Multi-parameter collaborative control method and system for chemical product production workshop

Abstract

The application provides a multi-parameter collaborative control method and system for a chemical product production workshop, and relates to the technical field of intelligent control. The method comprises the following steps: constructing a multi-parameter sparse coupling topology according to the dynamic mutual information between parameters; performing a topology path sensitivity evaluation, identifying the key coupling chain that has the greatest impact on reaction stability, and tracking the collaborative change trend of the parameters in the chain in real time; establishing a parameter prediction model, and performing automatic rearrangement of the adjustment priority of the parameters in the chain based on a constraint migration mechanism; constructing a self-evolution collaborative control strategy generator, generating collaborative control instructions for the reaction process through online evolution; and issuing the instructions to corresponding heating equipment, cooling equipment, pressure relief devices and feeding equipment. The application solves the technical problem that the prior art usually adjusts a single parameter and lacks in-depth analysis of the interaction between multiple parameters, which leads to the system being unable to maintain stability when multiple parameters fluctuate, and further leads to equipment failure or safety accidents.

Description

Multi-parameter collaborative control method and system for chemical product production workshop Technical Field

[0001] This invention relates to the field of intelligent control technology, specifically to a multi-parameter collaborative control method and system for chemical product production workshops. Background Technology

[0002] Chemical product production workshops typically involve multiple process parameters that interact with each other during production. Fluctuations in any one parameter can lead to a decline in the performance of the entire reaction process or even safety hazards. Therefore, achieving efficient and coordinated control of these multiple parameters is a key technological challenge in modern chemical production.

[0003] Traditional control methods typically adjust a single parameter, lacking in-depth analysis of the interactions between multiple parameters. Because these methods cannot integrate the coupling and interaction effects between parameters, they cannot effectively cope with situations where multiple parameters change simultaneously. This leads to slow system response and even an inability to maintain stability when multiple parameters fluctuate. Especially in dynamically changing production environments, the inability to adjust control strategies in real time can result in unstable reaction processes, unstable output, and even equipment failures or safety accidents. Summary of the Invention

[0004] This application provides a multi-parameter collaborative control method and system for chemical product production workshops, aiming to solve the technical problem that existing technologies usually adjust a single parameter and lack in-depth analysis of the interaction between multiple parameters, resulting in the system's inability to remain stable when multiple parameters fluctuate, which in turn leads to equipment failure or safety accidents.

[0005] The first aspect disclosed in this application provides a multi-parameter collaborative control method for a chemical product production workshop. The method includes: collecting temperature, pressure, concentration, and flow parameters through a distributed sensor network; constructing a multi-parameter sparse coupling topology based on the dynamic mutual information between parameters, wherein each node in the multi-parameter sparse coupling topology represents a single process parameter, and the edge weights represent the time-varying coupling strength between parameters; performing a topology path sensitivity assessment of the multi-parameter sparse coupling topology to identify the key coupling chain that has the greatest impact on reaction stability, and tracking the collaborative change trend of parameters on the key coupling chain in real time; establishing a parameter prediction model based on the key coupling chain and the real-time tracking results, and automatically rearranging the adjustment priorities of parameters within the chain based on a constraint migration mechanism; constructing a self-evolving collaborative control strategy generator based on the migrated dynamic constraints, using the prediction structure of the parameter prediction model and the rearranged adjustment priorities as input, and generating collaborative control instructions for the reaction process through online evolution; and issuing the collaborative control instructions to the corresponding heating equipment, cooling equipment, pressure relief device, and feeding equipment.

[0006] The second aspect of this application discloses a multi-parameter collaborative control system for a chemical product production workshop. This system is used in the aforementioned multi-parameter collaborative control method for chemical product production workshops. The system includes: a topology construction module, used to collect temperature, pressure, concentration, and flow parameters through a distributed sensor network, and construct a multi-parameter sparsely coupled topology based on the dynamic mutual information between parameters. In the multi-parameter sparsely coupled topology, each node represents a single process parameter, and the edge weights represent the time-varying coupling strength between parameters; and a trend tracking module, used to perform topology path sensitivity assessment of the multi-parameter sparsely coupled topology and identify the key parameters that have the greatest impact on reaction stability. The system includes a coupling chain and a real-time tracking module for the coordinated change trend of parameters on the key coupling chain; a priority reordering module for establishing a parameter prediction model based on the key coupling chain and real-time tracking results, and automatically reordering the adjustment priorities of parameters within the chain based on a constraint migration mechanism; an instruction generation module for constructing a self-evolving coordinated control strategy generator based on the migrated dynamic constraints, using the prediction structure of the parameter prediction model and the rearranged adjustment priorities as input, and generating coordinated control instructions for the reaction process through online evolution; and an instruction issuing module for issuing the coordinated control instructions to the corresponding heating equipment, cooling equipment, pressure relief device, and feeding equipment.

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

[0008] Real-time monitoring of key process parameters via distributed sensor networks ensures data timeliness and accuracy. By calculating the time-varying coupling strength between parameters based on dynamic mutual information, nonlinear and time-varying relationships between parameters can be identified and modeled. A sparsity strategy is used to remove redundant relationships, retaining only the coupling chains with the greatest impact on system stability, thereby improving the system's computational efficiency and controllability. Path sensitivity assessment of the topology identifies which parameter coupling relationships have a significant impact on the stability of the reaction process, especially in complex systems with multiple parameter interactions. This analysis effectively focuses on key parameters. After identifying key coupling chains, the coordinated change trends of parameters in these chains are tracked in real time, allowing for timely detection of anomalies or deviations, thus providing reliable input for subsequent control. Combined with real-time tracking... By establishing a dynamic parameter prediction model for key coupling chains, the changing trends of various parameters can be predicted more accurately, providing a basis for subsequent adjustment decisions. Based on the constraint migration mechanism, the adjustment priority of parameters within the chain is dynamically adjusted, ensuring that key parameters affecting reaction stability can be preferentially adjusted and intervened, thus improving the stability and safety of the process. By establishing a self-evolving collaborative control strategy generator, the control strategy is automatically adjusted using an online evolution mechanism. This strategy can be flexibly adjusted according to the real-time changes in the reaction process, ensuring optimal control of the reaction process. By precisely issuing collaborative control commands to each device, each device is ensured to be finely controlled as required, ensuring the stable operation of the reaction process in the entire chemical production workshop, optimizing energy and resource utilization in the production process, and improving overall production efficiency.

[0009] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0010] Figure 1 is a schematic diagram of the multi-parameter collaborative control method for a chemical product production workshop provided in an embodiment of this application.

[0011] Figure 2 is a schematic diagram of the structure of the multi-parameter collaborative control system for a chemical product production workshop provided in the embodiments of this application.

[0012] Figure labeling: Topology building module 10, trend tracking module 20, priority rearrangement module 30, instruction generation module 40, instruction issuance module 50. Detailed Implementation

[0013] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0014] Example 1, as shown in Figure 1, provides a multi-parameter collaborative control method for a chemical product production workshop. The method includes:

[0015] Temperature, pressure, concentration, and flow parameters are collected through a distributed sensor network. A multi-parameter sparse coupling topology is constructed based on the dynamic mutual information between the parameters. In the multi-parameter sparse coupling topology, each node represents a single process parameter, and the edge weight represents the time-varying coupling strength between the parameters.

[0016] Real-time data is collected by deploying a distributed sensor network in the chemical product production workshop. The sensors primarily monitor key process parameters such as temperature, pressure, concentration, and flow rate. Mutual information is used to measure the correlation and dependency between two parameters. In this step, the dynamic mutual information between these parameters is calculated based on their time-series data. Dynamic mutual information uses a sliding time window to estimate the local joint probability distribution of the parameters' time-series data, thereby capturing the nonlinear and time-varying coupling relationships between the parameters. A preliminary weighted graph is constructed using the calculated dynamic mutual information. Each node in the graph represents a process parameter, and the weight of each edge represents the coupling strength between two parameters. By removing connections with lower edge weights and retaining important coupling relationships, a multi-parameter sparse coupled topology is generated. Specifically, dynamic threshold pruning analysis is used to delete edges with weak coupling relationships, retaining only the top-ranked edges.

[0017] Perform topology path sensitivity assessment on multi-parameter sparse coupled topologies, identify the key coupling chains that have the greatest impact on reaction stability, and track the collaborative change trend of on-chain parameters of key coupling chains in real time.

[0018] Based on a multi-parameter sparse coupled topology, a dynamic coupling risk map is constructed by combining historical node perturbation responses, dynamic mutual information, and chain perturbation propagation. This map represents the coupling risk between different process parameters and indicates which coupling paths have the greatest impact on the stability of the reaction process. For each node, the sensitivity integral of all reachable paths originating from that node is calculated, i.e., the contribution of each path's response to system stability is calculated. Specifically, the path sensitivity integral of each node is calculated, and the path sensitivity integral is superimposed with the chain perturbation propagation to form a node contribution matrix to chain stability. This matrix reflects the importance of each node to system stability. Based on the node contribution matrix to chain stability, critical coupling chains are identified. Critical coupling chains are those node combinations with the largest perturbation propagation and second-order and multi-order coupling responses between nodes in the chain exceeding a set threshold. To ensure system stability, critical coupling chains are tracked in real time. Specifically, a joint deviation vector is constructed to quantify the deviation trends of all nodes in the chain. Combining this deviation information, an intra-chain collaborative deviation index is generated, which is used to monitor the collaborative change trends between parameters in real time.

[0019] Based on the key coupling chain and real-time tracking results, a parameter prediction model is established, and the adjustment priority of parameters within the chain is automatically rearranged based on the constraint migration mechanism.

[0020] Based on key coupling chains, a parameter prediction model is constructed using historical data, real-time tracking results, and dynamic coupling coefficients of parameters within the chain. This model can predict the changing trends of parameters within the chain. The establishment of the parameter prediction model relies on the historical time-series data of each parameter and the coupling relationships between them, improving the accuracy of predictions through a data-driven approach.

[0021] Based on the results of the parameter prediction model and combined with the constraint migration mechanism, the regulation priorities are automatically rearranged. The constraint migration mechanism dynamically adjusts the regulation priorities according to the impact of each parameter on the reaction stability. The steps of automatic regulation priority rearrangement include: for each parameter in the key coupling chain, determining their regulation priority based on the parameter prediction model and historical data; when some parameters become abnormal, the constraint migration mechanism is triggered to adjust their regulation priority within the chain, ensuring that important parameters are given priority for regulation, thereby maintaining the stability of the reaction process.

[0022] Based on the dynamic constraints after migration, a self-evolving collaborative control strategy generator is constructed. The predicted structure of the parameter prediction model and the rearranged adjustment priority are used as inputs to generate collaborative control instructions for the reaction process through online evolution.

[0023] The prediction structure of the parameter prediction model, namely the prediction relationship and dependency between each parameter, as well as the rearranged adjustment priority, are taken as input and passed to the control strategy generator. This information serves as the basis to ensure that the generated control strategy can reflect the most important control objectives and be optimized according to real-time changes.

[0024] The self-evolving cooperative control strategy generator generates control strategies through online evolution. Specifically, it fuses and encodes the predicted structure and rearranged regulation priorities into a specific control state gene, describing the interdependencies between parameters, the order of regulation, and how to respond to different disturbances. Using real-time monitoring data, it assesses the sensitivity of each parameter to disturbances throughout the entire reaction process. Specifically, by analyzing deviations within the current chain, it calculates disturbance sensitivity coefficients, which reflect the degree of response of each parameter within the chain to changes in disturbances. Based on these coefficients, directional mutations are triggered, adjusting certain parts of the strategy gene to better adapt to the current operating environment and objectives. The evolutionary process evaluates the effectiveness of these mutations through cross-scale assessments. Through continuous evolution, it ultimately generates cooperative control instructions for the reaction process—a set of precise operational instructions—instructing how to regulate each device to achieve the desired process control objectives. Self-evolution refers to the control system's ability to automatically adjust its control strategy based on real-time feedback data. This mechanism enables the system to continuously optimize and adapt to new operating states in the face of environmental changes or system disturbances.

[0025] The coordinated control command is sent to the corresponding heating equipment, cooling equipment, pressure relief device and feeding equipment.

[0026] The coordinated control commands are mapped and verified against the actual controlled equipment. A mapping verification window is established to ensure that the generated coordinated control commands accurately correspond to the actual equipment operations. Within the mapping verification window, each executing device, including heating equipment, cooling equipment, pressure relief devices, and feeding equipment, is evaluated to verify whether the equipment can operate according to the commands. Based on the evaluation feedback, deviation correction management is performed on the control parameters of the coordinated control commands. If the equipment execution result does not match the expectation, the relevant parameters in the coordinated control commands are automatically adjusted to reduce deviations and optimize control effects. The correction process is dynamic, meaning that whenever a deviation occurs in the equipment response, measures are immediately taken to adjust the command parameters to ensure that the entire production process is always maintained under optimal control.

[0027] Furthermore, based on the aforementioned key coupling chain, a parameter prediction model is established, and the adjustment priority of parameters within the chain is automatically rearranged based on a constraint migration mechanism, including:

[0028] For each parameter in the key coupling chain, a joint parameter prediction model is established using the parameter's historical time-series data, real-time tracking results, and dynamic coupling coefficients of the parameters within the chain. The time-series fitting prediction results of the parameter prediction model are obtained, and an overshoot analysis of the time-series fitting prediction results is performed using the target threshold range. When the target threshold range is exceeded at any time, a constraint migration mechanism is triggered, and the overshoot of the corresponding abnormal parameter is dynamically configured to the parameters within the chain, thereby completing the automatic rearrangement of adjustment priorities.

[0029] Modeling is performed for each parameter in the key coupled chain, including temperature, pressure, concentration, and flow rate. A joint prediction model for each parameter is built using historical time-series data, real-time tracking results, and dynamic coupling coefficients within the chain. Historical time-series data contains the trajectory of parameter changes over time, revealing long-term evolution trends. Real-time tracking results capture the current state of each parameter, providing immediate feedback on system status. Dynamic coupling coefficients within the chain describe the interrelationships and influence levels among parameters; different coefficients help assess the impact of each parameter on the entire reaction chain. The joint parameter prediction model integrates multiple input data sources to infer future parameter behavior. This model employs machine learning methods such as multiple regression, neural networks, and support vector machines, or time-series analysis methods based on dynamic systems theory. These methods can identify patterns in the complex relationships between multidimensional parameters and make accurate predictions.

[0030] By establishing a parameter prediction model, time-series fitting prediction results for each parameter are obtained. These results are estimates of parameter values ​​over a future period and are represented as time-series curves. Each parameter has a target threshold range, which represents the normal range that the parameter should maintain. The target threshold range is determined based on factors such as process specifications, equipment capacity, and reaction requirements. An exceedance analysis is performed on the time-series fitting prediction results based on the target threshold range. Specifically, it checks whether the time-series fitting prediction results exceed the set target threshold range. If the prediction results exceed the upper or lower limits of the target threshold, it indicates that the parameter is abnormal and its adjustment strategy needs to be adjusted.

[0031] When the prediction result exceeds the target threshold range at any given time, the constraint migration mechanism is automatically triggered. The goal is to dynamically adjust the parameter adjustment priority and ensure that critical parameters are adjusted first, preventing excessive impact on system stability. The constraint migration mechanism operates on a priority-based adjustment principle; that is, when some parameters are abnormal, the adjustment task of these abnormal parameters is transferred to higher-priority parameters, ensuring that these anomalies do not cause more serious problems.

[0032] Once the constraint migration mechanism is triggered, the excess values ​​of the out-of-limit parameters are dynamically configured. Specifically, this means identifying parameters that exceed their thresholds and reallocating their adjustment values ​​to other related parameters. This configuration process is achieved by adjusting control strategies, such as enhancing cooling, adjusting feed flow, or changing pressure, to help the out-of-limit parameters return to the normal range.

[0033] Automatic prioritization means that parameters most closely related to response stability and potentially causing greater problems will be given higher priority for adjustment. The prioritization process is dynamic and continuously updates priorities based on real-time data and the output of the predictive model to ensure optimal adjustment strategy.

[0034] Furthermore, the constraint migration mechanism includes:

[0035] A three-dimensional regulation feature for in-chain parameter adjustment is constructed, comprising the prediction bias feature of the parameter itself, the coupling strength feature of the parameter within the chain, and the sensitivity feature of the parameter to in-chain stability. The sensitivity feature is constructed based on the sensitivity matrix of the parameter prediction model. Parameter response analysis is performed based on historical in-chain regulation data and coupling relationships to calculate the in-chain parameter response propagation amount, which is used to quantify the diffusion effect of single-parameter perturbations within the chain. The regulation priority and magnitude of the parameters are determined using the three-dimensional regulation feature and the parameter response propagation amount.

[0036] The three-dimensional regulation characteristics are used to quantify and analyze the regulation properties of various parameters within the chain. These characteristics include three main dimensions: First, the parameter's prediction bias characteristic refers to the deviation between the predicted and actual values ​​of each parameter. These deviations reflect the accuracy of the parameter predictions, and parameters with larger deviations require higher priority regulation. Second, the parameter's coupling strength characteristic describes the strength of the coupling relationship between the parameter and other parameters. By analyzing the dynamic mutual information or coupling coefficients between parameters, the importance of each parameter in the entire reaction chain can be determined. Third, the parameter's sensitivity to chain stability is constructed based on the sensitivity matrix of each parameter. The sensitivity matrix describes the degree to which parameter changes affect reaction stability. Each parameter may have different effects on the stability of other parameters within the chain or the entire reaction system. By analyzing the sensitivity matrix, it is possible to identify which parameters contribute significantly to system stability and which parameters require priority regulation.

[0037] Historical regulation data within the chain is used to analyze the transitivity of parameter responses. This data includes previous regulation behaviors and results, indicating which parameter changes have significantly affected other parameters. Coupling analysis can reveal the interactions and transitivity between different parameters during past regulation processes; for example, whether regulation of one parameter causes significant fluctuations in other parameters within the chain.

[0038] Intra-chain parameter response transitivity is a quantitative indicator used to describe the effect of a disturbance in a single parameter propagating throughout the chain. Specifically, it measures the degree of influence of a parameter change (such as a control operation) on other parameters, reflecting the diffusion effect of the disturbance within the chain. Calculation methods can be based on transfer functions or coupling matrices, deriving the parameter response transitivity by analyzing the impact of each parameter's disturbance on other parameters within the chain. A larger parameter response transitivity indicates a greater impact of that parameter on the stability of the overall system and should be given priority during control.

[0039] By combining three-dimensional regulation characteristics with parameter response transitivity, the regulation priority and magnitude of each parameter are determined. Parameters with a greater impact on chain stability and higher sensitivity are given higher regulation priority. Simultaneously, parameters with strong coupling have a broad impact on the stability of the entire reaction process and should also be prioritized for regulation. Regulation magnitude refers to the degree to which each parameter needs to be adjusted. The determination of the regulation magnitude is based on two factors: the greater the prediction deviation of the parameter, the greater the regulation magnitude; and the response transitivity. If a disturbance in one parameter has a significant impact on other parts of the system, then even if its prediction deviation is not large, the regulation magnitude should be appropriately increased to reduce the effect of disturbance propagation.

[0040] Furthermore, a multi-parameter sparsely coupled topology is constructed based on the dynamic mutual information between parameters, including:

[0041] Calculate the dynamic mutual information between any two parameters. This dynamic mutual information is calculated based on a sliding time window to estimate the local joint probability distribution of the parameter time-series data, capturing the nonlinear and time-varying coupling relationships between the parameters. Construct an initial fully connected weighted graph using this dynamic mutual information. Each node in the initial fully connected weighted graph represents a single parameter, and the edge weights are assigned values ​​based on the dynamic mutual information between the corresponding parameters. Perform sparsity processing on the initial fully connected weighted graph to generate a multi-parameter sparsely coupled topology. This sparsity processing includes: retaining edges with a preset number of mutual information rankings for each node, deleting the remaining edges, and performing dynamic threshold pruning analysis on the edges after configuring a dynamic threshold. Perform topology consistency evaluation on the multi-parameter sparsely coupled topology, and update the coupling relationships of the multi-parameter sparsely coupled topology based on node degree distribution and intra-chain path strength analysis.

[0042] Dynamic mutual information is a measure of the nonlinear coupling and time-varying relationship between two time series. It reflects the interdependence between parameters over time, rather than just a linear relationship. Mutual information measures the degree of information sharing, that is, the extent to which the change in one parameter can predict the change in another parameter. In chemical production processes, the relationship between parameters is nonlinear and time-varying.

[0043] Sliding time windows are used to handle time-varying characteristics in time-series data. Within each time window, the joint probability distribution of time-series data for two selected parameters (e.g., temperature, pressure, etc.) is estimated. This joint probability distribution describes the likelihood that the two parameters will simultaneously exhibit certain states within a given time range. Based on this distribution, mutual information is estimated, which quantifies the information sharing between the two parameters. The method for calculating mutual information is based on Shannon entropy (i.e., information content metric).

[0044] Traditional linear correlation analysis, such as the Pearson correlation coefficient, can only capture the linear dependence between parameters, while dynamic mutual information can identify and quantify nonlinear and time-varying coupling relationships. Nonlinear coupling relationships refer to situations where the relationship between temperature and pressure is linear under certain conditions but exhibits a complex nonlinear relationship under others; dynamic mutual information can reveal these hidden connections. Time-varying coupling relationships refer to situations where the coupling relationship between parameters changes as the production process progresses; dynamic mutual information can track these changes in real time, reflecting the dynamic changes in the relationship between parameters.

[0045] Using the calculated dynamic mutual information, an initial fully connected weighted graph is constructed. In this graph, each node represents a single process parameter, such as temperature, pressure, or concentration. Each edge connects two parameter nodes, and the weight of the edge represents the coupling strength between the two parameters, specifically their dynamic mutual information. Fully connected means that there is an edge between every pair of parameters. This initial fully connected weighted graph is a fully connected network where each node has direct connections with every other node. In practical applications, the initial fully connected weighted graph is a dense graph because many parameters interact with each other in most complex chemical processes.

[0046] The sparsity strategy simplifies the initial fully connected weighted graph into a sparse graph by retaining the edges with the strongest coupling relationships and removing those with less influence or weaker relationships. This reduces computational complexity and avoids excessive parameter interference by removing redundant edges. Edge ranking is based on mutual information. The top N edges are retained according to the mutual information between all adjacent nodes of each node. The preset number of edges N is set according to actual needs; a larger N results in more edges being retained and a lower degree of sparsity, while a smaller N results in fewer edges being retained and a higher degree of graph sparsity.

[0047] The dynamic threshold is adjusted based on the state changes of each node, the coupling strength between parameters, and the real-time performance of the system. For example, the coupling relationships between some parameters change over time or become more important under different operating conditions. If the mutual information between two nodes falls below this dynamic threshold, the edge between them is deleted. In this way, redundant edges in the graph are removed, thereby improving the sparsity of the graph and retaining only the most critical coupling relationships. After sparsification, a multi-parameter sparse coupled topology is obtained. This topology graph only contains the coupling relationships between parameters that have a significant impact on the control and stability of the reaction process. This helps reduce unnecessary computational burden and allows the system to focus on critical coupling relationships.

[0048] The focus of topology consistency assessment is on node degree distribution and intra-chain path strength. Node degree distribution refers to the number of edges connected to each node, indicating the importance of a parameter. If a node has a high degree, it means that it has a strong coupling relationship with multiple other parameters and may be a key parameter in the reaction process. Intra-chain path strength refers to the coupling strength contained in a path in a sparse coupled topology. Strong paths mean that these parameters have a strong coupling relationship, so their coordinated adjustment has a significant impact on system stability.

[0049] Based on the results of the topology consistency assessment, the coupling relationships in the sparse coupled topology are updated to maintain the consistency between the graph structure and the actual coupling relationships in the reaction process, and to ensure that the coupling relationships of key parameters are preserved. This process involves adding or adjusting certain edges, especially when certain parameters become more important in actual operation. In such cases, new edges are added to the topology, or the weights of certain edges are increased. The final multi-parameter sparse coupled topology has an efficient and accurate representation of coupling relationships, which can be applied to the optimization of collaborative control in actual production processes.

[0050] Furthermore, a topology path sensitivity assessment of a multi-parameter sparsely coupled topology is performed to identify the key coupling chains that have the greatest impact on reaction stability, including:

[0051] For the multi-parameter sparse coupled topology, a dynamic coupling risk graph is constructed based on historical node perturbation responses, dynamic mutual information, and chain perturbation propagation. For the dynamic coupling risk graph, a multi-scale path sensitivity assessment is performed, which includes: a. calculating the sensitivity integral of all reachable paths originating from each node; b. superimposing the path sensitivity integral with the intra-chain perturbation propagation to form a node contribution matrix to chain stability. Based on the node contribution matrix to chain stability, key coupling chains are identified. These key coupling chains are combinations of nodes, and the total intra-chain perturbation propagation of the key coupling chain is the largest, with the second-order and multi-order coupling responses of nodes on the chain to other nodes exceeding a set threshold.

[0052] Historical node perturbation response refers to how each parameter responds when it is perturbed in history. This helps to identify which nodes react more strongly to perturbations, meaning they are potential risk points. Dynamic mutual information reflects the time-varying coupling relationship between parameters. Strongly coupled parameters have a greater mutual influence, generating more coupling risks. Chain perturbation propagation describes the propagation effect of parameter perturbations in the entire chain. Perturbations of certain parameters can cause chain reactions in the system.

[0053] A dynamic coupling risk graph is a graph structure designed to quantify the coupling risk between each node and its neighboring nodes. Each node in the graph represents a process parameter, and edges connect the nodes. The edge weights reflect the coupling strength and the magnitude of the coupling risk between nodes. By calculating the dynamic mutual information and perturbation propagation between each node and its neighboring nodes, the edge weights can describe the magnitude of the coupling risk. By constructing this dynamic coupling risk graph, it is possible to identify in real time which coupling relationships between parameters may lead to overall system instability or adverse reactions.

[0054] Multi-scale path sensitivity assessment quantifies the impact on system stability by calculating the sensitivity of each parameter and analyzing the perturbation propagation effect of different paths.

[0055] Specifically, the reachable path sensitivity integral is used to quantify the stability and sensitivity of all reachable paths from a given node to other nodes. This allows us to identify the impact of each parameter on other parameters or the entire response system under perturbation conditions. Specifically, starting from any node, the sensitivity of reachable paths from that node to all other nodes is calculated. Path sensitivity measures how the intensity of a perturbation changes as it propagates along a path. Each path may contain multiple nodes, and the sensitivity contribution of each node on this path to the perturbation is calculated. For each path, the sensitivity value can be calculated based on coupling relationships and the amount of perturbation propagation. The resulting reachable path sensitivity integral is obtained by integrating the sensitivity of all reachable paths originating from the given node, yielding a total sensitivity value. This sensitivity value reflects the node's influence within the entire network.

[0056] The contribution matrix is ​​used to quantify the contribution of each node to the stability of the entire system. Each element C in the matrix... ij This represents the stability contribution of node i to node j, and the magnitude of the contribution reflects the impact of node i on the overall stability of the system. By superimposing the path sensitivity integral and the in-chain perturbation propagation amount, the relative contribution of each node in the entire topology is obtained. Through the contribution matrix, the degree of influence of each node on stability can be quantified. For nodes with high contribution, their perturbations will have a greater impact on other parameters in the chain, so they need to be adjusted in the control strategy first.

[0057] A critical coupling chain refers to the group of parameter nodes that has the greatest impact on system stability in a multi-parameter coupled topology. The total propagation of disturbances along each coupling chain is an important indicator of that chain's impact on the system. By calculating the propagation of disturbances for all parameters within each chain, the chain with the largest total propagation of disturbances can be identified. A large total propagation of disturbances means that disturbances in some parameters can rapidly affect other critical parameters through that chain, making the system more prone to instability.

[0058] Beyond the coupling between a single node and other nodes, higher-order coupling responses are also considered. Specifically, second-order and multi-order coupling responses refer to situations where the coupling between certain parameters is not only manifested in the first-order (direct influence) but also indirectly affects other parameters through second-order or higher-order paths. Higher-order coupling responses mean that the influence of certain parameters on the system is not direct but rather generated through the synergistic effect of multiple nodes. For example, node A affects C through B, and then C affects D. Such second-order or multi-order coupling responses are more complex and difficult to control than direct coupling.

[0059] A threshold is set to filter out important coupling chains. Only when the second-order or multi-order coupling response of a chain is higher than the set threshold is it considered a critical coupling chain. This threshold is set based on the requirements of the actual production process and safety standards.

[0060] Furthermore, real-time tracking of the coordinated changes in on-chain parameters of key coupled chains includes:

[0061] Construct an intra-chain joint deviation vector to quantify the trend of intra-chain nodes simultaneously deviating from the target range; generate an intra-chain collaborative deviation index based on the joint deviation vector, and output the intra-chain collaborative deviation index as a real-time tracking result.

[0062] The intra-chain joint deviation vector is a vector used to quantify the trend of each node in the entire coupled chain deviating from the target value. The deviation of each node reflects the difference between the actual value and the target value of the parameter.

[0063] The intra-chain coordination deviation index is a numerical value representing the joint deviation degree of all nodes within the chain. This index integrates the deviations of individual nodes and reflects the coordinated change trend among them. The calculation of the intra-chain coordination deviation index includes: normalizing the joint deviation vector, weighted summing of the deviations (weights can be set according to the importance of the nodes or their contribution to stability), and evaluating the degree of coordination of the deviations of each node using Euclidean distance, weighted average, etc. The generated intra-chain coordination deviation index is output as a real-time tracking result for analysis and decision-making by operators or automatic control systems.

[0064] Furthermore, the coordinated control instructions for the reaction process are generated through online evolution, including:

[0065] The predicted structure and the rearranged regulatory priorities are fused and encoded into an intra-chain control state gene. This control state gene describes the predictive dependencies and regulatory order among intra-chain parameters and serves as the initial strategy representation for the evolutionary process. A perturbation sensitivity coefficient is generated on the initial strategy representation based on the intra-chain joint bias vector. This perturbation sensitivity coefficient measures the impact of the current intra-chain bias on the gene structure to determine the mutable location and magnitude of the initial strategy representation. The perturbation sensitivity coefficient is used to trigger directional mutations in the gene structure. The strategy candidates generated by these directional mutations are evaluated across scales, and the results of the cross-scale evaluation are used for cyclical evolution to generate coordinated regulatory instructions.

[0066] Intrachain control state genes are a coding method that describes the predictive dependencies and adjustment priorities among intrachain parameters. Here, a gene refers to a numerical or symbolic structure that can be used to describe a control strategy. It determines the behavior of the control system by encoding the predictive dependencies of various parameters in the reaction process and the order in which parameters are adjusted when disturbances occur.

[0067] By fusing the predicted structure with the rearranged regulation priorities, a unified in-chain control state gene is obtained, which reflects the predicted dependencies and regulation priorities of each parameter in the entire reaction chain. In the genetic algorithm, the in-chain control state gene serves as an initial policy representation, indicating the system's control decision policy at the current moment. The gene will serve as the starting point for the algorithm's optimization process, and subsequent policy optimization and evolution will unfold based on this initial gene.

[0068] The perturbation sensitivity coefficient is a parameter used to measure the degree of influence of changes in node deviations within the chain on the control gene structure. It reflects how deviations at different nodes within the chain affect the evolution and optimization of the entire control gene structure.

[0069] Based on the intra-chain joint deviation vector, the impact of each parameter deviating from the target value on the overall control strategy is calculated. Specifically, when the deviation of certain nodes is large, their influence on the control state gene is also large, thus affecting the adjustment direction of the control decision. By calculating the correlation between deviation and gene structure, a disturbance sensitivity coefficient for each node is generated. Nodes with high sensitivity have a greater influence on the system and should be given priority in adjustment, while nodes with low sensitivity can be adjusted later or not at all.

[0070] In other words, the disturbance sensitivity coefficient measures the feedback effect of disturbances in each node of the chain on the system stability. If a disturbance in a certain node triggers a chain reaction in other nodes of the chain, then its disturbance sensitivity coefficient will be high, which means that the node has a greater impact on the adjustment of the system control strategy.

[0071] Mutable locations refer to which parts of a gene will change. Parts with higher perturbation sensitivity coefficients will become the focus of mutation, thereby optimizing the system's control strategy. The mutation magnitude refers to the degree to which a gene changes during evolution. The greater the mutation magnitude, the more significant the adjustment of the control strategy. The mutation magnitude can be determined based on the perturbation sensitivity coefficient: nodes with high sensitivity will have larger mutation magnitudes, while nodes with low sensitivity will have smaller mutation magnitudes or remain unchanged.

[0072] In genetic algorithms, mutation is an operation used to explore new policy spaces. Directional mutation refers to using perturbation sensitivity coefficients to guide the direction and magnitude of mutations, rather than random mutations. This means that mutations occur in a directional manner, based on the analysis results of the current system state and parameter sensitivity, focusing on the parts that have a greater impact on system stability. Based on the calculated perturbation sensitivity coefficients, the nodes that have the greatest impact on the control policy are selected, and directional mutations are performed on these nodes. In other words, mutations will focus on those highly sensitive parameters to optimize the control effect.

[0073] Directional mutations generate new policy candidates from the current control policy genes. These candidates represent optimizations of existing control policies, aiming to improve the system's response to disturbances and its stability. Cross-scale evaluation involves assessing policy candidates at multiple different levels, including single nodes, parameter chains, and the overall response process. The ultimate goal is to select the optimal coordinated regulatory instruction, which is effective not only at local nodes but also maintains the overall stability of the system's response.

[0074] Furthermore, the cross-scale evaluation is performed through a cross-scale evaluation function, the evaluation characteristics of which include the deviation convergence result at the node level, the perturbation propagation reduction result at the chain level, and the global response stability maintenance result.

[0075] Cross-scale evaluation functions are mathematical tools for evaluating policy candidates. This function integrates evaluation indicators from multiple levels and calculates a comprehensive evaluation result based on the weights and relative importance of these indicators. Through this function, the effectiveness of each policy candidate can be quantified and ranked.

[0076] Evaluation methods for node-level deviation convergence results: Monitor the real-time deviation of each node and record whether the deviation gradually decreases over time until it reaches the preset target value. Evaluation methods for chain-level perturbation propagation reduction results: Analyze the propagation of perturbations within the chain based on the path and extent of perturbation propagation, and whether it can effectively reduce the spread of perturbations. If the perturbation can be quickly suppressed in the chain, the policy is evaluated as effective. Evaluation methods for maintaining global reaction stability results: Test the effectiveness of the policy candidate in maintaining system stability by simulating different types of perturbations. If global stability can be maintained, the policy is considered successful at the global level.

[0077] Furthermore, the coordinated control command is sent to the corresponding heating equipment, cooling equipment, pressure relief device, and feeding equipment, including:

[0078] Configure and verify the mapping of coordinated control commands; perform performance evaluation of the working equipment in the mapping verification window and establish evaluation feedback; perform deviation correction management of the control parameters of the coordinated control commands based on the evaluation feedback.

[0079] The mapping verification window is an operational framework that confirms and verifies the mapping relationship between coordinated control commands and actual working equipment, ensuring that commands can be correctly transmitted and applied to specific equipment.

[0080] The system will perform real-time evaluations of the equipment's performance. This means monitoring whether the equipment operates in accordance with the coordinated control instructions. The evaluation includes the actual working status of the equipment and any anomalies that occur during its execution. The results of the equipment performance evaluation will be translated into feedback, which includes whether the equipment's performance deviates from expectations, whether the execution efficiency meets the standards, and whether there are any abnormal disturbances.

[0081] Based on the obtained evaluation feedback, deviation correction of control parameters is performed. Deviation correction includes: adjusting control parameters in the coordinated control instructions to better adapt to the current state of the equipment or dynamic changes in the response process; adjusting the execution order of control instructions to ensure that high-priority parameters are adjusted first, while low-priority parameters are adjusted moderately according to the actual situation; and compensating for equipment deviations, for example, by correcting equipment response delays, calibrating sensor measurements, or compensating for coupling effects between equipment. Deviation correction management ensures more accurate execution of control instructions, continuously adapts to changes in equipment and fluctuations in the external environment, and continuously optimizes control instructions through closed-loop feedback, thereby ensuring the stability and efficiency of the production process.

[0082] Example 2: Based on the same inventive concept as the multi-parameter collaborative control method for chemical product production workshops in the preceding examples, as shown in Figure 2, this application provides a multi-parameter collaborative control system for chemical product production workshops, the system comprising:

[0083] The topology construction module 10 is used to collect temperature, pressure, concentration, and flow parameters through a distributed sensor network, and construct a multi-parameter sparse coupling topology based on the dynamic mutual information between parameters. In the multi-parameter sparse coupling topology, each node represents a single process parameter, and the edge weight represents the time-varying coupling strength between parameters. The change trend tracking module 20 is used to perform topology path sensitivity assessment of the multi-parameter sparse coupling topology, identify the key coupling chains that have the greatest impact on reaction stability, and track the collaborative change trend of parameters on the key coupling chains in real time. The priority reordering module 30 is used to establish a parameter prediction model based on the key coupling chains and real-time tracking results, and automatically reorder the adjustment priority of parameters within the chain based on a constraint migration mechanism. The instruction generation module 40 is used to construct a self-evolving collaborative control strategy generator based on the migrated dynamic constraints, using the prediction structure of the parameter prediction model and the rearranged adjustment priority as input, and generating collaborative control instructions for the reaction process through online evolution. The instruction issuing module 50 is used to issue the collaborative control instructions to the corresponding heating equipment, cooling equipment, pressure relief device, and feeding equipment.

[0084] Furthermore, the priority rearrangement module 30 is used to perform the following operation steps:

[0085] For each parameter in the key coupling chain, a joint parameter prediction model is established using the parameter's historical time-series data, real-time tracking results, and dynamic coupling coefficients of the parameters within the chain. The time-series fitting prediction results of the parameter prediction model are obtained, and an overshoot analysis of the time-series fitting prediction results is performed using the target threshold range. When the target threshold range is exceeded at any time, a constraint migration mechanism is triggered, and the overshoot of the corresponding abnormal parameter is dynamically configured to the parameters within the chain, thereby completing the automatic rearrangement of adjustment priorities.

[0086] Furthermore, the priority rearrangement module 30 is used to perform the following operation steps:

[0087] A three-dimensional regulation feature for in-chain parameter adjustment is constructed, comprising the prediction bias feature of the parameter itself, the coupling strength feature of the parameter within the chain, and the sensitivity feature of the parameter to in-chain stability. The sensitivity feature is constructed based on the sensitivity matrix of the parameter prediction model. Parameter response analysis is performed based on historical in-chain regulation data and coupling relationships to calculate the in-chain parameter response propagation amount, which is used to quantify the diffusion effect of single-parameter perturbations within the chain. The regulation priority and magnitude of the parameters are determined using the three-dimensional regulation feature and the parameter response propagation amount.

[0088] Furthermore, the topology building module 10 is used to perform the following operation steps:

[0089] Calculate the dynamic mutual information between any two parameters. This dynamic mutual information is calculated based on a sliding time window to estimate the local joint probability distribution of the parameter time-series data, capturing the nonlinear and time-varying coupling relationships between the parameters. Construct an initial fully connected weighted graph using this dynamic mutual information. Each node in the initial fully connected weighted graph represents a single parameter, and the edge weights are assigned values ​​based on the dynamic mutual information between the corresponding parameters. Perform sparsity processing on the initial fully connected weighted graph to generate a multi-parameter sparsely coupled topology. This sparsity processing includes: retaining edges with a preset number of mutual information rankings for each node, deleting the remaining edges, and performing dynamic threshold pruning analysis on the edges after configuring a dynamic threshold. Perform topology consistency evaluation on the multi-parameter sparsely coupled topology, and update the coupling relationships of the multi-parameter sparsely coupled topology based on node degree distribution and intra-chain path strength analysis.

[0090] Furthermore, the trend tracking module 20 is used to perform the following operation steps:

[0091] For the multi-parameter sparse coupled topology, a dynamic coupling risk graph is constructed based on historical node perturbation responses, dynamic mutual information, and chain perturbation propagation. For the dynamic coupling risk graph, a multi-scale path sensitivity assessment is performed, which includes: a. calculating the sensitivity integral of all reachable paths originating from each node; b. superimposing the path sensitivity integral with the intra-chain perturbation propagation to form a node contribution matrix to chain stability. Based on the node contribution matrix to chain stability, key coupling chains are identified. These key coupling chains are combinations of nodes, and the total intra-chain perturbation propagation of the key coupling chain is the largest, with the second-order and multi-order coupling responses of nodes on the chain to other nodes exceeding a set threshold.

[0092] Furthermore, the trend tracking module 20 is used to perform the following operation steps:

[0093] Construct an intra-chain joint deviation vector to quantify the trend of intra-chain nodes simultaneously deviating from the target range; generate an intra-chain collaborative deviation index based on the joint deviation vector, and output the intra-chain collaborative deviation index as a real-time tracking result.

[0094] Furthermore, the instruction generation module 40 is used to perform the following operation steps:

[0095] The predicted structure and the rearranged regulatory priorities are fused and encoded into an intra-chain control state gene. This control state gene describes the predictive dependencies and regulatory order among intra-chain parameters and serves as the initial strategy representation for the evolutionary process. A perturbation sensitivity coefficient is generated on the initial strategy representation based on the intra-chain joint bias vector. This perturbation sensitivity coefficient measures the impact of the current intra-chain bias on the gene structure to determine the mutable location and magnitude of the initial strategy representation. The perturbation sensitivity coefficient is used to trigger directional mutations in the gene structure. The strategy candidates generated by these directional mutations are evaluated across scales, and the results of the cross-scale evaluation are used for cyclical evolution to generate coordinated regulatory instructions.

[0096] Furthermore, the cross-scale evaluation is performed through a cross-scale evaluation function, the evaluation characteristics of which include the deviation convergence result at the node level, the perturbation propagation reduction result at the chain level, and the global response stability maintenance result.

[0097] Furthermore, the instruction issuing module 50 is used to perform the following operation steps:

[0098] Configure and verify the mapping of coordinated control commands; perform performance evaluation of the working equipment in the mapping verification window and establish evaluation feedback; perform deviation correction management of the control parameters of the coordinated control commands based on the evaluation feedback.

[0099] Through the foregoing detailed description of the multi-parameter collaborative control method for chemical product production workshops, those skilled in the art can clearly understand the multi-parameter collaborative control system for chemical product production workshops in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.

[0100] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A multi-parameter collaborative control method for chemical product production workshops, characterized in that, The method includes: collecting temperature, pressure, concentration, and flow parameters through a distributed sensor network; constructing a multi-parameter sparse coupling topology based on the dynamic mutual information between parameters, wherein each node in the multi-parameter sparse coupling topology represents a single process parameter, and the edge weight represents the time-varying coupling strength between parameters; performing a topology path sensitivity assessment of the multi-parameter sparse coupling topology to identify the key coupling chains that have the greatest impact on reaction stability, and tracking the collaborative change trend of parameters on the key coupling chains in real time; establishing a parameter prediction model based on the key coupling chains and the real-time tracking results, and automatically rearranging the adjustment priorities of parameters within the chains based on a constraint migration mechanism; constructing a self-evolving collaborative control strategy generator based on the migrated dynamic constraints, using the prediction structure of the parameter prediction model and the rearranged adjustment priorities as input, and generating collaborative control instructions for the reaction process through online evolution; issuing the collaborative control instructions to the corresponding heating equipment, cooling equipment, pressure relief device, and feeding equipment; constructing a multi-parameter sparse coupling topology based on the dynamic mutual information between parameters includes: calculating the dynamic mutual information between any two parameters, wherein the dynamic mutual information is estimated based on a sliding time window for local joint probability distribution of parameter time-series data. The calculation captures the nonlinear and time-varying coupling relationships between parameters; an initial fully connected weighted graph is constructed using the dynamic mutual information, where each node represents a single parameter, and the edge weights are assigned values ​​based on the dynamic mutual information between the corresponding parameters; the initial fully connected weighted graph is processed using a sparsification strategy to generate a multi-parameter sparsely coupled topology, wherein the sparsification strategy includes: retaining a preset number of edges with mutual information rankings for each node, deleting the remaining edges, and performing dynamic threshold pruning analysis on the edges after configuring a dynamic threshold; and performing topology consistency on the multi-parameter sparsely coupled topology. The evaluation process involves updating the coupling relationships of the multi-parameter sparse coupled topology based on node degree distribution and intra-chain path strength analysis; performing a topology path sensitivity assessment of the multi-parameter sparse coupled topology to identify the critical coupling chains that have the greatest impact on response stability, including: constructing a dynamic coupling risk graph for the multi-parameter sparse coupled topology based on historical node perturbation responses, dynamic mutual information, and chain perturbation propagation; and performing a multi-scale path sensitivity assessment on the dynamic coupling risk graph, which includes: a. calculating the sensitivity integral of all reachable paths originating from each node; b.The path sensitivity integral is superimposed with the intra-chain perturbation propagation amount to form a node contribution matrix to chain stability. Based on this matrix, key coupling chains are identified. These key coupling chains are node combinations with the largest intra-chain perturbation propagation amount, and the second-order and multi-order coupling responses of nodes on these chains to other nodes exceed a set threshold. Cooperative regulatory instructions for the reaction process are generated through online evolution, including: fusing the predicted structure with the rearranged regulatory priorities into an intra-chain control state gene, which describes the predictive dependency and regulatory order among intra-chain parameters, and uses this control state gene as the initial strategy representation for the evolutionary process; generating a perturbation sensitivity coefficient based on the intra-chain joint deviation vector, which measures the impact of the current intra-chain deviation on the gene structure to determine the mutable position and magnitude of the initial strategy representation; triggering directional mutations in the gene structure using the perturbation sensitivity coefficient; performing cross-scale evaluation on the strategy candidates generated by the directional mutations, and using the cross-scale evaluation results for cyclical evolution to generate cooperative regulatory instructions; and establishing a parameter prediction model based on the key coupling chains. The system automatically rearranges the adjustment priorities of in-chain parameters based on a constraint migration mechanism. This includes: establishing a joint parameter prediction model for each parameter in the key coupled chain using historical time-series data, real-time tracking results, and dynamic coupling coefficients of in-chain parameters; obtaining the time-series fitting prediction results of the parameter prediction model and performing an overshoot analysis of the time-series fitting prediction results using a target threshold range; triggering the constraint migration mechanism when any time exceeds the target threshold range, dynamically configuring the overshoot of the corresponding abnormal parameter to the in-chain parameters, and completing the automatic rearrangement of adjustment priorities; triggering the constraint migration mechanism includes: constructing three-dimensional adjustment features for in-chain parameters, including the prediction deviation features of the parameter itself, the coupling strength features of the parameter within the chain, and the sensitivity features of the parameter to in-chain stability, the sensitivity features being constructed based on the sensitivity matrix of the parameter prediction model; performing parameter response analysis based on historical adjustment data and coupling relationships within the chain, calculating the in-chain parameter response propagation amount, the parameter response propagation amount being used to quantify the diffusion effect of single-parameter perturbations within the chain; and determining the adjustment priority and adjustment magnitude of the parameters using the three-dimensional adjustment features and the parameter response propagation amount.

2. The multi-parameter collaborative control method for a chemical product production workshop as described in claim 1, characterized in that, Real-time tracking of the coordinated change trend of on-chain parameters of key coupled chains includes: constructing an intra-chain joint deviation vector to quantify the trend of intra-chain nodes simultaneously deviating from the target range; generating an intra-chain coordinated deviation index based on the joint deviation vector; and outputting the intra-chain coordinated deviation index as a real-time tracking result.

3. The multi-parameter collaborative control method for a chemical product production workshop as described in claim 1, characterized in that, The cross-scale evaluation is performed through a cross-scale evaluation function, whose evaluation characteristics include node-level deviation convergence results, chain-level perturbation propagation weakening results, and global response stability maintenance results.

4. The multi-parameter collaborative control method for a chemical product production workshop as described in claim 1, characterized in that, The coordinated control command is sent to the corresponding heating equipment, cooling equipment, pressure relief device, and feeding equipment, including: configuring a mapping verification window for the coordinated control command; performing an execution evaluation of the working equipment in the mapping verification window and establishing evaluation feedback; and performing deviation correction management of the control parameters of the coordinated control command based on the evaluation feedback.

5. A multi-parameter collaborative control system for a chemical product production workshop, characterized in that, The system, used to implement the multi-parameter collaborative control method for a chemical product production workshop according to any one of claims 1-4, comprises: a topology construction module for collecting temperature, pressure, concentration, and flow parameters through a distributed sensor network, and constructing a multi-parameter sparsely coupled topology based on the dynamic mutual information between parameters, wherein each node in the multi-parameter sparsely coupled topology represents a single process parameter, and the edge weight represents the time-varying coupling strength between parameters; a change trend tracking module for performing topology path sensitivity assessment of the multi-parameter sparsely coupled topology, identifying the key coupling chain that has the greatest impact on reaction stability, and tracking the collaborative change trend of parameters on the key coupling chain in real time; a priority reordering module for establishing a parameter prediction model based on the key coupling chain and the real-time tracking results, and automatically reordering the adjustment priority of parameters within the chain based on a constraint migration mechanism; an instruction generation module for constructing a self-evolving collaborative control strategy generator based on the migrated dynamic constraints, using the prediction structure of the parameter prediction model and the rearranged adjustment priority as input, and generating collaborative control instructions for the reaction process through online evolution; and an instruction issuing module for issuing the collaborative control instructions to the corresponding heating equipment, cooling equipment, pressure relief device, and feeding equipment.

Images