimensions.

[0037] This solution extracts and analyzes multi-source process data in real time to accurately determine changes in process status, optimizes control decisions for the reduction of ferrotitanium powder, improves process stability, reduces errors, ensures the process proceeds smoothly according to predetermined parameters, and improves production efficiency and product quality.

[0038] Step S103: Obtain preset control target data, perform dynamic correlation analysis based on specific operating stages, real-time status information and preset control target data, and dynamically determine the segmented optimization control strategy that matches the specific operating stage.

[0039] Preset control target data are target values ​​of control parameters that are pre-set during the process design and production planning stages, taking into account process requirements and production goals. This type of data includes ideal ranges or target values ​​for various process parameters such as temperature, pressure, flow rate, and atmosphere composition, used to guide the control and regulation of the production process. Its setting is typically based on historical operating data, actual process requirements, and engineering design standards, representing the ideal operating state that the process system is expected to achieve.

[0040] The segmented optimization control strategy divides the complete process into multiple independent stages, and then optimizes the control parameters independently for each stage based on its specific conditions and requirements. Each process stage has its own optimization goals and control strategies, which can dynamically adjust control parameters according to different operating conditions, thereby achieving overall process optimization and upgrading. The core of this strategy lies in developing targeted control plans based on different operating stages or states of the process, ensuring that optimal control effects are achieved at each stage.

[0041] When acquiring preset control target data, the original design specifications from the process design phase and historical production records of high-quality batches are first retrieved. Using big data statistical analysis technology, features are extracted from historical operating curves to accurately identify the parameter distribution patterns when the product quality reaches its optimal level. Subsequently, combined with the thermodynamic calculation results of the titanium-iron powder reduction reaction, a chemical equilibrium constraint scheme based on minimizing Gibbs free energy is established, clarifying the ideal mapping relationship between temperature, atmosphere ratio, and reduction rate at different stages. Through an expert knowledge base system, these physical constraints are logically integrated with production scheduling targets to generate a parameter matrix including target values ​​for each node in the entire process. Then, using database persistence technology, this matrix is ​​stored as a standardized control baseline sequence.

[0042] During dynamic correlation analysis, the system receives real-time labels of the identified specific operating stages, as well as real-time status information including parameter offsets and trends. Using multiple linear regression or grey relational analysis, it calculates the Euclidean distance and dynamic correlation between real-time status parameters and preset control target data, quantitatively determining the degree to which the current operating condition deviates from the ideal trajectory. Then, it applies a cross-sensitivity algorithm to analyze the weights of temperature, pressure, and flow rate on the reduction efficiency at the current stage, superimposing real-time fluctuation characteristics with the target baseline in a multi-dimensional spatial vector to identify the key control factors restricting the achievement of efficiency targets at the current stage. Through this correlation mapping method, the system can analyze in real-time the root causes of the discrepancy between the current operating stage and the preset target, transforming them into an optimization control deviation vector to be processed.

[0043] In the process of dynamically determining the segmented optimal control strategy, based on the deviation vector obtained from correlation analysis, a gain adaptive algorithm is used to match the most suitable control law from a preset strategy library according to the characteristics of the current operating stage. For the heating stage, Bang-Bang control based on the time-optimal principle combined with the PID algorithm is adopted to achieve rapid temperature response and ensure efficient progress of the heating process. In the stable reduction stage, a multivariate optimization technique based on model predictive control is introduced. By solving the quadratic programming problem online, the control gain and command step size are dynamically adjusted under the premise of satisfying pressure and atmosphere fluctuation constraints to ensure the stability and efficiency of the reduction process. The determination of the control strategy for each stage requires priority screening from a preset set of control schemes through a heuristic search algorithm. Finally, a segmented optimal control strategy that is highly consistent with the current physicochemical properties and has segmented differentiated characteristics is formed to ensure that the titanium iron powder reduction process can approach the optimal operating state at each stage. The pre-set control scheme set refers to a structured strategy reserve library that is pre-constructed based on the kinetic characteristics of the titanium iron powder reduction process and expert experience, and includes multi-dimensional process parameter linkage logic. It consists of a series of control templates customized for different reaction stages, such as heating and dehydration, surface reduction and deep diffusion. It aims to provide a limited and optimized decision space for the search algorithm, and ensure that the system can accurately select and call the logic combination with the highest matching degree from the library according to the changes in the real-time physical state.

[0044] Based on the above technical solution, optionally, dynamic correlation analysis can be performed based on specific operating stages, real-time status information, and preset control target data to dynamically determine a segmented optimization control strategy that matches the specific operating stage, including: Dynamic process parameters are extracted based on specific operational stages and real-time status information to continuously determine the real-time process parameter set. Dynamic dependency analysis is performed based on real-time process parameter sets and preset control target data to evaluate the dependency relationship between each real-time process parameter and the preset control target data in real time. Real-time sensitivity analysis is performed based on the dependency data to evaluate the impact of each real-time process parameter on the preset control target data. The data on the impact of each real-time process parameter on the preset control target data are sorted in real time, and key process parameters are dynamically filtered out based on preset screening thresholds to obtain a real-time updated list of key process parameters. Based on the list of key process parameters and the target parameter requirements data corresponding to specific operating stages, dynamic target value optimization is performed to dynamically determine a segmented optimization control strategy that matches the specific operating stage.

[0045] In this solution, the real-time process parameter set is a dataset composed of various real-time data collected during the process, including parameters such as temperature, pressure, and flow rate, reflecting the real-time status of the production environment and process operation.

[0046] Real-time process parameters are various variables in the process collected at specific time points, including parameters such as temperature, pressure, and flow rate. They are always in a dynamic state and can accurately capture the instantaneous state of process conditions during production.

[0047] Dependency data is used to describe the interaction and relationship between different process parameters. By analyzing the correspondence between real-time process parameters and preset control targets, the specific influence mechanism of each process parameter on the control target can be clarified.

[0048] Impact data is a numerical indicator that quantifies the degree to which each real-time process parameter affects the control target. It can reflect the proportion of contribution of process parameter changes to the preset control target, and provide data support for screening parameters with high impact weight and carrying out targeted optimization and adjustment.

[0049] The preset screening threshold is a parameter judgment standard formulated by combining process objectives and actual operating requirements. It is specifically used to screen process parameters that have a significant impact on control objectives. Only process parameters whose impact exceeds the threshold will be included in the subsequent process optimization process.

[0050] Key process parameters are those that play a decisive role in the process objectives at a specific stage of the process. They occupy a core position in the entire process, and fluctuations in their parameters directly affect the quality level and production efficiency of the final product.

[0051] The critical process parameter list is a set of key process parameters obtained through scientific screening for a specific process stage. The parameters in this list have a significant impact on the process objectives and should be given priority control during real-time process monitoring and parameter adjustment.

[0052] The target parameter requirements data are parameter standards that are pre-set based on the technical requirements of a specific process stage, including key indicators such as temperature, pressure, and flow rate.

[0053] Based on the identified specific operational stages and real-time status information, the system employs attribute mapping technology to accurately select real-time process parameters from the sensor array that match the current thermodynamic logic. High-frequency polling sampling is performed via the industrial bus to capture the instantaneous values ​​of various parameters, which are then integrated and packaged into a standardized set of real-time process parameters. During this process, Z-score standardization is simultaneously employed to eliminate numerical differences between parameters of different dimensions, such as temperature and pressure, ensuring a clear representation of the immediate physical characteristics of the process conditions under a unified benchmark. Combining the real-time process parameter set with preset control target data, the system comprehensively utilizes mutual information algorithms and Pearson correlation analysis to quantify the logical relationships between variables. By calculating the statistical correlation between fluctuations in each process parameter and target yield and quality indicators, and simultaneously using grey relational analysis to deeply explore the nonlinear mapping relationships between variables, the system ultimately generates data reflecting the dependency of input variables on the target's guiding effect, thereby clearly defining the logical support strength of each process dimension.

[0054] Based on the generated dependency data, the system performs a Sobol global sensitivity analysis to quantify the contribution of each process parameter to the control target variance. By calculating partial derivatives near the current operating point, the system assesses the magnitude of the system response caused by small changes in each variable, thereby acquiring real-time impact data. This data precisely calibrates the real-time sensitivity of factors such as heating power and atmosphere ratio to the achievement of preset targets in numerical form. The system uses a fast sorting algorithm to arrange the impact data in descending order of contribution, while introducing a preset screening threshold as a logical filtering boundary. By automatically eliminating secondary variables with impact below the threshold, the system dynamically retains the parameter characteristics that play a dominant role in the reduction reaction, generating a list of key process parameters that can be updated in real time. This filtering mechanism ensures that control resources are concentrated on high-sensitivity process links, avoiding ineffective adjustments to inefficient parameters.

[0055] Finally, the system cross-matches the list of key process parameters with the target parameter requirements for the current stage. Using constrained genetic algorithms or linear programming techniques, it searches for the parameter combination that optimizes the restoration efficiency for the current stage within the physical operating limits of the equipment. Through iterative calculations to optimize the target values ​​of each key process parameter, the system generates precise parameter setpoints and adjustment step instructions in real time. These optimization results are then mapped to the process logic of the current operating stage, ultimately dynamically formulating a segmented optimization control strategy highly adapted to the specific operating stage.

[0056] This solution continuously adjusts control strategies by dynamically extracting and optimizing real-time process parameters, thereby improving the accuracy and stability of the production process, increasing production efficiency, reducing unnecessary energy waste, ensuring product quality, meeting process objectives, and optimizing the production process.

[0057] Based on the above technical solution, optionally, real-time multivariate decoupling analysis and parameter optimization can be performed based on a piecewise optimization control strategy and current control parameters to dynamically determine the optimal control parameters, including: Based on the segmented optimization control strategy and the current control parameters, the coupling relationship analysis of key process parameters is continuously carried out to dynamically determine the coupling correlation matrix; The adaptive range of the current control parameters is dynamically derived based on the coupling correlation matrix, resulting in the real-time adaptive range of the current control parameters and the real-time constraint conditions. Dynamic adaptive adjustment is performed based on the real-time adaptive range and real-time constraints to obtain the real-time adjustment range of the current control parameters; Based on the real-time adjustment range, the current control parameters are adjusted in real time to dynamically determine the optimal control parameters.

[0058] In this scheme, the coupling correlation matrix is ​​used to characterize the mutual influence and dependence among the key process parameters. Based on the analysis of the dynamic coupling effect between parameters, this matrix clarifies the degree of influence between each parameter. Each value in the matrix corresponds to a characterization of the coupling strength between two process parameters.

[0059] The real-time adaptive range is the fluctuation range within which control parameters maintain stable system operation under specific operating conditions. This range is dynamically updated based on real-time collected operating data and system response results, ensuring that system control parameters remain within the effective control range under different operating conditions.

[0060] Real-time constraints are a series of limiting rules that the system must follow to ensure safe and efficient operation during the process. These conditions mainly include elements such as the extreme range of control parameters and the threshold of the rate of change. By specifying the operating boundaries of the parameters, the system is ensured to be in a safe operating state at each stage of operation.

[0061] The real-time adjustment range is the dynamic adjustment amplitude of the control parameters after combining the real-time adaptability interval and real-time constraints. This range clarifies the dynamic adjustment boundary of the parameters under actual operating conditions, supporting the control parameters to adapt and adjust in a timely manner according to changes in process conditions, ultimately achieving optimized operation of the process.

[0062] Guided by a clearly defined control objective based on a segmented optimization control strategy, real-time values ​​of control parameters are retrieved synchronously. Using relative gain matrix and partial correlation analysis techniques, the degree of cross-interference between parameters is quantitatively analyzed. Multivariate sensitivity calculations are performed around the current operating point to quantify the mutual mapping strength between heating power increment, atmosphere flow fluctuation, and furnace temperature response. Subsequently, according to the variable topological relationships, the calculated influence coefficients of each dimension are matrixed, ultimately dynamically constructing a coupling correlation matrix to characterize the physical dependence strength between parameters. Based on the constructed coupling correlation matrix, singular value decomposition (SVD) is used to spatially decouple the multidimensional control system, analyzing the system's stability margin under specific operating conditions. By solving the Lyapunov stability criterion of the control law, the maximum deviation limit of the parameters without inducing system oscillation or runaway reduction reaction is derived in reverse, thereby dynamically defining the real-time adaptability range of the current control parameters. Simultaneously, combining the physical range of the reduction equipment and process safety specifications, the extreme value boundaries and rate of change limits of the actuators are clarified using dynamic threshold extraction techniques, forming real-time constraints to ensure the safe operation of the equipment.

[0063] The system performs Boolean logic intersection operations on the real-time adaptive range and real-time constraints, and uses interval arithmetic techniques to eliminate invalid fluctuation regions that do not comply with safety regulations or exceed stability margins. During this process, a multi-dimensional spatial projection algorithm is used to verify the coincidence between the stability boundary and the physical boundary of the equipment, locking in a legal and compliant subset of parameters that can be adjusted under the current operating conditions. By calculating the upper and lower limits and dynamic envelope of this subset, a real-time adjustment range is dynamically generated. This range reflects both the flexibility of parameter adjustment and a clear safety boundary, ensuring that subsequent parameter corrections can balance process optimization needs with equipment operational safety. Using the real-time adjustment range as the constraint boundary and the current control parameters as the starting point of iteration, the optimal solution of the objective function is searched within the constrained space using the interior point method or the Lagrange multiplier method. The system calculates the deviation vector between the current state and the ideal restoration target in real time, and performs gradient optimization iterations within the real-time adjustment range to dynamically determine the optimal control parameters that maximize the restoration rate and minimize energy consumption fluctuations. This process uses a closed-loop feedback mechanism to correct parameter increments in real time, ensuring that the final optimized values ​​can directly drive the precise actions of the restoration equipment actuators, achieving adaptive and fine-tuned control of the entire titanium-iron powder reduction process.

[0064] In this solution, by dynamically analyzing coupling relationships and adjusting control parameters in real time, the system is ensured to maintain an optimized working state under changing process conditions, thereby improving process stability and accuracy, meeting real-time production requirements, and enhancing the system's adaptability and efficiency.

[0065] Step S104: Obtain and dynamically update the current control parameters, perform real-time multivariate decoupling analysis and parameter optimization based on the segmented optimization control strategy and the current control parameters, and dynamically determine the optimized control parameters.

[0066] Current control parameters refer to various parameters that are adjusted and controlled in real time according to actual production needs during process operation. These parameters are directly related to the immediate state of the production process and are flexibly adjusted according to process progress and actual operating conditions to ensure smooth production flow. For example, in the reduction of ferrotitanium powder, the current control parameters involved include heating power, atmosphere flow rate, and reactor temperature setting.

[0067] Optimized control parameters are parameter values ​​obtained by adjusting and optimizing the current control parameters. For example, increasing the heating power or adjusting the atmosphere composition at certain stages can optimize the efficiency and stability of the entire process.

[0068] When acquiring and dynamically updating current control parameters, the system periodically polls the operating status of the actuators via OPCUA or ModbusTCP protocol, using the real-time database interface of the distributed control system. It directly reads key data such as the heater output power ratio, the real-time opening degree of the mass flow meter, and the current setpoint temperature of the reactor. These parameters are integrated and encapsulated into the current control vector in real time. Simultaneously, an exponentially weighted moving average technique within a sliding window is used to smooth the data stream, effectively eliminating transient glitches during signal transmission. By performing differential calculations between these real-time acquired physical quantities and the command values ​​from the previous adjustment cycle, the system can capture incremental changes in control variables in real time and store them in a memory database with millisecond-level response capabilities. Subsequently, based on a piecewise optimization control strategy and considering the physical characteristics of the current operating stage, the system invokes a multivariate decoupling algorithm system composed of relative gain analysis, singular value decomposition, and feedforward compensation. In this process, the relative gain matrix technique is first used to quantitatively analyze the coupling strength between heating power, atmosphere flow rate, furnace temperature, and reduction rate, clearly identifying the mutual interference relationships between various parameters. Furthermore, by employing feedforward compensation technology, the highly coupled multi-input multi-output system is decomposed into multiple independent single-loop logics, effectively avoiding drastic temperature fluctuations caused by adjusting the atmosphere flow rate. Within this decoupling framework, the algorithm can accurately isolate the cross-influence between variables and establish an independent response evaluation axis for each controlled variable.

[0069] After real-time decoupling, the system enters the parameter optimization phase, dynamically determining the optimal control parameters. This process employs online rolling optimization technology from model predictive control. Guided by preset control objectives, and under the premise of strictly meeting the equipment's rated power and physical range limitations, it searches for the optimal parameter combination by solving a nonlinear programming problem. For real-time state deviations, a heuristic search is conducted in the solution space using a genetic algorithm or particle swarm optimization algorithm to calculate the parameter correction amount that maximizes reduction efficiency and minimizes energy consumption. This step uses the current control parameters as the starting point for iteration, performing rapid convergence calculations using gradient descent or Newton's method to dynamically derive the optimal control parameters for the next stage, such as precise heating power settings or atmosphere composition ratios down to the unit increment, achieving refined control of the ferrotitanium powder reduction process while considering efficiency and energy-saving requirements. Finally, the generated optimal control parameters undergo a logical consistency check to ensure that their change step size meets the stability requirements of the process system, avoiding disruptions to normal production operations due to parameter mutations. The system uses deviation prediction technology to estimate the potential system response after the optimized parameters are issued. If the predicted output is within a preset safety boundary, the optimized setpoint is written to the controller in the execution layer in real time via a reverse communication link. This dynamic update cycle continuously collects new current control parameters and repeats the above optimization process, enabling the system to automatically adjust control parameters according to real-time fluctuations in operating conditions at different stages of ferrotitanium powder reduction, such as heating and stable reduction. Through this closed-loop optimization mechanism, the optimized control parameters can dynamically evolve in real time to follow the drift of the process state.

[0070] Step S105: Obtain historical process information, and perform dynamic analysis of the coupling lag effect and synergistic relationship between each process step based on the optimized control parameters and historical process information, and dynamically determine the synergistic effect data between each process step and the optimized control parameters.

[0071] Historical process information is a comprehensive record of all relevant data from past process operations. This data includes multiple aspects such as temperature, pressure, flow rate, reaction time, equipment operating status, and environmental conditions, fully presenting the process's operating status at different time periods. It also records the interactions between various process steps, changes in process parameters, and corresponding production results.

[0072] A process step refers to the various operations and processing steps in the production process, divided according to a predetermined sequence or time stage. Each process step has its unique process characteristics and clear operational requirements, and there are close mutual influences and dependencies between the steps. Taking the reduction process of ferrotitanium powder as an example, common process steps include the heating stage, the stabilization reduction stage, and the cooling stage.

[0073] Synergy effect data is obtained by analyzing the interactions between multiple process steps and control parameters, reflecting the interdependence and coordination among these steps. Synergy effect data can reveal the joint response of each process step under specific conditions, demonstrating cooperation and mutual intervention between different steps, and promoting process optimization and the achievement of production goals.

[0074] During the historical process information acquisition phase, the system is equipped with a dedicated industrial data acquisition engine. Using extraction, transformation, and loading tools, it retrieves long-term operational records from a historical database. These records cover thousands of batches of ferrotitanium powder reduction data, specifically including temperature fluctuation curves, pressure gradients, reducing agent consumption rates, and corresponding product metallization rates under different operating conditions. During data import, the system employs a time-series cleaning algorithm to remove dead values ​​and outliers caused by sensor malfunctions, while using cubic spline interpolation to complete missing sampling points. The processed data, after normalization and transformation, is uniformly stored as a structured historical archive supporting high-concurrency queries.

[0075] When conducting dynamic analysis of the coupling lag effects and synergistic relationships among various process stages, the system performs multi-dimensional comparisons between the current optimized control parameters and retrieved similar historical operating conditions. This process uses cross-correlation functions to calculate the time delay distribution between heating power changes and furnace temperature response and reduction reaction rate, thereby accurately locating the lag time constants of different process stages under the dual influence of thermal inertia and chemical kinetics. Subsequently, the system uses partial least squares regression to establish a multivariate synergistic contribution matrix, removing the interaction weights of temperature, pressure, and flow rate under specific process stages. This allows the system to quantify the chain reactions of single-stage parameter adjustments on downstream stages, thereby revealing the implicit dynamic intervention logic within the process flow.

[0076] When dynamically determining the synergistic effect data of each process step and optimized control parameters, the system, based on the aforementioned lag analysis results, uses structural equation modeling or path analysis techniques to construct synergistic paths of control variables to production results. This step uses optimized control parameters as the driving source, combining feedback patterns from historical process information to calculate the joint gain coefficient of each process step under synergistic effects. The system then uses nonlinear sensitivity analysis to simulate the contribution of each parameter combination to the global restoration efficiency under the current operating state, extracting quantitative indicators that reflect the coordination between steps. Finally, the calculated synergistic path coefficients and mutual influence matrices are packaged into standardized synergistic effect data.

[0077] Based on the above technical solution, optionally, dynamic analysis of the coupling lag effect and synergistic relationship between each process step can be performed based on optimized control parameters and historical process information to dynamically determine the synergistic effect data between each process step and the optimized control parameters, including: Based on the periodic evolution patterns and spatiotemporal correlation attributes in historical process information, we continuously extract cross-process time sequence features and dynamically determine the coupling lag factor between each process step. Based on the coupling lag factor, a multi-stage synergistic response correlation analysis is continuously performed on each process stage to dynamically determine the synergistic correlation matrix between each process stage. Based on the collaborative correlation matrix, the cross-process collaborative deviation of the optimized control parameters is continuously evaluated and quantitatively determined, and the collaborative fluctuation characteristics and evolution law data of each process step are dynamically determined. Based on the data of synergistic fluctuation characteristics and evolution laws, we continuously perform in-depth calculations and verifications of synergistic effects, and dynamically determine the synergistic effect data of each process link and optimized control parameters.

[0078] In this scheme, the periodic evolution law is that the changes of various parameters in the process have periodic characteristics, specifically manifested in the regular fluctuations of key parameters such as temperature and pressure in different time periods.

[0079] Spatiotemporal correlation attributes refer to the inherent correlation between process parameters at different time points and spatial locations. This attribute describes the objective laws governing how various parameters in a process interact and change in tandem with the passage of time and spatial transformation. Parameter changes at different production stages exhibit a time lag effect, and various process parameters in the spatial dimension may also form interdependent relationships.

[0080] The coupling lag factor is used to characterize the dynamic coupling relationship between different process steps, while incorporating the time delay or lag effect that exists during parameter changes. Parameter changes in preceding process steps often require a certain delay before having an actual impact on subsequent steps; the coupling lag factor can accurately quantify these lag effects.

[0081] A synergistic correlation matrix is ​​a data structure used to characterize the interactions and synergistic effects between multiple process stages. This matrix can reflect the interaction intensity and synergistic relationship between different process stages within the same time period, while clearly presenting the positive or negative synergistic patterns between each stage.

[0082] Synergistic fluctuation characteristics are specific fluctuation patterns formed under the synergistic effect of multiple process stages. This characteristic reveals the degree of synchronous fluctuation among multiple process stages, and clarifies the propagation path and amplification mechanism of such fluctuations between different stages.

[0083] Evolutionary patterns are the long-term evolutionary trends of process parameters or system characteristics over time. These patterns are used to describe the dynamic changes in system state over time, and can also reveal the gradual accumulation and dynamic evolution mechanism of characteristics such as cooperative fluctuations within the system.

[0084] Based on the cyclical evolution patterns inherent in historical process information, the system first employs Fast Fourier Transform (FFT) to perform frequency domain analysis on long-period data, extracting significant fluctuation periods of key parameters such as temperature and pressure. Combining spatiotemporal correlation attributes, a cross-correlation function is used to slide and compare the characteristic sequences of adjacent process steps along the time axis, calculating the displacement when the correlation coefficient reaches its peak. This step quantitatively measures the average time difference between the response of preceding steps to subsequent steps, accurately pinpointing the dynamic delay between each step, and then dynamically determining the coupling lag factor used to characterize the time delay pattern. Based on the coupling lag factor, the system uses a dynamic time warping algorithm to align the process sequences across steps, eliminating waveform misalignment caused by lag effects. Furthermore, partial correlation analysis is used to eliminate spurious correlation interference from third-party parameters, deeply mining the net correlation strength between heating, reduction, and cooling steps. These time-delay-compensated correlation coefficients are then filled into a multidimensional tensor to dynamically construct a collaborative correlation matrix. This matrix clearly reflects the interaction response density between each process step, explicitly presenting the positive and negative feedback logic of interdependence among nodes in the process flow.

[0085] After obtaining the synergistic correlation matrix, wavelet packet decomposition is used to analyze the operational deviations of each stage at multiple scales, based on the current optimized control parameters, to remove the synergistic oscillation components caused by parameter adjustments. By calculating the signal energy distribution in different frequency bands, the synergistic fluctuation characteristics caused by the synergistic effect are quantified, and the propagation gain of the deviation between each stage is clarified. Simultaneously, the Mann-Kendall test is introduced to perform nonparametric trend analysis on the time-series data, identifying the monotonic change trend of the synergistic effect with production progress, and thus deriving the long-term evolution law data of the process state over time. Based on the synergistic fluctuation characteristics and evolution law data, the system further uses structural equation modeling to conduct deep synergistic calculations on the path contributions between each stage, establishing a causal mapping network between control variables and the overall system objective. Monte Carlo simulation is used to perform large-scale random sampling verification within the parameter range determined by the evolution law data to verify the robustness of the synergistic effect under different operating conditions. Path coefficient reestimation is performed on the simulation results, and finally, the synergistic effect data, including multi-stage linkage gain and interference offsetting coefficients, is dynamically determined.

[0086] In this solution, by analyzing the coupling relationship and synergistic effect between process links, control parameters are dynamically adjusted to optimize the process, thereby achieving precise control and optimization, improving production efficiency and reducing energy consumption.

[0087] Step S106: Based on the synergistic effect data, perform priority adjustment analysis of each process link and optimized control parameters, dynamically generate adaptive control commands, and perform closed-loop control and dynamic correction of the actuator of the reduction equipment based on the adaptive control commands until the titanium iron powder reduction process is completed.

[0088] Adaptive control commands are a dynamically generated set of instructions that combine real-time process status, synergistic effect data, and optimized control objectives. It first analyzes the real-time adjustment needs of each process stage, then flexibly adjusts control parameters to adapt to various changes occurring during the process. These control commands can automatically adjust control strategies based on feedback data and system operating status, ensuring the system always maintains optimal operating levels. It not only optimizes each control parameter but also responds in real-time to process changes and abnormal situations.

[0089] The reduction equipment is the core production equipment in the reduction process of ilmenite powder, mainly composed of a reactor, heating system, and atmosphere control device. Its function is to reduce ilmenite into ilmenite powder under specific temperature, pressure, and atmosphere conditions.

[0090] The actuator is a key component of the restoration equipment, primarily responsible for adjusting the equipment's physical state according to control commands. It comprises a series of mechanical or electrical devices used to regulate temperature, airflow, and pressure, such as electric heaters, valves, pumps, and fans. The actuator receives adaptive control commands from the control system and alters the equipment's physical parameters to meet process requirements.

[0091] In the priority adjustment analysis phase, the system uses multi-criteria decision analysis technology to weight and map the collected synergistic effect data with real-time process deviations. This process incorporates the analytic hierarchy process (AHP) to quantitatively assess the sensitivity of optimized control parameters such as temperature, pressure, and atmosphere flow rate to the current reduction stage, thereby identifying the key links that have the most significant impact on overall reduction efficiency. The system combines the positive and negative feedback patterns accumulated from historical synergistic effects and prioritizes the adjustment order of each process link according to the Pareto optimality principle. For complex operating conditions with multiple parameter conflicts, the system constructs a competitive weight matrix to prioritize the adjustment space of high-sensitivity parameters, determining the optimal adjustment sequence and step size allocation scheme for each process link under the current operating state.

[0092] In the stage of dynamically generating adaptive control commands, the system, based on the aforementioned priority adjustment sequence, integrates fuzzy control algorithms with proportional, integral, and derivative gain adaptive techniques to transform optimized control parameters into low-level control signals with physical dimensions. This process dynamically corrects the proportional and integral terms of the controller by calculating the correlation gain in the synergistic effect data online in real time, thereby compensating for adjustment deviations caused by process lags. The system introduces a heuristic constraint optimization operator to generate an adaptive control command set, including heating system power settings, atmosphere regulating valve opening, and fan speed, while satisfying the rated power and safety boundary constraints of the restoration equipment. These command sets undergo logical fusion and boundary verification to adapt to the dynamic changes in the process while ensuring that the fluctuation range of each executed variable remains within the physical tolerance range of the equipment.

[0093] During the closed-loop control and dynamic correction phase, the generated adaptive control commands are sent in real time to the actuators of the reduction equipment via an industrial fieldbus. Electric heaters, regulating valves, and pumps adjust their voltage frequency or mechanical displacement according to the commands, thereby altering the thermal field distribution and atmosphere flow state within the furnace. During this process, the system uses a state observer to collect real-time displacement feedback data from the actuators and process response data monitored by sensors, and calculates the deviation vector between the actual output and the ideal trajectory using residual analysis. Once a deviation from the expected target is detected, the system immediately applies feedback linearization or sliding mode control techniques to correct the gain of subsequent commands in real time, thereby compensating for the mechanical hysteresis or nonlinear friction effects of the actuators. This dynamic correction process covers all stages of ferrotitanium powder reduction, including heating, preheating, stable reduction, and cooling, forming a continuously iterative closed-loop control chain. As the reduction reaction progresses, the system continuously updates the synergistic effect evaluation results through real-time data streams and dynamically adjusts the response priority of the actuators using a heuristic search algorithm. When the key quality indicators such as atmosphere composition and metallization rate monitored online reach the preset shutdown threshold, the system issues the final instruction set for stable cooling and system reset, completing the end control of the actuator, and the entire titanium iron powder reduction process ends.

[0094] Based on steps S101-S106 above, by dynamically acquiring multi-source process parameters, optimizing the control strategy in real time, and adjusting each process link, the stable and efficient reduction process of ferrotitanium powder is ensured. Adaptive control commands realize closed-loop regulation, improving production accuracy and flexibility, and enhancing product quality and efficiency.

[0095] Based on the above technical solution, optionally, based on the synergistic effect data, priority adjustment analysis of each process step and optimized control parameters is performed to dynamically generate adaptive control commands, including: Based on synergistic effect data, the influence weight of each process step on the optimization control parameters is continuously quantified, and the synergistic correlation strength of each process step is prioritized based on the influence weight, so as to dynamically determine the ranking result of the influence of multiple steps. Based on the ranking results of the multi-stage influence, the control gain and adjustment step size of each process stage under the current operating conditions are dynamically configured to continuously obtain the adaptive adjustment weight matrix of each process stage. Based on the adaptive adjustment weight matrix and optimized control parameters, cross-stage joint compensation calculations for control deviations are continuously performed to dynamically determine the instruction coordination correction values ​​for each process stage. Based on the instruction collaborative correction value and the preset constraints of each process step, instruction fusion and boundary verification are continuously performed to dynamically generate adaptive control instructions for each process step.

[0096] In this scheme, the influence weight is a quantitative indicator used to quantify the degree of influence of each process step on the optimized control parameters, intuitively reflecting the role of each process step in the control process. The larger the weight value, the more significant the influence of that process step on the optimized control parameters.

[0097] Synergy strength refers to the specific degree to which multiple process steps work together to influence the control objective. It quantifies the interaction and coordination between different process steps. The higher the synergy strength, the more prominent the synergistic effect between these process steps, and the more obvious the positive driving effect on the control objective.

[0098] The multi-stage impact ranking result is obtained by ranking the priority of the impact optimization control objectives of each process stage based on the quantification of the impact weight and synergistic correlation strength of each process stage.

[0099] Control gain is the amplification factor of the control signal in a control system. It directly determines the system's response to changes in input and is mainly used to adjust the response speed and amplitude of process parameters to the optimization target.

[0100] The adjustment step size is the specific amount of change made each time the control parameter is adjusted. The size of the adjustment step size directly affects the control effect. The larger the step size, the larger the range of parameter adjustment in a single operation, and the faster the adjustment speed; the smaller the step size, the finer the parameter adjustment and the higher the control precision.

[0101] The adaptive adjustment weight matrix is ​​a data matrix used to represent the adjustment weight and adjustment step size of each process step. It dynamically adjusts according to the real-time process operation, continuously optimizing the entire control process and ensuring that the various process steps can maintain coordination and balance.

[0102] The instruction coordination correction value is a parameter adjustment amount calculated by combining the coordination effects of each process step. Its function is to correct and optimize the control parameters. Through this correction value, it is possible to ensure that the control instructions of each process step achieve the best coordination state.

[0103] Preset constraints are a series of limiting rules set in advance, taking into account the characteristics of the process and the system's operational requirements. Their purpose is to ensure that the operating range and parameter changes during the control process remain within a safe and efficient range. These include restrictions on key parameters such as temperature and pressure.

[0104] Based on the acquired synergistic effect data, the Sobol global sensitivity analysis method is first used to quantitatively analyze the contribution of each process step to the overall control objective. The calculated contribution rate is then converted into the influence weight of each step. Based on this weight, the fuzzy comprehensive evaluation method is used to quantify the coupling response strength between different process steps, clarifying the synergistic correlation strength when each step works together. Finally, the rank-sum ratio method is used to prioritize and weight the influence of each step, dynamically generating a multi-step influence ranking result, clearly defining the adjustment priority of each step, and providing clear logical support for the rational allocation of subsequent adjustment resources. Combining the multi-step influence ranking result, the system uses gain scheduling technology to dynamically configure the control gain of each process step according to the urgency and sensitivity characteristics of the current restoration stage. Simultaneously, considering the fluctuation frequency of the current operating conditions, the adjustment step size is calculated through a nonlinear mapping function, balancing the dual requirements of system response speed and operational stability. These control gain and adjustment step size data are arranged as tensors according to the process dimension and dynamically encapsulated to form an adaptive adjustment weight matrix, including the adjustment rhythm of the entire process. Through differentiated parameter allocation, it is ensured that each process step maintains consistency in the dynamic operating environment, improving control synergy.

[0105] The adaptive adjustment weight matrix and optimized control parameters are multiplied together, and a feedforward compensation algorithm is used to coordinate and align execution deviations across stages on the time axis. Multivariate joint residual analysis is employed to identify command conflicts generated during the adjustment process in different stages, and dynamic deviation weighted compensation is performed based on the priority set by the weight matrix. In this process, real-time calculations can offset the adjustment increments that cancel dynamic mutual interference between stages, thereby dynamically determining the command coordination correction value to align with the global control target, ensuring that adjustments to a single variable do not cause overall system instability and guaranteeing the stability of the control process. The generated command coordination correction value is superimposed on the basic control parameters, while pre-set constraints such as temperature, pressure, and power are retrieved. Logic threshold judgment and saturation limiting techniques are used to perform real-time safety verification of the control commands. An anti-integral saturation operator is used to limit the rate of change and filter extreme values ​​of the corrected data stream, ensuring that the output signal is strictly controlled within the physical boundaries of the equipment. Through this multi-dimensional command fusion and rigorous boundary verification, the system ultimately dynamically generates adaptive control commands, driving the precise actions of the reduction equipment's actuators, achieving closed-loop precise control of the entire titanium-iron powder reduction process.

[0106] In this scheme, by dynamically quantifying and adjusting the influence weights and synergistic relationships of process steps, it is helpful to accurately optimize control parameters, improve the coordination and response efficiency of the process, and thus improve the stability and production efficiency of the titanium iron powder reduction process.

[0107] See appendix Figure 2 , Figure 2This is a schematic flowchart of the second main step of a dynamic coupling control method for the reduction process of ferrotitanium powder according to an embodiment of the present invention. Figure 2 As shown, the dynamic coupling control method for the reduction process of titanium-iron powder in an embodiment of the present invention mainly includes the following steps S201-S209.

[0108] Step S201: Continuously acquire multi-source process parameters during the reduction of ferrotitanium powder, continuously clean and preprocess the multi-source process parameters, and generate a real-time optimized dataset.

[0109] Step S202: Perform dynamic feature extraction based on the real-time optimization dataset to obtain the key features of each process parameter. Based on the selected key features, perform stratification and sorting to generate a real-time structured dataset.

[0110] Step S203: Perform dynamic time series analysis based on the real-time structured dataset, extract the time series characteristics of each process parameter, analyze the dynamic evolution trend of the process parameters based on the time series characteristics, and generate a real-time time series feature dataset.

[0111] Step S204: Perform dynamic multidimensional data fusion based on the real-time time series feature dataset to obtain a dynamic process status dataset.

[0112] Step S205: Based on the preset multidimensional phase transformation determination rules, the dynamic process state dataset is mapped to the preset process state domain, and real-time pattern matching and classification are performed to dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information.

[0113] Step S206: Obtain preset control target data, perform dynamic correlation analysis based on specific operating stages, real-time status information and preset control target data, and dynamically determine the segmented optimization control strategy that matches the specific operating stage.

[0114] Step S207: Obtain and dynamically update the current control parameters, perform real-time multivariate decoupling analysis and parameter optimization based on the piecewise optimization control strategy and the current control parameters, and dynamically determine the optimized control parameters.

[0115] Step S208: Obtain historical process information, and perform dynamic analysis of the coupling lag effect and synergistic relationship between each process step based on the optimized control parameters and historical process information, and dynamically determine the synergistic effect data between each process step and the optimized control parameters.

[0116] Step S209: Based on the synergistic effect data, perform priority adjustment analysis of each process link and optimized control parameters, dynamically generate adaptive control commands, and perform closed-loop control and dynamic correction of the actuator of the reduction equipment based on the adaptive control commands until the titanium iron powder reduction process is completed.

[0117] In this embodiment, the real-time optimization dataset is a collection of data extracted from multi-source process parameters through cleaning and preprocessing. It is mainly used for dynamically adjusting and optimizing control parameters. It includes various key data points in the process and their corresponding optimization requirements, with key data points including temperature, pressure, and flow rate.

[0118] Process parameters are various measurable indicators and environmental conditions that can affect the process progress during production. Common process parameters include temperature, pressure, flow rate, atmosphere composition, and time.

[0119] Key features are representative variables extracted from process parameter data that have a significant impact on the production process. By systematically processing and deeply analyzing process parameters, the indicators that best reflect the real-time status and development trend of the process are selected; these are the key features.

[0120] Real-time structured datasets are collections of data formed based on process parameters, after feature extraction, hierarchical organization, and ordered sorting. This dataset is organized according to a predetermined structure and hierarchy, and it possesses real-time feedback capabilities, mapping the actual state and dynamic trends of the current process.

[0121] Time-series characteristics refer to the changing properties of process parameters over time. These characteristics include various types such as fluctuations, periodic changes, and trends. By mining and analyzing time-series characteristics, we can clearly understand the differences in the state of process parameters at different points in time.

[0122] The real-time time series feature dataset is an integrated collection of time series feature data extracted from process parameters, mainly used to reflect the dynamic changes of process parameters at different time points.

[0123] In the continuous acquisition of multi-source process parameters, the system first uses a Kalman filter algorithm to denoise the raw current and voltage signals transmitted from the underlying sensors in real time, while simultaneously employing median filtering to remove transient outliers caused by electromagnetic interference. After cleaning, the data is then standardized using standard deviation to eliminate numerical differences between different units such as temperature, pressure, and flow rate, ensuring comparability across all process dimensions. Combined with real-time emission reduction and efficiency improvement optimization requirements from production tasks, the system correlates and maps the processed data stream with the required indicators, encapsulating them into a high-fidelity real-time optimization dataset.

[0124] Based on a real-time optimized dataset, the system employs principal component analysis or random forest feature importance assessment techniques to screen out the key features that contribute most to the reduction reaction rate and product quality from a massive number of process variables. After obtaining the key features, semantic tagging technology is used to organize them hierarchically according to the physical entity topology of the reduction furnace, dividing the features into equipment layer, location layer, and measurement point layer. Simultaneously, information entropy theory is used to rank the variables within each layer based on their significance. Through this hierarchical data arrangement and priority definition, the system integrates scattered feature variables into a logically related real-time structured dataset, realizing the digital reconstruction of the physical logic and operating status of the process site.

[0125] Based on the real-time structured dataset, the system applies a sliding window algorithm to slice each process parameter along the time axis, extracting time-series features such as mean, variance, and skewness that reflect data distribution fluctuations in real time. For non-stationary signals such as atmosphere components, discrete wavelet transform is used for multi-scale decomposition to separate low-frequency components representing long-term evolution trends and high-frequency detail components capturing instantaneous disturbances. By performing first-order difference operations on these components, the instantaneous slope and acceleration of parameter changes over time are calculated, accurately quantifying the dynamic evolution of each process indicator and generating a real-time time-series feature dataset that reflects the time evolution attributes of the reduction process.

[0126] Finally, the system utilizes a vector space mapping mechanism to dynamically fuse multidimensional variables in the real-time time-series feature dataset. This process encapsulates temperature gradients, pressure fluctuations, and atmospheric composition features at different spatial measurement points and time scales into a unified high-dimensional feature vector. Simultaneously, it introduces label indexing technology from the time-series database to achieve real-time association between data points and physical topology and production batches. Through the superposition and fusion of multidimensional tensors, the system transforms a single time series into a dynamic process state dataset with self-describing capabilities and process context relationships, achieving a comprehensive and accurate characterization of the titanium-iron powder reduction process from microscopic instantaneous fluctuations to macroscopic stage evolution.

[0127] In this embodiment, by continuously acquiring and analyzing multi-source process parameters, dynamically extracting key features, performing time-series analysis and data fusion, and generating an accurate process status dataset, real-time monitoring and optimized control are achieved, thereby improving the accuracy and efficiency of the production process and ensuring the stability of the titanium iron powder reduction process.

[0128] It should be noted that although the steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effects of the present invention, different steps do not necessarily have to be executed in such an order. They can be executed simultaneously (in parallel) or in other orders, and these variations are all within the scope of protection of the present invention.

[0129] Furthermore, the present invention also provides a dynamic coupling control system for the reduction process of ferrotitanium powder.

[0130] See appendix Figure 3 , Figure 3 This is a main structural block diagram of a dynamic coupling control system for the reduction process of ferrotitanium powder according to an embodiment of the present invention. Figure 3 As shown, it specifically includes: Data acquisition module 301 is used to continuously acquire multi-source process parameters during the reduction process of ferro-titanium powder, and dynamically reconstruct hierarchically and extract time-series features based on the multi-source process parameters to obtain a dynamic process state dataset. The matching and classification module 302 is used to map the dynamic process state dataset to the preset process state domain based on the preset multi-dimensional phase transition determination rules, perform real-time pattern matching and classification, and dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information. The dynamic correlation analysis module 303 is used to acquire preset control target data, perform dynamic correlation analysis based on the specific operating stage, real-time status information and preset control target data, and dynamically determine the segmented optimization control strategy that matches the specific operating stage. Optimization module 304 is used to acquire and dynamically update the current control parameters, perform real-time multivariate decoupling analysis and parameter optimization based on the segmented optimization control strategy and the current control parameters, and dynamically determine the optimized control parameters. The coupling analysis module 305 is used to acquire historical process information, and to perform dynamic analysis of the coupling lag effect and synergistic relationship between each process step based on the optimized control parameters and historical process information, and to dynamically determine the synergistic effect data between each process step and the optimized control parameters. The correction module 306 is used to perform priority adjustment analysis of each process link and optimized control parameters based on the synergistic effect data, dynamically generate adaptive control commands, and perform closed-loop control and dynamic correction of the actuator of the reduction equipment based on the adaptive control commands until the titanium iron powder reduction process is completed.

[0131] The dynamic coupling control system for the reduction process of ferrotitanium powder provided in this application embodiment can achieve... Figure 1 The various processes implemented in the method implementation examples will not be described again here to avoid repetition.

[0132] Those skilled in the art will understand that all or part of the processes in the method of the above embodiment of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium can include any entity or device capable of carrying the computer program code, a medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content included in the computer-readable storage medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0133] Furthermore, the present invention also provides an electronic device 400, including a processor 401, a memory 402, and a program or instructions stored in the memory 402 and executable on the processor 401. When the program or instructions are executed by the processor 401, they implement the various processes of the above-described embodiment of the dynamic coupling control method for the reduction process of titanium-iron powder, and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0134] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0135] Furthermore, the present invention also provides a computer-readable storage medium. In one embodiment of the computer-readable storage medium according to the present invention, the computer-readable storage medium can be configured to store a program for executing a dynamic coupling control method for a titanium-iron powder reduction process according to the above-described method embodiments. This program can be loaded and run by a processor to implement the above-described dynamic coupling control method for a titanium-iron powder reduction process. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The computer-readable storage medium can be a storage device comprising various electronic devices. Optionally, in the embodiments of the present invention, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0136] Furthermore, it should be understood that since the various modules are only provided to illustrate the functional units of the device of the present invention, the physical devices corresponding to these modules may be the processor itself, or a part of the processor's software, hardware, or a combination of software and hardware. Therefore, the number of modules shown in the figures is merely illustrative.

[0137] Those skilled in the art will understand that the various modules in the device can be adaptively split or combined. Such splitting or combining of specific modules will not cause the technical solution to deviate from the principles of the present invention; therefore, the technical solutions after splitting or combining will fall within the protection scope of the present invention.

[0138] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A dynamic coupling control method for the reduction process of ferrotitanium powder, characterized in that, The method includes: Continuously acquire multi-source process parameters during the reduction of ferro-titanium powder, and dynamically reconstruct hierarchically and extract time-series features based on the multi-source process parameters to obtain a dynamic process state dataset; Based on the preset multidimensional phase transformation determination rules, the dynamic process state dataset is mapped to the preset process state domain, and real-time pattern matching and classification are performed to dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information. Acquire preset control target data, perform dynamic correlation analysis based on specific operating stages, real-time status information and preset control target data, and dynamically determine segmented optimization control strategies that match specific operating stages; Acquire and dynamically update the current control parameters, perform real-time multivariate decoupling analysis and parameter optimization based on the segmented optimization control strategy and the current control parameters, and dynamically determine the optimized control parameters. Historical process information is acquired, and dynamic analysis of coupling lag effects and synergistic relationships between various process steps is conducted based on optimized control parameters and historical process information. The synergistic effect data between each process step and optimized control parameters is dynamically determined. Based on the synergistic effect data, priority adjustment analysis of each process step and optimized control parameter is performed, adaptive control commands are dynamically generated, and the actuator of the reduction equipment is controlled and dynamically corrected in a closed loop based on the adaptive control commands until the titanium iron powder reduction process is completed.

2. The dynamic coupling control method for the reduction process of ferrotitanium powder according to claim 1, characterized in that, in, Based on the dynamic hierarchical reconstruction of multi-source process parameters and the extraction of time-series features, a dynamic process state dataset is obtained, including: Continuously clean and preprocess multi-source process parameters to generate real-time optimization datasets; Dynamic feature extraction is performed based on the real-time optimization dataset to obtain key features of each process parameter. Based on the selected key features, the dataset is hierarchically and sorted to generate a real-time structured dataset. Dynamic time series analysis is performed based on real-time structured datasets to extract time series features of each process parameter. Based on the time series features, the dynamic evolution trend of the process parameters is analyzed to generate a real-time time series feature dataset. Dynamic process status datasets are obtained by performing dynamic multidimensional data fusion based on real-time time-series feature datasets.

3. The dynamic coupling control method for the reduction process of ferrotitanium powder according to claim 1, characterized in that, in, Based on preset multidimensional phase transition determination rules, the dynamic process state dataset is mapped to a preset process state domain, and real-time pattern matching and classification are performed to dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information, including: Based on the preset multidimensional phase transition determination rules, the system continuously performs real-time feature extraction and pattern recognition on dynamic process state datasets to dynamically determine the key determination conditions for process state changes. Based on key judgment conditions, the dynamic process state dataset is mapped to the preset process state domain in real time, and the process state is classified to obtain the preliminary matching label for process state matching. Based on the initial matching tags, time-series consistency comparison and rule reasoning are performed to dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information.

4. The dynamic coupling control method for the reduction process of ferrotitanium powder according to claim 1, characterized in that, in, Based on dynamic correlation analysis of specific operational phases, real-time status information, and preset control target data, a segmented optimization control strategy matching the specific operational phase is dynamically determined, including: Dynamic process parameters are extracted based on specific operational stages and real-time status information to continuously determine the real-time process parameter set. Dynamic dependency analysis is performed based on real-time process parameter sets and preset control target data to evaluate the dependency relationship between each real-time process parameter and the preset control target data in real time. Real-time sensitivity analysis is performed based on the dependency data to evaluate the impact of each real-time process parameter on the preset control target data. The data on the impact of each real-time process parameter on the preset control target data are sorted in real time, and key process parameters are dynamically filtered out based on preset screening thresholds to obtain a real-time updated list of key process parameters. Based on the list of key process parameters and the target parameter requirements data corresponding to specific operating stages, dynamic target value optimization is performed to dynamically determine a segmented optimization control strategy that matches the specific operating stage.

5. The dynamic coupling control method for the reduction process of ferrotitanium powder according to claim 4, characterized in that, in, Based on a piecewise optimization control strategy and current control parameters, real-time multivariate decoupling analysis and parameter optimization are performed to dynamically determine the optimal control parameters, including: Based on the segmented optimization control strategy and the current control parameters, the coupling relationship analysis of key process parameters is continuously carried out to dynamically determine the coupling correlation matrix; The adaptive range of the current control parameters is dynamically derived based on the coupling correlation matrix, resulting in the real-time adaptive range of the current control parameters and the real-time constraint conditions. Dynamic adaptive adjustment is performed based on the real-time adaptive range and real-time constraints to obtain the real-time adjustment range of the current control parameters; Based on the real-time adjustment range, the current control parameters are adjusted in real time to dynamically determine the optimal control parameters.

6. The dynamic coupling control method for the reduction process of ferrotitanium powder according to claim 1, characterized in that, in, Dynamic analysis of coupling lag effects and synergistic relationships among various process steps is conducted based on optimized control parameters and historical process information. This dynamically determines the synergistic effect data between each process step and the optimized control parameters, including: Based on the periodic evolution patterns and spatiotemporal correlation attributes in historical process information, we continuously extract cross-process time sequence features and dynamically determine the coupling lag factor between each process step. Based on the coupling lag factor, a multi-stage synergistic response correlation analysis is continuously performed on each process stage to dynamically determine the synergistic correlation matrix between each process stage. Based on the collaborative correlation matrix, the cross-process collaborative deviation of the optimized control parameters is continuously evaluated quantitatively, and the collaborative fluctuation characteristics and evolution law data of each process are dynamically determined. Based on the data of synergistic fluctuation characteristics and evolution laws, we continuously perform in-depth calculations and verifications of synergistic effects, and dynamically determine the synergistic effect data of each process link and optimized control parameters.

7. The dynamic coupling control method for the reduction process of ferrotitanium powder according to claim 1, characterized in that, in, Based on the aforementioned synergistic effect data, priority adjustment analysis of each process step and optimized control parameter is performed, and adaptive control commands are dynamically generated, including: Based on synergistic effect data, the influence weight of each process step on the optimization control parameters is continuously quantified, and the synergistic correlation strength of each process step is prioritized based on the influence weight, so as to dynamically determine the ranking result of the influence of multiple steps. Based on the ranking results of the multi-stage influence, the control gain and adjustment step size of each process stage under the current operating conditions are dynamically configured to continuously obtain the adaptive adjustment weight matrix of each process stage. Based on the adaptive adjustment weight matrix and optimized control parameters, cross-stage joint compensation calculations for control deviations are continuously performed to dynamically determine the instruction coordination correction values ​​for each process stage. Based on the instruction collaborative correction value and the preset constraints of each process step, instruction fusion and boundary verification are continuously performed to dynamically generate adaptive control instructions for each process step.

8. A dynamic coupling control system for the reduction process of ferrotitanium powder, characterized in that, The system includes: The data acquisition module is used to continuously acquire multi-source process parameters during the reduction of ferro-titanium powder, and to dynamically reconstruct the hierarchical structure and extract time-series features based on the multi-source process parameters to obtain a dynamic process state dataset. The matching and classification module is used to map the dynamic process state dataset to the preset process state domain based on the preset multi-dimensional phase transition determination rules, perform real-time pattern matching and classification, and dynamically determine the specific operating stage of the current reduction process of ferrotitanium powder and the corresponding real-time status information. The dynamic correlation analysis module is used to acquire preset control target data, perform dynamic correlation analysis based on specific operating stages, real-time status information and preset control target data, and dynamically determine the segmented optimization control strategy that matches the specific operating stage. The optimization module is used to acquire and dynamically update the current control parameters. Based on the segmented optimization control strategy and the current control parameters, it performs real-time multivariate decoupling analysis and parameter optimization to dynamically determine the optimized control parameters. The coupling analysis module is used to acquire historical process information, and to perform dynamic analysis of the coupling lag effect and synergistic relationship between various process links based on the optimized control parameters and historical process information, so as to dynamically determine the synergistic effect data between each process link and the optimized control parameters. The correction module is used to perform priority adjustment analysis of each process link and optimized control parameters based on the synergistic effect data, dynamically generate adaptive control commands, and perform closed-loop control and dynamic correction of the actuator of the reduction equipment based on the adaptive control commands until the titanium iron powder reduction process is completed.

9. An electronic device comprising a processor, a memory, and a program or instructions stored in the memory and executable on the processor, characterized in that, The program or instructions are adapted to be loaded and run by the processor to perform a dynamic coupling control method for the reduction process of ferro-titanium powder according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by a processor to perform a dynamic coupling control method for the reduction process of ferro-titanium powder according to any one of claims 1 to 7.

Images