A controlled vacuum concentration method and system for cyclic adenosine extraction

By constructing and generating a basic thermo-mass migration map, the problem of difficulty in identifying the material state during the vacuum concentration of cyclic adenosine extract was solved, enabling a more accurate control strategy and improving concentration efficiency and product quality stability.

CN123101979APending Publication Date: 2026-11-17GUANGDONG NANSHAN MEDICINE INNOVATION INST +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611245928.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-17
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing vacuum concentration processes, the material state of cyclic adenosine monophosphate extract is difficult to identify accurately, and the control strategy response is lagging, resulting in unstable concentration efficiency and product quality.

Method used

Construct a basic thermal mass transfer map, generate thermal mass transfer maps and predicted thermal mass transfer maps, identify material state changes through material state trajectories, and generate adaptive control strategies.

Benefits of technology

It improves the control adaptability of the vacuum concentration process, reduces the lag judgment of single-point threshold control, and ensures that the concentration process is more in line with the changes in the state of the material.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123101979A_ABST
    Figure CN123101979A_ABST
Patent Text Reader

Abstract

This invention relates to the field of extraction and concentration process control technology, specifically to a method and system for vacuum concentration control in cyclic adenosine extraction. The method includes: obtaining the process connection relationship of a double-effect vacuum concentration unit and constructing a basic thermo-mass migration (TMM) map; generating a TTM map based on current measured operating data in each monitoring cycle; starting from the second monitoring cycle, using the TTM map generated in the previous monitoring cycle as the state map for the previous cycle, and combining it with current control parameters to generate a predicted TTM map; converting the thermo-mass state deviation between the TTM map and the predicted TTM map within consecutive monitoring cycles into a material state trajectory, determining the current material state based on the material state trajectory, and generating a vacuum concentration control strategy. This method can reflect changes in the material state during the concentration process, improving the adaptability and stability of vacuum concentration control.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of extraction and concentration process control technology, specifically to a method and system for controlled decompression concentration of cyclic adenosine extract. Background Technology

[0002] Cyclic adenosine monophosphate (cAMP) is a class of active ingredients with important applications. In the deep processing of medicinal and edible raw materials such as jujubes, it is usually necessary to obtain a concentrated solution containing cAMP through processes such as extraction, filtration, membrane separation, and concentration. Vacuum concentration can remove some water at lower temperatures, which helps to reduce the loss of heat-sensitive components and improve the efficiency of subsequent finished product processing. Therefore, it is widely used in the extraction concentration process.

[0003] Existing vacuum concentration processes largely rely on empirical settings or threshold controls for operating parameters such as temperature, vacuum level, liquid level, steam pressure, and circulation flow rate. For double-effect vacuum concentration units, there are coupling relationships between the first and second effects, including steam transfer, condensate discharge, and material circulation. Furthermore, the solid content, sugar content, viscosity, and membrane concentration state of the extract can fluctuate from batch to batch, resulting in inconsistent actual concentration effects at different concentration stages for the same control parameters.

[0004] Especially during the gradual concentration of cyclic adenosine monophosphate (cAMP) extract, phenomena such as increased material viscosity, decreased heat transfer efficiency, changes in evaporation rate, and localized stagnation can easily overlap, making it difficult to stably monitor cAMP content and material quality status online. Existing methods typically fail to reflect the continuous evolution of material status in a timely manner, and control strategies tend to lag behind changes in the concentration process, thus affecting concentration efficiency and product quality stability.

[0005] Therefore, how to accurately identify changes in the state of materials during vacuum concentration and formulate a control strategy that adapts to the current concentration state is a technical problem that urgently needs to be solved. Summary of the Invention

[0006] To address the shortcomings of existing technologies in the vacuum concentration process of cyclic adenosine monophosphate (cAMP) extract, such as difficulty in identifying the material state and lag in control response, this invention provides a vacuum concentration control method and system for cAMP extraction. By constructing a basic thermo-mass migration map, generating a thermo-mass migration map and a predicted thermo-mass migration map, the deviation between the two is converted into a material state trajectory, and a control strategy is generated accordingly, thereby improving the adaptability of the concentration process control.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for controlled vacuum concentration in cyclic adenosine extraction, the method comprising:

[0009] Obtain the process connection relationship of the double-effect vacuum concentrator, determine the graph nodes and graph edges based on the process connection relationship, and construct a basic graph of heat and mass transfer.

[0010] In each monitoring cycle, the current measured operating data is acquired, and a heat mass transfer map is generated based on the current measured operating data and the heat mass transfer baseline map;

[0011] Starting from the second monitoring cycle, the control parameters of the current monitoring cycle are obtained, and the heat mass migration map generated in the previous monitoring cycle is used as the state map of the previous cycle. Based on the state map of the previous cycle and the control parameters, a predicted heat mass migration map is generated.

[0012] The thermal mass state deviation between each thermal mass migration map and the corresponding predicted thermal mass migration map within a continuous monitoring period is converted into a material state trajectory.

[0013] The current material state is determined based on the material state trajectory, a reduced pressure concentration control strategy corresponding to the current material state is generated, and the thermo-mass migration diagram corresponding to the current monitoring cycle is used as the previous cycle state diagram for the next monitoring cycle.

[0014] Furthermore, based on the process connection relationships, graph nodes and edges are determined, and a basic graph of heat and mass transfer is constructed, including:

[0015] Obtain the material flow path, steam transfer path, condensate discharge path, and circulation return path in the double-effect vacuum concentrator unit;

[0016] The process locations in the material flow path where material residence, evaporation, heat exchange, and circulation state changes are determined as graph nodes;

[0017] The material flow relationship, steam transfer relationship, condensate discharge relationship and circulation return relationship between adjacent graph nodes are defined as graph edges;

[0018] Connect the nodes and edges of the graph according to the connection sequence corresponding to the material flow path, steam transfer path, condensate discharge path and circulation return path to obtain the basic graph of heat and mass transfer.

[0019] Further, generating a heat-mass transfer map based on the current measured operating data and the basic heat-mass transfer map includes:

[0020] Acquire the operating data of the first effect, the second effect, the condensate, and the circulation data for the current monitoring cycle;

[0021] The current thermo-mass state value of each node in the diagram is determined based on the first-effect operation data and the second-effect operation data;

[0022] The current migration state value of each graph edge is determined based on the condensate data and the cycle operation data;

[0023] The current thermo-mass state value and the current migration state value are mapped to the corresponding graph nodes and graph edges in the thermo-mass migration base graph to obtain the thermo-mass migration graph for the current monitoring period.

[0024] Further, based on the previous cycle state diagram and the control parameters, a predicted thermo-mass migration diagram is generated, including:

[0025] Extract the previous cycle thermo-mass state value of each graph node and the previous cycle migration state value of each graph edge from the previous cycle state diagram.

[0026] Based on the control parameters, the previous cycle thermo-mass state value, and the previous cycle migration state value, a migration prediction value for each graph edge is generated.

[0027] Along the connection direction of each graph edge, the migration prediction value is accumulated to the adjacent graph node to obtain the thermal mass prediction value of each graph node;

[0028] The predicted thermal mass and the predicted migration values ​​are mapped to the corresponding graph nodes and edges in the basic thermal mass migration graph to obtain the predicted thermal mass migration graph.

[0029] Further, based on the control parameters, the previous cycle thermomass state value, and the previous cycle migration state value, migration prediction values ​​for each graph edge are generated, including:

[0030] According to the edge type of each graph edge, extract the corresponding control parameter components from the control parameters;

[0031] Calculate the control increment of the control parameter component relative to the corresponding control parameter component in the previous monitoring cycle;

[0032] The state response coefficients of each graph edge are generated based on the migration state value of each graph edge in the previous cycle and the thermo-mass state value of its adjacent graph nodes in the previous cycle.

[0033] The control increment and the state response coefficient are used to correct the migration state value of the corresponding graph edge in the previous cycle, so as to obtain the migration prediction value of each graph edge.

[0034] Further, the step of converting the thermo-mass migration map deviation between each thermo-mass migration map and the corresponding predicted thermo-mass migration map within a continuous monitoring period into a material state trajectory includes:

[0035] Starting from the second monitoring period, the thermal mass migration map and the predicted thermal mass migration map for the same monitoring period will be paired with graph nodes and graph edges.

[0036] The node deviation is obtained by subtracting the predicted thermal mass value of the graph node from the current thermal mass state value of the paired graph node.

[0037] The edge deviation is obtained by subtracting the predicted edge migration value from the current migration state value of the paired edge.

[0038] According to the connection direction of each graph edge, the node deviation and the edge deviation are converted into material status source items of each graph node in the corresponding monitoring period;

[0039] The material status source items are arranged in order of monitoring cycle to obtain the material status trajectory.

[0040] Further, according to the connection direction of each graph edge, the node deviation and the edge deviation are converted into material status source items for each graph node in the corresponding monitoring period, including:

[0041] Taking any graph node as the target graph node, the graph edges connected to the target graph node are divided into inbound graph edges and outbound graph edges according to the connection direction.

[0042] The edge deviations of the incoming graph edges and the outgoing graph edges are summarized separately to obtain the incoming migration deviation and the outgoing migration deviation.

[0043] The difference between the inbound migration deviation and the outbound migration deviation is taken as the net migration deviation of the target graph node;

[0044] The net migration deviation is subtracted from the node deviation of the target graph node to obtain the material status source item of the target graph node in the corresponding monitoring period.

[0045] Further, determining the current material state based on the material state trajectory includes:

[0046] The trajectory components of each graph node within the continuous monitoring period are extracted from the material state trajectory, and the trajectory components are used as the material state source items of the corresponding graph node.

[0047] For the material state source item corresponding to the same graph node, the sign consistency, duration period number and change slope are calculated to obtain the state duration characteristics of the corresponding graph node.

[0048] Dominant state nodes are selected from each graph node according to the described state persistence characteristics;

[0049] The current material state is determined based on the path position of the dominant state node in the thermal mass transfer baseline diagram and the change direction of the material state source item corresponding to the dominant state node.

[0050] Furthermore, the generation of a reduced-pressure concentration control strategy corresponding to the current material state includes:

[0051] Obtain the dominant state node and material state source item corresponding to the current material state;

[0052] By tracing back along the graph edges connected to the dominant state node in the basic thermal mass transfer graph, candidate controlled graph edges that have a migration association with the dominant state node are obtained.

[0053] Based on the path distance, edge type, and historical adjustment status between the candidate controlled graph edge and the dominant state node, the control priority of the candidate controlled graph edge is generated;

[0054] According to the control priority, the intensity and change slope of the material state source item are converted into the control adjustment amount of each candidate controlled graph edge;

[0055] The decompression and concentration control strategy is generated based on the control adjustment amount of each candidate controlled graph edge.

[0056] A reduced-pressure concentration control system for cyclic adenosine extraction, used to implement the aforementioned reduced-pressure concentration control method for cyclic adenosine extraction, the system comprising:

[0057] The basic graph construction module is used to obtain the process connection relationship of the double-effect vacuum concentrator, determine the graph nodes and graph edges based on the process connection relationship, and construct the basic graph of heat and mass transfer.

[0058] The migration map generation module is used to acquire the current measured operating data in each monitoring cycle and generate a thermo-mass migration map based on the current measured operating data and the thermo-mass migration baseline map.

[0059] The prediction map generation module is used to obtain the control parameters of the current monitoring cycle starting from the second monitoring cycle, use the heat mass migration map generated in the previous monitoring cycle as the state map of the previous cycle, and generate a prediction heat mass migration map based on the state map of the previous cycle and the control parameters.

[0060] The trajectory conversion module is used to convert the thermal mass migration map and the corresponding predicted thermal mass migration map within a continuous monitoring period into a material state trajectory.

[0061] The control update module is used to determine the current material state based on the material state trajectory, generate a depressurization concentration control strategy corresponding to the current material state, and use the thermo-mass migration diagram corresponding to the current monitoring cycle as the previous cycle state diagram for the next monitoring cycle.

[0062] Compared with the prior art, the beneficial effects of the present invention are:

[0063] This invention constructs a thermo-mass migration baseline diagram of the material flow, steam transfer, condensate discharge, and circulation reflux relationships in a double-effect vacuum concentrator unit, and generates a thermo-mass migration diagram in each monitoring cycle, so that dispersed temperature, vacuum degree, liquid level, condensation, and circulation data can be expressed under a unified process topology, which is convenient for reflecting the thermo-mass coupling relationship between the first and second effects.

[0064] This invention generates a predicted thermo-mass migration diagram based on the previous cycle state diagram and the current control parameters, converts the deviation between the measured state and the predicted state into a material state trajectory, and generates a control strategy in reverse based on the dominant state node. This can reduce the lag judgment caused by single-point threshold control and make the reduced pressure concentration control more adaptable to the material state changes during the concentration process of cyclic adenosine monophosphate extract. Attached Figure Description

[0065] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0066] Figure 1 This is a schematic diagram illustrating an application scenario of the dual-effect vacuum concentrator provided in an embodiment of the present invention.

[0067] Figure 2 A schematic flowchart of a method for controlled decompression concentration of cyclic adenosine extract provided in an embodiment of the present invention;

[0068] Figure 3 This is a schematic diagram of the structure of a reduced pressure concentration control system for cyclic adenosine extraction provided in an embodiment of the present invention.

[0069] Attached reference numerals: 1. Feed tank; 2. First-effect heat exchanger; 3. First-effect separator; 4. Second-effect heat exchanger; 5. Second-effect separator; 6. Condenser; 7. Vacuum pump; 8. Discharge tank; 9. Control unit. Detailed Implementation

[0070] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0071] Example 1

[0072] Please see Figure 1 This invention provides a method for controlling the reduced pressure concentration of cyclic adenosine extract, which can be applied to the dual-effect reduced pressure concentration production scenario after membrane separation or pre-concentration of cyclic adenosine extract. Figure 1The application scenario of a double-effect vacuum distillation unit is illustrated. Feed tank 1, serving as the feed end of the material flow path, is used to contain the cyclic adenosine monophosphate (cAMP) extract to be concentrated. A first-effect heat exchanger 2 and a first-effect separator 3 together constitute a first-effect concentration unit. The first-effect heat exchanger 2 heats the cAMP extract entering the first-effect concentration unit, while the first-effect separator 3 separates the heated material into gas and liquid phases, and allows the generated secondary steam to migrate along the steam transfer path to the second-effect concentration unit. A second-effect heat exchanger 4 and a second-effect separator 5 together constitute a second-effect concentration unit. The second-effect heat exchanger 4 uses the secondary steam generated in the first-effect concentration unit to exchange heat with the material entering the second-effect concentration unit, while the second-effect separator 5... Separator 5 is used for gas-liquid separation of the material after the double-effect heat exchange; condenser 6 is used for condensing the gas phase vapor discharged from the double-effect separator 5 and forming a corresponding condensate discharge process; vacuum pump 7 and condenser 6 together form a vacuum maintenance unit, used to maintain the negative pressure state required for the first-effect and second-effect concentration units to perform reduced pressure concentration; discharge tank 8, as the discharge end of the material flow path, is used to receive the cyclic adenosine monophosphate extract after evaporation and concentration by the first-effect and second-effect concentration units; control unit 9 is used to acquire the current measured operating data of each detection point of the double-effect vacuum concentration unit and output control parameters to the corresponding actuators according to the generated reduced pressure concentration control strategy. The solid lines and arrows in the figure represent the migration path and migration direction of the material or vapor, and the dashed lines represent the signal connection relationship between each detection point or actuator and control unit 9.

[0073] In actual production, cyclic adenosine monophosphate (cAMP) extract is usually derived from the extract or membrane concentrate of raw materials such as jujubes. These materials, in addition to the target active ingredient, also contain sugars, polysaccharides, pigments, and other soluble solids. As the vacuum concentration process progresses, the material volume gradually decreases, while the solid content and viscosity gradually increase, leading to changes in heat transfer resistance, circulation flow state, and evaporation intensity. Therefore, even if external control parameters such as temperature, vacuum level, steam pressure, or circulation flow rate are maintained within the set range, the actual material state inside the unit may still deviate. For example, changes in the heat input on the first-effect side may affect the evaporation state on the second-effect side through secondary steam; unstable condensate discharge may alter the vacuum maintenance effect; and insufficient circulation reflux may lead to localized material retention or heat transfer attenuation.

[0074] Current vacuum distillation control typically relies on single-point temperature, pressure, liquid level, or flow rate as the basis for judgment, and then adjusts steam, vacuum, circulation, or discharge parameters based on empirical thresholds. This method is applicable to common low-viscosity materials, but in the concentration of cyclic adenosine monophosphate (cAMP) extracts, the material's quality state and the retention status of its active ingredients are difficult to directly reflect through a single online detection value. Especially in double-effect vacuum distillation units, there is a significant thermo-mass migration coupling relationship between the first and second effects, condensation, and circulation. Fluctuations at a single detection point may originate from changes in control parameters, or from increased material viscosity, decreased heat transfer efficiency, localized stagnation, or evaporation load shifts. If control is based solely on whether the detection value exceeds the limit, it is easy to misjudge normal unit response as material abnormality, or fail to adjust the control strategy in a timely manner even when the material condition has continuously deteriorated.

[0075] Based on the aforementioned operating characteristics, this embodiment does not simply interpret the vacuum concentration process as a combination of multiple independent control loops. Instead, it views the material flow, steam transfer, condensate discharge, and recirculation in the double-effect vacuum concentration unit as a heat and mass transfer process with directional relationships. Specifically, the locations in the unit where material residence, evaporation, heat exchange, and circulation state changes occur can be abstracted as graph nodes, and the material flow, steam transfer, condensate discharge, and recirculation relationships between adjacent locations can be abstracted as graph edges, thereby forming a basic heat and mass transfer graph that can express the inherent coupling relationships of the double-effect vacuum concentration process.

[0076] Furthermore, this embodiment generates a thermo-mass migration map for the current monitoring cycle using current measured operating data to reflect the actual thermo-mass migration state of the unit. Simultaneously, it uses the thermo-mass migration map from the previous cycle as the previous cycle's state map, combined with current control parameters, to generate a predicted thermo-mass migration map, representing the theoretically expected thermo-mass migration state under the existing state and current control actions. Deviations between the actual and predicted states are no longer directly treated as single-point anomalies, but are further converted into material state trajectories along the connection relationships of graph nodes and edges. These material state trajectories reflect the evolution direction and continuous characteristics of the material state within the continuous monitoring cycle, thus providing a basis for identifying the current material state and generating appropriate vacuum concentration control strategies.

[0077] Therefore, the core principle of this embodiment is to use the process connection relationship of the double-effect vacuum concentration unit to constrain the data interpretation process, and to transform discrete operating data such as temperature, pressure, condensation, and circulation into a material state expression with spatial migration relationship and temporal continuity, so as to avoid relying solely on single-point thresholds for empirical judgment, and to make vacuum concentration control more closely match the actual state changes of cyclic adenosine monophosphate extract during the concentration process.

[0078] Please see Figure 2The present invention provides an embodiment of a method for controlled vacuum concentration in the extraction of cyclic adenosine monophosphate (cAMP), the specific steps of which are as follows:

[0079] S1: Obtain the process connection relationship of the double-effect vacuum concentrator, determine the graph nodes and graph edges based on the process connection relationship, and construct the basic graph of heat and mass transfer.

[0080] In this embodiment, the double-effect vacuum concentrator unit is not composed of a single concentrator tank or a single control loop, but rather includes interconnected process components such as first-effect concentration, second-effect concentration, steam transfer, condensate discharge, material conveying, and recirculation. During the concentration process of the cyclic adenosine monophosphate extract, material flow, heat input, water evaporation, secondary steam transfer, and condensate discharge occur simultaneously, with clear migration directions between different process locations. Therefore, a basic thermo-mass migration diagram is first established based on the unit's actual pipeline connections, valve connections, and process flow direction, allowing subsequent operating data to be interpreted within the specific process locations and migration relationships.

[0081] Specifically, the nodes in the diagram can correspond to locations that reflect the material's residence, heat exchange, evaporation, or phase change states, such as the first-effect heating zone, second-effect heating zone, material storage zone, recirculation zone, and condensate discharge zone. The edges can correspond to material flow edges, steam migration edges, condensate discharge edges, and recirculation edges. The setting of the nodes and edges is not required to be limited to a specific equipment structure, as long as it can express the material, heat, and condensate migration relationships in the double-effect vacuum concentration process. Those skilled in the art will understand that the basic heat and mass transfer diagram can be established based on the unit process flow diagram, pipeline layout diagram, or control point table; this application does not impose further limitations on this.

[0082] S2: In each monitoring cycle, acquire the current measured operating data, and generate a thermal mass transfer map based on the current measured operating data and the thermal mass transfer baseline map.

[0083] In this embodiment, the currently measured operating data is used to reflect the actual operating status of the double-effect vacuum concentrator unit during the current monitoring cycle. The measured operating data may include first-effect temperature, first-effect pressure or vacuum, first-effect liquid level, second-effect temperature, second-effect pressure or vacuum, second-effect liquid level, steam pressure, condensate flow rate, condensate temperature, circulating pump operating frequency, and circulating flow rate. In practical applications, the collected data can be first processed by time alignment, outlier removal, sliding filtering, or dimensional normalization to reduce the impact of instantaneous sensor fluctuations on status judgment.

[0084] When generating a thermo-mass migration map, the values ​​at each detection point are not simply recorded. Instead, the detected values ​​are mapped to corresponding nodes and edges in the basic thermo-mass migration map. The current thermo-mass state value of a node can be used to represent the temperature, pressure, liquid level, evaporation intensity, or heating state at the corresponding process location; the current migration state value of an edge can be used to represent the material migration intensity, vapor migration intensity, condensate discharge intensity, or circulation reflux intensity between adjacent process locations. In this way, the originally dispersed sensor data is transformed into a graph state with process location and migration direction, which is beneficial for identifying coupling changes in the double-effect concentration process, rather than judging solely based on whether a single detected quantity exceeds the limit.

[0085] S3: Starting from the second monitoring cycle, obtain the control parameters for the current monitoring cycle, use the heat mass migration map generated in the previous monitoring cycle as the state map of the previous cycle, and generate a predicted heat mass migration map based on the state map of the previous cycle and the control parameters.

[0086] In this embodiment, the predicted thermo-mass migration map is used to represent the theoretically expected thermo-mass migration result caused by the current control parameters based on the state of the previous cycle. The control parameters may include control quantities related to the vacuum concentration process, such as heating steam regulation, vacuum regulation, circulating pump regulation, condensate discharge regulation, and material feed / discharge regulation. Since the state map of the previous cycle does not exist in the first monitoring cycle, the prediction process begins from the second monitoring cycle.

[0087] It should be noted that the viscosity and solids content of the cyclic adenosine monophosphate extract gradually change during the concentration process, and the same control parameter does not produce entirely the same effect at different concentration stages. For example, the same steam valve adjustment range may rapidly increase the evaporation intensity at low viscosity stages, while at high viscosity stages it may mainly manifest as increased local heating or decreased heat transfer efficiency. Therefore, in this embodiment, when generating the predicted thermo-mass migration map, the thermo-mass state values ​​of the nodes in the previous period and the migration state values ​​of the edges are used as the state basis, and then combined with the current control parameters for calculation, so that the prediction result is simultaneously affected by the control effect and the material state of the previous period.

[0088] S4: Convert the thermal mass migration map deviation between each thermal mass migration map and the corresponding predicted thermal mass migration map within the continuous monitoring period into a material state trajectory.

[0089] In this embodiment, deviation from the thermomass transfer state is not directly equivalent to equipment malfunction or material malfunction. For the double-effect vacuum concentration process, the difference between the measured state and the predicted state may originate from normal control response, sensor fluctuations, changes in steam load, or from increased material viscosity, heat transfer attenuation, local stagnation, or poor condensation. Therefore, in this embodiment, the thermomass transfer map and the predicted thermomass transfer map are paired in terms of nodes and edges to obtain the node deviation and edge deviation, which are then converted based on the connection direction of the edges.

[0090] The essence of this transformation lies in using the directional relationships in the thermo-mass migration baseline diagram to constrain and interpret the sources of deviation. For example, when the state deviation of a certain graph node can be explained by the migration deviation of the inbound or outbound graph edges, this deviation is more likely to be a transitive change caused by the migration process; when the node deviation persists after deducting the net migration deviation, it better reflects the change in the material state itself corresponding to that process location. The resulting material state source terms are arranged in chronological order to form a material state trajectory, which is used to express the evolution trend of the material state within a continuous monitoring period.

[0091] S5: Determine the current material state based on the material state trajectory, generate a depressurization concentration control strategy corresponding to the current material state, and use the thermo-mass migration diagram corresponding to the current monitoring cycle as the previous cycle state diagram for the next monitoring cycle.

[0092] In this embodiment, the material state trajectory is used to identify the dominant state change in the current concentration stage. The sign consistency, duration period number, and change slope of the material state source items corresponding to the same graph node can be analyzed to determine whether the state change is continuous and dominant. If a material state source item at a certain location maintains the same direction of change over multiple monitoring periods, and the change slope continuously increases, it indicates that this location may become the dominant state node in the current concentration process. Based on the path position of the dominant state node in the thermo-mass migration baseline graph and the direction of its source item change, it can be determined whether the current material state belongs to a state type such as evaporation load shift, decreased heat transfer efficiency, enhanced circulation retention, insufficient condensate discharge, or approaching concentration endpoint.

[0093] When generating a vacuum concentration control strategy, the dominant state nodes can be traced backward along the thermo-mass migration baseline to find candidate controlled graph edges that can regulate that state. Control priorities are then determined based on path distance, edge type, and historical regulation states. This results in a control strategy that is not simply triggered by temperature, pressure, or level thresholds, but rather adjusts based on the dominant changes reflected in the material state trajectory. For example, when the material state trajectory shows increased circulation retention, priority can be given to adjusting control quantities related to circulation reflux; when the state trajectory shows an abnormal shift of evaporation load to the second-effect side, priority can be given to coordinating control quantities related to steam transfer and condensate discharge. The thermo-mass migration graph of the current monitoring cycle serves as the previous cycle's state graph after the control strategy is generated, ensuring continuous recursion in the control process and thus improving the vacuum concentration process's adaptability to changes in material state.

[0094] Preferably, based on the process connection relationships, graph nodes and graph edges are determined to construct a basic heat and mass transfer graph, including:

[0095] S1.1: Obtain the material flow path, steam transfer path, condensate discharge path, and circulation return path in the double-effect vacuum concentrator unit.

[0096] S1.2: The process locations in the material flow path where material residence, evaporation, heat exchange, and circulation state changes are determined as graph nodes.

[0097] S1.3: The material flow relationship, steam transfer relationship, condensate discharge relationship and circulation return relationship between adjacent graph nodes are determined as graph edges.

[0098] S1.4: Connect the nodes and edges of the graph according to the connection sequence corresponding to the material flow path, steam transfer path, condensate discharge path and circulation return path to obtain the basic heat and mass transfer graph.

[0099] In this embodiment, the basic thermal mass transfer diagram is used to represent the fixed process connections within the double-effect vacuum distillation unit, and it does not change with the operating data of a single monitoring cycle. The construction of this basic diagram can be based on the process flow diagram, pipeline connection diagram, valve location table, sensor layout table, and control actuator list of the double-effect vacuum distillation unit. All of the above information is conventional engineering data obtainable during the installation and commissioning of the concentration equipment. Those skilled in the art can organize this data according to the actual equipment configuration; this application does not limit the specific equipment model.

[0100] When determining the process connections, the flow direction of the media within the unit should be distinguished first. The material flow path refers to the flow path formed between the first-effect concentration unit, the second-effect concentration unit, and the discharge end after the cyclic adenosine monophosphate extract enters the concentrator from the feed end; the steam transfer path refers to the transfer path of heating steam entering the first-effect heat exchange position, and the transfer path of secondary steam generated in the first effect entering the second-effect heat exchange position; the condensate discharge path refers to the path where condensate is formed and discharged after heat exchange with heating steam or secondary steam; the circulation return path refers to the path where the material returns to the heated position within the corresponding concentration unit via a circulation pump or return pipeline. If the actual unit is equipped with a buffer tank, gas-liquid separator, condenser, vacuum interface, or sampling port, these can be categorized into the corresponding paths based on whether they participate in the migration relationships of materials, steam, condensate, or circulation return.

[0101] The fundamental graph of heat and mass transfer can be represented as a directed graph. .in, Represents the set of graph nodes. This represents the set of graph edges. Each graph node corresponds to a process location capable of handling thermo-mass conditions, and each graph edge corresponds to a directional migration relationship between two adjacent graph nodes. The direction is determined based on the design flow direction of materials, steam, condensate, or recirculation. When a path experiences changes in its open / closed state under different operating conditions, that path can still be included as a graph edge in the thermo-mass migration baseline graph. The validity of a graph edge within a specific monitoring period is determined by the corresponding valve status, pump operating status, or flow rate detection value.

[0102] The division of graph nodes is based on the principles of state attribution, data mapping, and connectivity of migration relationships. State attribution means that the process location can correspond to thermo-mass states such as temperature, pressure, liquid level, vacuum degree, circulation flow rate, and evaporation intensity; data mapping means that the process location can directly receive sensor data or can have its state value calculated from adjacent detection data; connectivity of migration relationships means that there is a clear material, steam, condensate, or circulation return relationship between the process location and adjacent locations. For example, the first-effect heating zone, the second-effect heating zone, the first-effect gas phase outlet, the second-effect gas phase outlet, the condensate outlet, the material circulation inlet, and the material circulation outlet can all be set as graph nodes. When there are both liquid phase residence areas and gas phase migration areas within the same equipment, they can be split into two graph nodes according to the detection accuracy requirements to avoid mixing the liquid phase material state and the gas phase vapor state in the same node.

[0103] The division of graph edges is based on the actual migration medium and direction between adjacent graph nodes. Material flow edges represent the transport relationship of cyclic adenosine monophosphate extract between adjacent material nodes; steam transfer edges represent the transfer relationship of heating steam or secondary steam between adjacent gas phase nodes; condensate discharge edges represent the migration relationship of condensate from the heat exchange position or condensation position to the discharge end; and circulation reflux edges represent the relationship of material returning from the circulation outlet to the heating position or mixing position. If there are migration relationships of different media between two nodes, different types of graph edges should be established separately, rather than merging them into a single graph edge. For example, there may be both secondary steam transfer relationships and material transfer relationships between a first-effect concentration unit and a second-effect concentration unit; in this case, steam transfer edges and material flow edges can be set separately.

[0104] Each node in the basic heat and mass transfer diagram can be configured with a node number, node type, corresponding process location, mappable operational data type, and associated path; each edge can be configured with an edge number, starting node, ending node, edge type, migration medium, designed flow direction, associated actuator, and associated detection data. The node types can include material residence nodes, evaporation heat exchange nodes, gas phase migration nodes, condensate discharge nodes, and circulation return nodes; the edge types can include material flow edges, steam transfer edges, condensate discharge edges, and circulation return edges. These settings ensure that operational data from different sources can be mapped to the corresponding nodes or edges, and are not intended to limit the equipment structure.

[0105] When determining the connection sequence, the process design flow direction is used as the primary basis, and the layout of detection points and the position of the actuators are used as verification criteria. For material flow paths, the connection sequence can be: feed end, first material residence position, first heating position, second material residence position, second heating position, and discharge end. For steam transfer paths, the connection sequence can be: heating steam inlet, first-effect heat exchange position, first-effect secondary steam outlet, second-effect heat exchange position, and second-effect gas phase outlet. For condensate discharge paths, the connection sequence can be: heat exchange condensation position, condensate collection position, and condensate outlet. For circulation return paths, the connection sequence can be: circulation outlet, circulation pump, return inlet, and heating position. If the actual unit uses parallel branches or bypass branches, corresponding graph edges can be added according to the actual start and end positions of the branches, and the valve status or pump operating status corresponding to the branch can be recorded in the graph edge attributes.

[0106] After the above processing, the resulting heat and mass transfer baseline diagram reflects the fixed topological relationships of the double-effect vacuum concentrator unit. This baseline diagram is established before production or during equipment commissioning and remains unchanged as long as the equipment structure, piping connections, or monitoring points remain unchanged. When equipment maintenance leads to adjustments in piping connections, valve positions, or monitoring points, the heat and mass transfer baseline diagram can be regenerated based on the adjusted process connections. The purpose of this setup is to fix the heat transfer, material flow, and condensate discharge relationships within the unit into a calculable graphical structure, enabling the operational data collected in each monitoring cycle to be compared, predicted, and analyzed for deviations within the same topological framework.

[0107] Preferably, generating a heat-mass transfer map based on the current measured operating data and the heat-mass transfer baseline map includes:

[0108] S2.1: Obtain the first-effect operation data, second-effect operation data, condensate data, and circulation operation data for the current monitoring cycle.

[0109] S2.2: Determine the current thermo-mass state value of each node in the diagram based on the first-effect operation data and the second-effect operation data.

[0110] S2.3: Determine the current migration state value of each graph edge based on the condensate data and the cycle operation data.

[0111] S2.4: Map the current thermo-mass state value and the current migration state value to the corresponding graph nodes and graph edges in the thermo-mass migration base graph to obtain the thermo-mass migration graph for the current monitoring period.

[0112] In this embodiment, the current measured operating data refers to the operating data collected from the double-effect vacuum concentrator unit within the same monitoring cycle and after time alignment. The monitoring cycle can be set according to the unit's thermal inertia, sensor sampling frequency, and control actuator response time. For example, when the sampling interval of sensors such as temperature, pressure, liquid level, and flow rate is 1 second, the monitoring cycle can be set to 30 seconds. Within each monitoring cycle, the sampled values ​​at the same detection point are processed by median filtering and arithmetic mean, and the processed value is used as the valid measured value for that monitoring cycle. If a sampled value exceeds the instrument's range, or the variation of adjacent sampled values ​​exceeds three times the allowable variation of that detection point within a sampling interval, then that sampled value is discarded as an abnormal sampled value. The allowable variation range can be determined based on continuous stable operating data during the equipment commissioning phase, or it can be set according to the process safety control range.

[0113] The operating data for the first effect may include the evaporation temperature, vacuum level, liquid level, heating steam pressure, material inlet temperature, and material outlet temperature. The operating data for the second effect may include the evaporation temperature, vacuum level, liquid level, gas inlet temperature, gas outlet temperature, and material outlet temperature. Condensate data may include the instantaneous flow rate, cumulative condensate volume within the monitoring period, condensate temperature, and condensate discharge pressure. Circulation operating data may include the circulation pump frequency, circulation pump current, circulation flow rate, circulation inlet pressure, and circulation outlet pressure. If some of these monitoring points are not configured in the actual equipment, data from adjacent monitoring points or equivalent operating quantities recorded by the equipment controller can be used as substitutes. Those skilled in the art can determine the specific data source based on the instrumentation configuration of the double-effect vacuum concentrator unit; this application does not impose further limitations.

[0114] To map operational data with different dimensions to the same thermomass transfer baseline diagram, normalization can be performed on each valid measured value. For any operational data... Its normalized value It can be determined in the following way:

[0115]

[0116] in, This represents the lower limit of the operating data under the corresponding concentration process. This represents the upper limit of the operating data under the corresponding concentration process. This means that the calculation result is restricted to a closed interval between 0 and 1. and It can be determined based on process specifications, equipment operating range, or historical qualified batch data. For example, the evaporation temperature of the first effect... and The lower and upper limits of the permissible evaporation temperature range for the process are acceptable; the instantaneous flow rate of the condensate is... and The lower and upper limits of the allowable flow rate of condensate during the stable concentration stage can be used. For vacuum data, it can be uniformly converted to absolute pressure or to negative absolute pressure before normalization to avoid inconsistencies in direction caused by different instrument display methods.

[0117] The current thermomass state (TMS) value describes the heat state, pressure state, liquid level state, and evaporation state at the corresponding graph node. For evaporation heat transfer nodes such as single-effect or double-effect heating zones, the normalized evaporation temperature, vacuum degree, liquid level, and material inlet / outlet temperature difference can be used to form a node state vector, which serves as the current TMS value for that graph node. For gas phase migration nodes, the normalized gas phase temperature, gas phase pressure, and evaporation intensity of adjacent evaporation zones can be used to form the current TMS value. For condensate discharge nodes, the condensate temperature, condensate discharge pressure, and cumulative condensate volume can be used to form the current TMS value. If a graph node corresponds to only a single detection quantity, the current TMS value of that node can be represented by the normalized value of that detection quantity; if it corresponds to multiple detection quantities, the current TMS value is represented in vector form.

[0118] The current migration state value describes the actual intensity of the migration relationship corresponding to the graph edge within the current monitoring period. For material flow edges, the current migration state value can be determined based on material flow rate, liquid level changes at adjacent material nodes, and the operating status of the material transfer pump. For steam transfer edges, the current migration state value can be determined based on steam pressure, vapor phase temperature, pressure difference between adjacent vapor phase nodes, and steam valve opening. For condensate discharge edges, the current migration state value can be determined based on instantaneous condensate flow rate, cumulative condensate volume, and condensate discharge pressure. For circulation return edges, the current migration state value can be determined based on circulation pump frequency, circulation flow rate, circulation inlet pressure, and circulation outlet pressure.

[0119] In one specific implementation, the current transition state value of a graph edge can be obtained by weighting the normalized running data associated with that edge. For any graph edge... Its current migration status value It can be represented as:

[0120]

[0121] in, To be with the edge of the graph The associated first Normalized operational data, for The corresponding weights The weights can be set based on the directness of the corresponding migration relationship to the operating data. For example, for the circulation backflow side, the circulation flow rate directly reflects the backflow intensity, and its weight can be set to 0.5; the circulation pump frequency reflects the degree of action of the actuator, and its weight can be set to 0.3; the pressure difference between the circulation inlet pressure and the circulation outlet pressure reflects the backflow resistance, and its weight can be set to 0.2. For the condensate discharge side, the instantaneous condensate flow rate and the cumulative condensate volume within the monitoring period directly reflect the discharge status, and their weights can be set to 0.5 and 0.3 respectively, while the condensate discharge pressure weight can be set to 0.2. These weights can be verified based on stable batch operating data during the equipment commissioning phase, or they can be set by process engineers based on the correlation between the detected quantity and the migration status.

[0122] When mapping the current thermomass state value and the current migration state value to the thermomass migration baseline graph, the data is written according to the node number, edge number, corresponding process location, and associated detection data pre-recorded in the graph. Each graph node is written with its current thermomass state value, and each graph edge is written with its current migration state value, while simultaneously recording the corresponding monitoring cycle number and sampling time. If a graph edge's corresponding valve is closed or its corresponding pump is stopped during the current monitoring cycle, the current migration state value of that graph edge is set to 0. If the graph edge has no direct flow detection value but the associated actuator is open, an equivalent migration state value can be calculated based on the pressure difference, level difference, or temperature difference between adjacent graph nodes.

[0123] The resulting thermo-mass migration map is a dynamic graph structure that overlays the measured conditions of the current monitoring period onto the basic thermo-mass migration map. This thermo-mass migration map retains the fixed process connections of the double-effect vacuum concentrator unit while also carrying the actual operating data of each process location and migration path within the current monitoring period.

[0124] Preferably, based on the previous cycle state diagram and the control parameters, a predicted thermo-mass migration diagram is generated, including:

[0125] S3.1: Extract the previous cycle thermo-mass state value of each graph node and the previous cycle migration state value of each graph edge from the previous cycle state diagram.

[0126] S3.2: Generate the migration prediction value of each graph edge based on the control parameters, the thermo-mass state value of the previous cycle, and the migration state value of the previous cycle.

[0127] S3.3: Along the connection direction of each graph edge, the migration prediction value is accumulated to the adjacent graph node to obtain the thermal mass prediction value of each graph node.

[0128] S3.4: Map the predicted thermal mass values ​​and the predicted migration values ​​to the corresponding graph nodes and edges in the thermal mass migration baseline graph to obtain the predicted thermal mass migration graph.

[0129] In this embodiment, the predicted calorific value transfer (CVP) map is used to represent the theoretical CVP state of the double-effect vacuum concentrator unit, recursively derived from the operating state of the previous cycle under the current control parameters. This prediction result is not directly given based on a fixed threshold, but rather uses the previous cycle's state map as the starting point, incorporating the influence of the current control parameters on the migration relationships of each graph edge into the CVP base map, thereby obtaining the predicted map state for the current monitoring cycle. Since a referable previous cycle state map has not yet been formed for the first monitoring cycle, the predicted CVP map is generated starting from the second monitoring cycle.

[0130] The previous cycle state diagram already includes the previous cycle thermo-mass state values ​​of each graph node and the previous cycle migration state values ​​of each graph edge. The previous cycle thermo-mass state values ​​of graph nodes can be state vectors formed from normalized data such as temperature, pressure, liquid level, and evaporation state; the previous cycle migration state values ​​of graph edges can be material flow intensity, steam transfer intensity, condensate discharge intensity, or circulation reflux intensity. To maintain consistency in time recursion, the previous cycle thermo-mass state values ​​and previous cycle migration state values ​​adopt the same normalization scale as the current measured operating data, with values ​​ranging from 0 to 1.

[0131] The control parameters for the current monitoring cycle can include control quantities related to the vacuum concentration process, such as the opening degree of the heating steam valve, the opening degree of the vacuum regulating valve, the frequency of the circulating pump, the opening degree of the condensate discharge valve, the opening degree of the feed valve, and the opening degree of the discharge valve. In practical applications, the control parameters can be either target values ​​issued by the control unit or actual action values ​​fed back by the actuator. If both target and actual action values ​​exist, the actual action values ​​are used first; if the actual action values ​​are missing, the target values ​​are used as the control parameters for the current monitoring cycle. All control parameters are also normalized to ensure their value range matches the graph edge transition state values.

[0132] For any graph edge First, determine which control parameters affect the edge of the graph based on its edge type. Material flow edges are mainly affected by the opening of the feed valve, the opening of the discharge valve, or the frequency of the conveying pump; steam transfer edges are mainly affected by the opening of the heating steam valve, the steam pressure setpoint, or the vacuum regulation; condensate discharge edges are mainly affected by the opening of the condensate discharge valve, the condenser operating status, or the vacuum regulation; and circulation return edges are mainly affected by the circulation pump frequency and the circulation valve opening. These correspondences can be established as a control association table during the equipment commissioning phase based on the actuator's position. The control association table records the edge number, edge type, associated control parameters, and direction of action.

[0133] In one specific implementation, graph edges In the current monitoring cycle The migration prediction value can be expressed as:

[0134]

[0135] in, Represents graph edges Migration predictions This represents the transition state value of edge e in the previous cycle of the previous cycle state graph. Represents graph edges The corresponding state response coefficient, Indicates the relationship between the graph edge and the current monitoring period. Associated control increments, This means that the calculation result is restricted to a closed interval between 0 and 1.

[0136] Control increment Based on the edges of the graph The associated control parameter components are calculated. For any control parameter component, using the same normalization upper and lower limits, the control parameter component of the current monitoring period and the control parameter component corresponding to the previous monitoring period are normalized to 0 to 1 respectively, and the change of the corresponding control parameter component is obtained by subtracting the latter from the former. For example, if a graph edge is affected by two control parameters simultaneously, the changes of the two normalized control parameter components are weighted according to their corresponding action weights to obtain the control increment corresponding to the graph edge. The weights can be determined based on the degree of direct association between the actuator and the graph edge. For example, the circulation pump frequency corresponding to the circulation return edge is a direct action parameter, and its weight can be set to 0.7; the circulation valve opening is a coordinating action parameter, and its weight can be set to 0.3. If there is only one direct action parameter, then the weight of that parameter is 1. When the action direction is positive, increasing the control parameter increases the migration prediction value; when the action direction is negative, increasing the control parameter decreases the migration prediction value.

[0137] State response coefficient This parameter is used to represent the impact of the material state in the previous cycle on the effectiveness of the current control action. For the vacuum concentration process of cyclic adenosine monophosphate (cAMP) extract, the same control increment can produce different migration effects at different concentration stages; therefore, changes in control parameters should not be directly equated with changes in migration state. The state response coefficient can be determined jointly by the migration state value of the graph edge in the previous cycle and the thermo-mass state value of its adjacent graph nodes in the previous cycle. For example, when the migration state value of the circulating reflux edge in the previous cycle is within a stable range and the liquid level difference between adjacent graph nodes is within an allowable range, changes in the circulating pump frequency can effectively change the circulating migration state; when the material viscosity state reflected by adjacent graph nodes increases or the heat transfer state decreases, the migration change corresponding to the same circulating pump frequency change should be weakened. The state response coefficient can be set to a range of 0 to 1, where a value of 1 indicates that the control increment fully affects the migration prediction of the graph edge, and a value of 0 indicates that the control increment is not included in the migration prediction under the current state.

[0138] After obtaining the migration prediction values ​​for each graph edge, the thermal mass prediction values ​​for each graph node are calculated according to the connection direction of the graph edges in the thermal mass migration baseline graph. For any graph node i, its set of incoming graph edges and set of outgoing graph edges can be determined separately. The migration prediction values ​​of the incoming graph edges serve as the input influence quantity of the thermal mass state of that graph node, and the migration prediction values ​​of the outgoing graph edges serve as the output influence quantity of the thermal mass state of that graph node. For graph nodes... Any component of the thermomass state can be recursively derived in the following manner:

[0139]

[0140] in, Represents graph nodes The predicted calorific value, This represents the thermomass state value of node i in the previous cycle of the graph. This represents the cumulative input impact of the incoming graph edge on graph node i. This represents the cumulative output impact of outgoing graph edges on graph node i. and These are the inflow and outflow influence coefficients, respectively. The inflow and outflow influence coefficients are set based on the graph edge type, the migrating medium, and the relationship between the corresponding graph edge and the target graph node's thermo-mass state component. For graph edges where the migrating medium directly changes the target thermo-mass state component, the influence coefficient is set according to the contribution ratio of the migrating medium to the change in the target state component. For graph edges that couple to the target thermo-mass state component through changes in liquid level, flow rate, or pressure, the influence coefficient is set according to the contribution ratio of the corresponding coupling effect. The sum of the inflow and outflow influence coefficients corresponding to the same graph node does not exceed 1. For example, regarding the temperature prediction component of the double-effect heating zone node, the secondary steam transfer edge directly inputs heat into the double-effect heating zone, with its inward influence coefficient set to 0.6; the material flow edge entering the double-effect heating zone affects the temperature of the double-effect heating zone through the material inlet temperature and material flow rate, with its inward influence coefficient set to 0.2; the gas phase output edge of the double-effect heating zone carries away the heat generated by evaporation, with its outward influence coefficient set to 0.3; and the material output edge removes the sensible heat of the material, with its outward influence coefficient set to 0.2. Therefore, the sum of the inward influence coefficients is 0.8, and the sum of the outward influence coefficients is 0.5, both not exceeding 1.

[0141] It should be noted that the predicted calorific value can be a scalar or a vector. When a graph node corresponds to only a single calorific state, a normalized numerical value can be used. When a graph node corresponds to multiple state components such as temperature, pressure, liquid level, and evaporation state, the above recursion can be performed on each state component separately, and then the calorific value vector can be formed. For example, for a single-effect heating zone node, temperature prediction component, pressure prediction component, liquid level prediction component, and evaporation intensity prediction component can be generated separately; for a condensate discharge node, condensate temperature prediction component, condensate discharge pressure prediction component, and condensate accumulation prediction component can be generated separately.

[0142] The calculated predicted migration values ​​of each graph edge are written into the corresponding graph edge in the basic thermo-mass migration graph, and the predicted thermo-mass migration values ​​of each graph node are written into the corresponding graph node in the basic thermo-mass migration graph, thus obtaining the predicted thermo-mass migration graph for the current monitoring period. This predicted thermo-mass migration graph has the same graph nodes, graph edges, and connection directions as the thermo-mass migration graph for the current monitoring period. The difference is that the predicted thermo-mass migration graph records the theoretical state recursively derived from the state graph of the previous period and the current control parameters, while the thermo-mass migration graph records the actual state mapped from the current measured operating data. Maintaining the same topology structure for both facilitates pairing and comparison on the same graph node and the same graph edge.

[0143] Preferably, based on the control parameters, the previous cycle thermo-mass state value, and the previous cycle migration state value, the migration prediction value for each graph edge is generated, including:

[0144] S3.2.1: Extract the corresponding control parameter components from the control parameters according to the edge type of each graph edge.

[0145] S3.2.2: Calculate the control increment of the control parameter component relative to the corresponding control parameter component in the previous monitoring cycle.

[0146] S3.2.3: Generate the state response coefficients of each graph edge based on the migration state value of each graph edge in the previous cycle and the thermo-mass state value of its adjacent graph nodes in the previous cycle.

[0147] S3.2.4: The control increment and the state response coefficient are used to correct the migration state value of the corresponding graph edge in the previous cycle to obtain the migration prediction value of each graph edge.

[0148] In this embodiment, the migration prediction value of each graph edge is used to describe the theoretical migration intensity that the corresponding migration relationship should reach within the current monitoring period after the current control parameter is applied. Since different types of graph edges correspond to different migration media and controlled objects, not all control parameters can be applied uniformly to all graph edges. Instead, a correspondence between control parameters and graph edge types should be established first. This correspondence can be determined during the equipment commissioning phase based on the installation location of the actuator, the process flow, and the control logic, forming a control association table. The control association table should at least record the graph edge number, edge type, associated control parameter, direction of action, and weight of action.

[0149] For material flow edges, associated control parameters may include the opening degree of the feed valve, the opening degree of the discharge valve, or the frequency of the material conveying pump; for steam transfer edges, associated control parameters may include the opening degree of the heating steam valve, the steam pressure setpoint, or the opening degree of the vacuum regulating valve; for condensate discharge edges, associated control parameters may include the opening degree of the condensate discharge valve, the condenser cooling water regulation rate, or the opening degree of the vacuum regulating valve; for circulation return edges, associated control parameters may include the circulation pump frequency and the circulation valve opening. When a graph edge is associated with multiple control parameters simultaneously, the control parameter that directly changes the migration intensity of the graph edge is used as the primary control parameter, and the control parameter that changes the migration boundary conditions of the graph edge is used as the secondary control parameter. For example, the circulation return edge uses the circulation pump frequency as the primary control parameter and the circulation valve opening as the secondary control parameter; the condensate discharge edge uses the condensate discharge valve opening as the primary control parameter and the vacuum regulating valve opening as the secondary control parameter.

[0150] Control parameter components are sets of parameters extracted from the control parameters of the current monitoring cycle that have a control correlation with a certain graph edge. When extracting control parameter components, the edge type of the target graph edge is first determined according to the control correlation table, and then the main control parameter and auxiliary control parameter corresponding to that edge type are read. If a control parameter is expressed as a percentage opening, it is directly normalized to a range of 0 to 100%; if it is expressed as a frequency, it is normalized according to the allowable frequency range of the actuator. For example, if the allowable frequency of the circulating pump is 0Hz to 50Hz, 25Hz corresponds to a normalized value of 0.5; if it is expressed as a pressure setpoint, it is normalized according to the allowable pressure setpoint range of that process section. After normalization, different control parameters uniformly fall into the value range of 0 to 1, facilitating same-scale calculations with the migration state values ​​of the previous cycle.

[0151] The control increment reflects the control change in the current monitoring cycle relative to the previous monitoring cycle, rather than directly using the absolute value of the current control parameter. This is because the state diagram of the previous cycle already includes the transition state after the control action of the previous monitoring cycle; the current prediction only needs to superimpose the newly generated control change in the current monitoring cycle. For any control parameter component, the control increment is obtained by subtracting the normalized control parameter component of the previous monitoring cycle from the normalized control parameter component of the current monitoring cycle. If the absolute value of the control increment does not reach the dead zone, the control increment is set to 0. The dead zone can be set according to the actual response accuracy of the actuator. For example, the dead zone for valve opening can be set to a normalized value of 0.02, and the dead zone for pump frequency can be set to a normalized value of 0.01.

[0152] When multiple control parameter components exist for the same graph edge, the control increments are synthesized according to their respective weights to obtain the equivalent control increment corresponding to that graph edge. The weights are set based on the directness of the control parameter's effect on the graph edge migration intensity. The weight of the main control parameter can be set to 0.6 to 0.8, and the weight of the auxiliary control parameter can be set to 0.2 to 0.4, with the sum of the weights corresponding to the same graph edge being 1. For example, in the reflux edge, the weight of the reflux pump frequency is set to 0.7, and the weight of the reflux valve opening is set to 0.3; in the steam transfer edge, the weight of the heating steam valve opening is set to 0.6, and the weight of the vacuum regulating valve opening is set to 0.4. If the control parameter has a reverse effect on the graph edge migration intensity, the control increment takes the opposite sign during synthesis.

[0153] The state response coefficient is used to correct the actual migration effect of the same control increment under different material states. This coefficient is generated jointly by the migration state value of the target graph edge in the previous cycle and the thermo-mass state values ​​of its adjacent graph nodes in the previous cycle. Specifically, the migration state value of the target graph edge in the previous cycle's state graph is read first, followed by the thermo-mass state values ​​of the starting and ending graph nodes of that edge. For material flow edges and circulation return edges, the focus is on liquid level, pressure, and circulation resistance-related states; for steam transfer edges, the focus is on temperature, pressure, and evaporation states; for condensate discharge edges, the focus is on condensate temperature, condensate discharge pressure, and condensate accumulation states. All of the above states use the aforementioned normalized thermo-mass state values, ranging from 0 to 1.

[0154] The state response coefficient ranges from 0 to 1. A value close to 1 indicates that the current material state allows the control increment to be fully converted into a change in migration state, while a value close to 0 indicates that the current material state significantly inhibits the control effect. This coefficient can be segmented based on historical qualified batch data. For example, for the reflux edge, when the migration state value of the previous cycle is between 0.3 and 0.8, and the normalized value of the liquid level difference between adjacent graph nodes is between 0.1 and 0.4, the state response coefficient can be set to 0.8 to 1; when the migration state value of the previous cycle exceeds 0.8, or the normalized value of the pressure difference between adjacent graph nodes exceeds 0.6, it indicates that further increasing the cyclic control quantity has limited effective change in migration state, and the state response coefficient can be set to 0.3 to 0.6. For the steam transfer edge, when the thermo-mass state value of the previous cycle shows that the evaporation state of the corresponding heated node is close to the process limit, the state response coefficient can be reduced to avoid interpreting the steam control increment as an enhancement of steam migration.

[0155] After obtaining the equivalent control increment and state response coefficient, the migration state value of the corresponding graph edge in the previous cycle is corrected to form the migration prediction value of that graph edge in the current monitoring cycle. The correction direction is determined by the direction of action in the control association table, and the correction magnitude is jointly limited by the equivalent control increment and state response coefficient. The corrected migration prediction value is still limited to the range of 0 to 1; when the calculation result exceeds 1, it is taken as 1, and when the calculation result is less than 0, it is taken as 0. For graph edges that correspond to valve closure or pump stoppage in the current monitoring cycle, the migration prediction value is directly set to 0; for graph edges that are not affected by the current control parameters, the migration prediction value remains the migration state value of the previous cycle, or is inertially decayed according to the changes in the thermo-mass state of adjacent graph nodes.

[0156] Through the above processing, the predicted migration value of each graph edge simultaneously reflects the edge type, changes in control parameters, the migration state of the previous cycle, and the thermomass state of adjacent nodes. This processing method avoids mechanically equating changes in control parameters with changes in migration intensity, enabling the predicted thermomass migration map to better reflect the response differences of cyclic adenosine monophosphate extract during vacuum concentration due to changes in concentration, viscosity, and heat transfer state.

[0157] Preferably, the step of converting the thermo-mass migration map deviation between each thermo-mass migration map and the corresponding predicted thermo-mass migration map within a continuous monitoring period into a material state trajectory includes:

[0158] S4.1: Starting from the second monitoring cycle, perform graph node pairing and graph edge pairing for the thermal mass migration map and the predicted thermal mass migration map for the same monitoring cycle.

[0159] S4.2: Subtract the predicted thermal mass value of the graph node from the current thermal mass state value of the paired graph node to obtain the node deviation.

[0160] S4.3: Subtract the predicted migration value of the graph edge from the current migration state value of the paired graph edge to obtain the edge deviation.

[0161] S4.4: Based on the connection direction of each graph edge, convert the node deviation and the edge deviation into material status source items for each graph node in the corresponding monitoring period.

[0162] S4.5: Arrange the material status source items in the order of the monitoring cycle to obtain the material status trajectory.

[0163] In this embodiment, both the thermal mass transfer map and the predicted thermal mass transfer map are built on the same thermal mass transfer base map, and they have the same node numbers, edge numbers, and connection directions. Therefore, node and edge pairing does not require re-spatial search or similarity matching, but is performed one-to-one according to the node and edge numbers pre-recorded in the thermal mass transfer base map. If a node or edge has missing detection data in the current monitoring cycle, the state value of the most recent valid monitoring cycle can be used for short-term retention; if the number of consecutive missing cycles reaches a preset missing limit, the node or edge will not participate in the material state source item calculation in the current monitoring cycle. The preset missing limit can be set according to the monitoring cycle length and the production control allowable delay. For example, when the monitoring cycle is 30s, the preset missing limit can be set to 3 consecutive monitoring cycles.

[0164] After the graph nodes are paired, the current thermo-mass state value and the predicted thermo-mass state value of the same graph node are compared item by item. If the state value of the graph node is a single normalized value, the node deviation is directly obtained; if the state value of the graph node is a state vector containing multiple components such as temperature, pressure, liquid level, and evaporation state, the deviation of each component is calculated separately, and then combined according to the preset node component weights to form the node deviation. The node component weights can be set according to the degree of correlation between different state components and changes in material state. For example, for an evaporation heat exchange node, the weight of the evaporation state component can be set to 0.4, the weight of the temperature component can be set to 0.25, the weight of the pressure component can be set to 0.2, and the weight of the liquid level component can be set to 0.15; for a condensate discharge node, the weight of the condensate accumulation component can be set to 0.4, the weight of the condensate temperature component can be set to 0.3, and the weight of the condensate discharge pressure component can be set to 0.3. The sum of the weights of each component is 1.

[0165] After edge pairing is completed, the current migration status value of the same edge is compared with the predicted migration value to obtain the edge deviation. The edge deviation is a signed value; a positive value indicates that the actual migration intensity of the edge in the current monitoring period exceeds the predicted migration intensity, and a negative value indicates that the actual migration intensity does not reach the predicted migration intensity. Since the edge type and connection direction are recorded, the edge deviation not only represents the numerical difference but also the location where the difference occurs along the material flow, steam transfer, condensate discharge, or circulation return direction. For edge where valve closure or pump stoppage results in both the predicted migration value and the current migration status value being 0, the edge deviation is recorded as 0; for edge where migration is predicted but the measured migration is 0, the edge deviation is recorded as a negative value to indicate that migration is blocked or has not occurred.

[0166] The calculated results of node deviation and edge deviation can be first screened for validity to avoid interpreting minor sensor fluctuations as changes in material state. Validity screening can be performed using a combination of deviation thresholds and duration. For example, when the absolute value of node deviation is less than 0.05, it can be recorded as 0; when the absolute value of edge deviation is less than 0.03, it can also be recorded as 0. These thresholds can be determined based on the natural fluctuation range between the measured and predicted states in historical qualified batches, or based on the statistical distribution of stable operating data during the equipment commissioning phase. If a statistical method is used, the 95th percentile of the absolute value of the deviation from the stable operating data can be taken as the corresponding threshold.

[0167] The material state source term represents the state deviation at a graph node that cannot be fully explained by migration deviations from adjacent graph edges. Its physical meaning is as follows: if the current state deviation of a graph node is mainly caused by migration deviations from inbound or outbound graph edges, then the deviation is a result of migration propagation; if a persistent deviation exists after deducting the effects of migration propagation, it indicates that there is a change in the material state itself at the corresponding process location of the graph node, such as increased viscosity leading to circulation obstruction, local stagnation leading to heat transfer attenuation, or a shift in evaporation load leading to phase change state shift. Therefore, the material state source term is not a normal error value, but rather a state change quantity processed through the directional relationships of the thermo-mass migration baseline graph.

[0168] During the specific conversion, taking the target graph node as the center, the incoming and outgoing graph edges connected to the target graph node are first identified. The deviation of the incoming graph edge represents the migration change flowing into the target graph node, and the deviation of the outgoing graph edge represents the migration change leaving the target graph node. If the incoming migration deviation can explain the positive node deviation of the target graph node, then this deviation is not considered as a material state source item; if the outgoing migration deviation can explain the negative node deviation of the target graph node, then this deviation is also not considered as a material state source item. Only the remaining deviations that cannot be explained by the incoming and outgoing migrations are considered as the material state source items of the target graph node in the corresponding monitoring period.

[0169] Material state source terms retain their sign direction. Positive material state source terms indicate that the actual thermo-mass state of the target graph node shows an accumulation, enhancement, or retention trend relative to the predicted state; negative material state source terms indicate that the actual thermo-mass state of the target graph node shows a decay, release, or insufficient migration trend relative to the predicted state. The meaning of the symbols is interpreted in conjunction with the node type for different node types. For evaporative heat transfer nodes, positive source terms may correspond to actual heating or evaporation exceeding the predicted state, while negative source terms may correspond to insufficient heat transfer efficiency or insufficient evaporation intensity; for recirculation nodes, positive source terms may correspond to accumulated recirculation pressure or liquid level, while negative source terms may correspond to insufficient recirculation intensity; for condensate discharge nodes, positive source terms may correspond to accumulated condensate discharge, while negative source terms may correspond to insufficient condensate discharge.

[0170] After obtaining the material state source items for each monitoring period, a trajectory component is established for each graph node according to the monitoring period sequence. The trajectory component includes the material state source item value, sign direction, and corresponding time stamp for that graph node within the continuous monitoring period. The trajectory components of multiple graph nodes together constitute the material state trajectory. The material state trajectory can be stored using a time series table, a node trajectory matrix, or a graph state sequence with time index, as long as the one-to-one correspondence between graph nodes, monitoring periods, and material state source items can be maintained.

[0171] To facilitate continuous state identification, the material state trajectory can be maintained using a sliding window method. The length of the sliding window can be set according to the thermal inertia of the concentration process and the control response time. For example, when the monitoring cycle is 30 seconds, and the time for the double-effect vacuum thickener to reach a stable response in temperature, pressure, and condensate flow rate after a change in control parameters is approximately 3 minutes, the sliding window length can be set to 6 monitoring cycles. After generating a material state source item for each new monitoring cycle, it is added to the sliding window, and the earliest monitoring cycle data in the window is removed, thereby forming a material state trajectory that can be used for material state determination at the current moment.

[0172] Preferably, according to the connection direction of each graph edge, the node deviation and the edge deviation are converted into material status source items for each graph node in the corresponding monitoring period, including:

[0173] S4.4.1: Taking any graph node as the target graph node, divide the graph edges connected to the target graph node into incoming graph edges and outgoing graph edges according to the connection direction.

[0174] S4.4.2: Summarize the edge deviation of the incoming graph edge and the edge deviation of the outgoing graph edge respectively to obtain the incoming migration deviation and the outgoing migration deviation.

[0175] S4.4.3: The difference between the inbound migration deviation and the outbound migration deviation is taken as the net migration deviation of the target graph node.

[0176] S4.4.4: Subtract the net migration deviation from the node deviation of the target graph node to obtain the material status source item of the target graph node in the corresponding monitoring period.

[0177] In this embodiment, the calculation object of the material state source item is each graph node in the thermal mass transfer (TMT) baseline diagram. During each calculation, a graph node is selected as the target graph node. All graph edges directly connected to the target graph node in the TMT baseline diagram are read, and the direction of the graph edge is determined based on the pre-recorded starting and ending graph nodes. Graph edges whose ending point is the target graph node are classified as inbound graph edges, and those whose starting point is the target graph node are classified as outbound graph edges. If a certain actual pipeline may have bidirectional flow under different operating conditions, two graph edges with opposite directions can be pre-established in the TMT baseline diagram, and effective graph edges can be selected for calculation based on the current valve status, pump operating status, or flow direction.

[0178] The distinction between inbound and outbound edges is used to differentiate the direction of the impact of migration deviations on target nodes. The deviation of an inbound edge represents the difference between the actual migration state and the predicted migration state of materials, steam, condensate, or recirculation entering the target node; the deviation of an outbound edge represents the difference between the migration state and the predicted migration state leaving the target node. Regarding the sign of the deviation, a positive value indicates that the actual migration state exceeds the predicted migration state, while a negative value indicates that the actual migration state does not reach the predicted migration state. Therefore, a positive deviation of an inbound edge typically corresponds to an increase in the input to the target node, and a positive deviation of an outbound edge typically corresponds to an increase in the output of the target node.

[0179] Before summarizing the edge deviations, the edge deviations can be scaled according to the edge type. Since material flow edges, steam transfer edges, condensate discharge edges, and circulation return edges reflect different migration media, although their edge deviations have all been normalized, their impact on the thermo-mass state of the target graph node is not entirely the same. Therefore, edge influence coefficients can be set for different edge types, with values ​​ranging from 0 to 1. For example, for an evaporative heat exchange node, the influence coefficient for the steam transfer edge directly connected to the node can be set to 0.4 to 0.6, the influence coefficient for the material flow edge can be set to 0.2 to 0.4, and the influence coefficient for the condensate discharge edge can be set to 0.1 to 0.3. The edge influence coefficients can be determined based on equipment commissioning data, qualified batch operation data, or the degree of influence confirmed by process personnel. The sum of the inbound edge influence coefficients corresponding to the same target graph node should not exceed 1, and the sum of the outbound edge influence coefficients should not exceed 1.

[0180] The inbound migration deviation is obtained by summing the effective edge deviations of all inbound edges of the target graph node. Specifically, for any inbound edge, the current migration state value of the edge is first subtracted from the corresponding migration prediction value to obtain an edge deviation with positive and negative directions; then, the absolute value of the edge deviation is compared with the corresponding edge deviation threshold. When the absolute value of the edge deviation does not reach the edge deviation threshold, the effective edge deviation of the edge is recorded as 0; when the absolute value of the edge deviation reaches the edge deviation threshold, the edge deviation is multiplied by the corresponding edge influence coefficient to obtain the effective edge deviation of the edge. The effective edge deviation retains the positive or negative sign of the original edge deviation. The inbound migration deviation is obtained by algebraically summing the effective edge deviations of all inbound edges of the target graph node. The outbound migration deviation is obtained by summing the effective edge deviations of all outbound edges in the same way. If the target graph node has no inbound edges, the inbound migration deviation is recorded as 0; if there are no outbound edges, the outbound migration deviation is recorded as 0.

[0181] Net migration deviation represents the net input impact of adjacent graph edge migration deviations on the target graph node. When the inward migration deviation exceeds the outward migration deviation, it indicates that the external migration relationship creates a net input to the target graph node; when the outward migration deviation exceeds the inward migration deviation, it indicates that the external migration relationship creates a net output to the target graph node. This net migration deviation is a state change caused by graph edge propagation relationships and is not directly considered as a material state change of the target graph node itself.

[0182] The node deviation of a target graph node contains two components: one is the deviation caused by the migration and propagation of adjacent graph edges, and the other is the deviation caused by the material state change of the corresponding process location of the target graph node itself. To isolate the migration and propagation effects from the node deviation, the net migration deviation is subtracted from the node deviation of the target graph node. The remaining deviation after subtraction is used as the material state source term for that target graph node in the corresponding monitoring period. This material state source term indicates that there are state changes at the target graph node that cannot be explained by inbound and outbound migration.

[0183] Material state source terms retain their positive and negative directions. For evaporation heat exchange nodes, a positive material state source term indicates a cumulative trend in the heating, evaporation, or material retention state at that location relative to the predicted state; a negative material state source term indicates that the actual evaporation or heat exchange response has not reached the predicted state. For recirculation nodes, a positive material state source term indicates a cumulative trend in recirculation-related pressure, level, or recirculation resistance; a negative material state source term indicates insufficient actual recirculation. For condensate discharge nodes, a positive material state source term indicates a cumulative increase or discharge retention at the condensate discharge end; a negative material state source term indicates that condensate generation or discharge has not reached the predicted state.

[0184] For example, if the node deviation of a certain double-effect evaporative heat exchange node is positive, and the edge deviation of its inlet steam transfer edge is also positive, then the positive deviation of the node may partly originate from the secondary steam input exceeding the predicted value. Only when a positive residual deviation remains after deducting this inlet migration deviation is the remaining portion recorded as the material state source term of the double-effect evaporative heat exchange node. If this material state source term remains in the same direction within a continuous monitoring period, it can be further used to determine whether there is a continuous material state evolution at that location, rather than an instantaneous deviation caused by a single control action or migration fluctuation.

[0185] The above processing ensures that node deviations are not directly interpreted as material anomalies. Instead, it first deducts the portion that can be explained by migration relationships according to the graph edge connection direction, and then retains the remaining portion related to the material state of the target graph node itself. The resulting material state source term can more accurately reflect the local state changes of the cyclic adenosine monophosphate extract during the double-effect vacuum concentration process.

[0186] Preferably, determining the current material state based on the material state trajectory includes:

[0187] S5.1.1: Extract the trajectory components of each graph node within the continuous monitoring period from the material state trajectory, and use the trajectory components as the material state source items of the corresponding graph node.

[0188] S5.1.2: Calculate the sign consistency, duration period number and change slope of the material state source item corresponding to the same graph node to obtain the state duration characteristics of the corresponding graph node.

[0189] S5.1.3: Select the dominant state node from each graph node according to the state persistence characteristics.

[0190] S5.1.4: Determine the current material state based on the path position of the dominant state node in the thermal mass transfer baseline diagram and the change direction of the material state source item corresponding to the dominant state node.

[0191] In this embodiment, the material state trajectory is not an instantaneous deviation of a single monitoring cycle, but rather a continuous sequence of changes in the material state source items of each graph node over time. To avoid misjudgments caused by single sampling disturbances, instantaneous valve actions, or sensor fluctuations, the current material state is determined based on the trajectory changes within the continuous monitoring window. The length of the continuous monitoring window can be set according to the thermal inertia of the double-effect vacuum concentrator and the response time of the actuator. For example, when the monitoring cycle is 30 seconds and the time for steam regulation, vacuum regulation, and circulation regulation to have a stable impact on the thermo-mass state is 3 minutes, the continuous monitoring window can be set to 6 monitoring cycles.

[0192] For each graph node, its trajectory component within the continuous monitoring window is extracted from the material state trajectory. The trajectory component includes the material state source item, source item symbol, and time sequence corresponding to each monitoring cycle within the window for that graph node. When the absolute value of a material state source item does not reach the effective source item threshold, the source item corresponding to that cycle is not included in the calculation of symbol consistency and the number of consecutive cycles. The effective source item threshold can be set based on the natural fluctuation range of material state source items in a stable operating batch. For example, the 95th percentile of the absolute value of source items in a stable operating batch can be used. In the initial debugging phase where historical data is lacking, the effective source item threshold can be set to 0.05, and corrected after accumulating qualified batch data.

[0193] Sign consistency is used to determine whether material state source items of the same graph node maintain the same direction of change within a continuous monitoring window. If the number of positive source items reaching the valid threshold within the continuous monitoring window reaches a preset consistency number, then the graph node has positive sign consistency; if the number of negative source items reaches the preset consistency number, then the graph node has negative sign consistency. For example, when the continuous monitoring window is 6 monitoring cycles, the preset consistency number can be set to 4 monitoring cycles, and it is required that there are no valid source items with opposite signs for two consecutive monitoring cycles. This rule can eliminate the influence of a single reverse fluctuation on the determination of the state direction.

[0194] The duration of a graph node is used to represent the length of time a valid material state source item exists continuously within the same graph node. For any graph node, the duration of the node is calculated by tracing back from the current monitoring period and counting the number of monitoring periods that consecutively reach the valid source item threshold with the same sign direction. If a monitoring period fails to reach the valid source item threshold, but the number of invalid periods does not exceed one, it can be considered a short-term sampling interruption, and the duration of the node is not interrupted. If two consecutive invalid periods or valid source items with opposite signs appear, the duration of the node is recounted. For example, when the duration of the node reaches three monitoring periods, it can be considered that the node has experienced a confirmed continuous state change.

[0195] The slope of change indicates the strengthening or weakening trend of a material state source term at the same graph node within a continuous monitoring window. Specifically, the changes in source term intensity can be compared sequentially at the beginning, middle, and end of the window. If the absolute value of the source term in two consecutive monitoring cycles at the end of the window is greater than that in two consecutive monitoring cycles at the beginning of the window, the slope of the source term change at that graph node is determined to be an strengthening trend; if the absolute value of the source term in two consecutive monitoring cycles at the end of the window is less than that in two consecutive monitoring cycles at the beginning of the window, it is determined to be a weakening trend; otherwise, it is determined to be a maintaining trend. Source term intensity indicates the absolute value of the material state source term; the slope of change does not change the direction of the source term sign.

[0196] The state persistence characteristic is composed of sign consistency, number of consecutive effective periods, and slope of change. Sign consistency is used to determine the direction of material state change, the number of consecutive periods is used to determine whether the change has temporal continuity, and the slope of change is used to determine whether the change is still developing. For each graph node, the sign direction, number of consecutive effective periods, and whether the trend is increasing or decreasing can be recorded as the state persistence characteristic of that node. This characteristic is not directly equivalent to anomaly detection, but is used to identify the node that best represents the source of the current concentration state change among multiple graph nodes.

[0197] The dominant state node is selected from graph nodes with valid state persistence characteristics. During selection, graph nodes that meet the symbol consistency requirement and whose persistence period reaches a preset persistence threshold are given priority. When multiple graph nodes meet the requirements, they are sorted according to source term strength, persistence period, and trend. For example, the cumulative results of source term strength within the continuous monitoring window can be compared first, with the graph node having the largest cumulative result given priority. If the cumulative results are the same, the graph node with more persistence periods is given priority. If the persistence period is the same, graph nodes with an increasing trend are given priority over those maintaining a trend, and those maintaining a trend are given priority over those with a decreasing trend. If two adjacent graph nodes simultaneously meet the dominant condition and are located at consecutive positions on the same material flow path, the graph node closer to the source of thermal mass migration can be designated as the dominant state node to determine the control tracing direction.

[0198] The current material state is determined based on the path position of the dominant state node in the thermo-mass transfer diagram and the direction of change of the material state source term. The path position indicates whether the state change occurs in the material flow path, steam transfer path, condensate discharge path, or circulation return path; the direction of change indicates whether the actual state at that location is cumulative, enhanced, diminished, or insufficient relative to the predicted state. The combination of these two factors forms the basis for determining the current material state.

[0199] For example, when the dominant state node is located at the evaporation heat exchange node and the material state source term is negative and continuously increasing, the current material state can be determined to be insufficient heat transfer response or insufficient evaporation intensity; when the dominant state node is located at the circulation reflux node and the material state source term is positive and continuously increasing, the current material state can be determined to be enhanced circulation retention; when the dominant state node is located at the condensation discharge node and the material state source term is negative and persistent, the current material state can be determined to be insufficient condensation discharge; when the dominant state node is located at the material residence node before discharge and the material state source term is positive and maintained within the continuous monitoring window, the current material state can be determined to be approaching the concentration endpoint or enhanced material accumulation. The above state types can be named according to the process path and detection point settings of the actual double-effect vacuum concentrator unit, and the determination basis is still the path position of the dominant state node and the direction of change of the material state source term.

[0200] Preferably, the generation of a reduced-pressure concentration control strategy corresponding to the current material state includes:

[0201] S5.2.1: Obtain the dominant state node and material state source item corresponding to the current material state.

[0202] S5.2.2: Perform reverse tracing along the graph edges connected to the dominant state node in the basic thermal mass migration graph to obtain candidate controlled graph edges that have migration associations with the dominant state node.

[0203] S5.2.3: Generate the control priority of the candidate controlled graph edge based on the path distance, edge type, and historical adjustment status between the candidate controlled graph edge and the dominant state node.

[0204] S5.2.4: According to the control priority, the intensity and change slope of the material state source item are converted into the control adjustment amount of each candidate controlled graph edge.

[0205] S5.2.5: Generate the decompression and concentration control strategy based on the control adjustment amount of each candidate controlled graph edge.

[0206] In this embodiment, the reduced pressure concentration control strategy is not directly triggered by the current material state name. Instead, it first determines the dominant state node and material state source term corresponding to the current material state, and then searches the thermo-mass migration baseline diagram for controlled migration relationships that can regulate the dominant state node. The dominant state node indicates the process location where the current state change is concentrated, and the material state source term indicates the state deviation direction, source term intensity, and change trend at that location.

[0207] The determination of candidate controlled graph edges is based on the connection direction in the thermo-mass transfer baseline graph. Starting from the dominant state node, a reverse tracing is performed in the direction opposite to the formation cause of the material state source item to find graph edges that have a direct or indirect migration association with the dominant state node. If a graph edge is associated with an executable control parameter, such as the opening degree of a steam valve, the opening degree of a vacuum regulating valve, the frequency of a circulating pump, the opening degree of a condensate discharge valve, the opening degree of a feed valve, or the opening degree of a discharge valve, then that graph edge can be considered a candidate controlled graph edge. If a graph edge is only used for state transmission and does not have an adjustable actuator, then that graph edge is only used as a tracing path and is not considered a candidate controlled graph edge.

[0208] The scope of reverse tracing can be limited by path distance. Path distance refers to the number of graph edges traversed between a candidate controlled graph edge and the dominant state node. For example, the path distance can be limited to no more than 3; a candidate controlled graph edge with a distance of 1 indicates that it directly connects to the dominant state node, and a candidate controlled graph edge with a distance of 2 or 3 indicates that it has a migration effect on the dominant state node through intermediate nodes. If no controlled graph edge is found after exceeding the preset path distance, upstream controlled graph edges of the same type as the path where the dominant state node is located can be used as candidate objects.

[0209] Control priority is determined by path distance, edge type, and historical adjustment status. A shorter path distance indicates a more direct adjustment transmission link between the candidate controlled graph edge and the dominant state node, resulting in a higher priority. Edge type is used to determine whether the control action matches the current material state. For example, if the dominant state node is located on a circulation return path and the source term exhibits enhanced circulation retention, the circulation return edge takes precedence over the steam transfer edge; if the dominant state node is located on a condensate discharge path and the source term exhibits insufficient condensate discharge, the condensate discharge edge and vacuum-related steam transfer edge take precedence over the material flow edge. Historical adjustment status is used to prevent the same actuator from repeatedly making large adjustments within consecutive monitoring cycles. If a candidate controlled graph edge has reached its set adjustment limit within the last two monitoring cycles, its control priority is reduced.

[0210] The strength and slope of the material state source term are used to determine the control adjustment amount. The source term strength can be divided into three adjustment levels according to its normalized absolute value: the first adjustment level is when the source term strength reaches 0.05 and is less than 0.15; the second adjustment level is when the source term strength reaches 0.15 and is less than 0.30; and the third adjustment level is when the source term strength reaches 0.30. When the slope of change is an increasing trend, the upper limit adjustment range is selected within the corresponding adjustment level; when the slope of change is a maintaining trend, the middle adjustment range is selected; and when the slope of change is a decreasing trend, the lower limit adjustment range is selected. The single control adjustment amount can be limited to 1% to 5% of the allowable range of the corresponding control parameter, with the specific range set according to the actuator response accuracy and process safety range.

[0211] For different candidate controlled graph edges, the control adjustment direction is determined according to the direction of the material state source term and the edge type. If a positive and continuously increasing material state source term appears in the circulation reflux node, the circulation pump frequency or circulation valve opening corresponding to the circulation reflux edge can be increased; if a negative and continuously increasing material state source term appears in the evaporation heat exchange node, the heating steam valve opening corresponding to the steam transfer edge can be increased, or the vacuum regulating valve opening can be adjusted to restore evaporation conditions; if a negative and persistent material state source term appears in the condensate discharge node, the discharge valve opening corresponding to the condensate discharge edge can be increased, or the relevant control quantities of the condenser can be adjusted; if a positive and sustained material state source term appears in the material residence node before discharge, the relevant control quantities of the feed can be reduced or a discharge preparation control quantity can be generated.

[0212] When multiple candidate controlled graph edges require adjustment simultaneously, control adjustment amounts are allocated sequentially according to control priority. The candidate controlled graph edge with the highest priority first receives the primary adjustment amount corresponding to the dominant source term; the remaining candidate controlled graph edges receive auxiliary adjustment amounts based on the strength of the remaining source terms and their own priorities. If the control directions corresponding to two candidate controlled graph edges conflict, such as simultaneously requiring an increase in steam input and a decrease in evaporation load, the candidate controlled graph edge corresponding to the path of the dominant state node is given priority, and the control adjustment amount of the conflicting graph edge is limited to 0.

[0213] The generated vacuum concentration control strategy can include target control parameters, control adjustment direction, control adjustment amount, and execution sequence. The target control parameter corresponds to a specific actuator; the control adjustment direction includes increasing, decreasing, or maintaining; the control adjustment amount is expressed using normalized amplitude or engineering amplitude; and the execution sequence is determined according to control priority. To ensure smooth control action, the control adjustment amount can be limited by amplitude, rate of change restriction, and dead zone processing before issuance. For example, the single adjustment amplitude of valve opening is limited to 1% to 3%, and the single adjustment amplitude of circulating pump frequency is limited to 1Hz to 3Hz; when the calculated adjustment amount does not reach the actuator's dead zone, the control parameter remains unchanged. Thus, the vacuum concentration control strategy can backtrack from the material state trajectory to an adjustable migration path, and then convert the adjustable migration path into specific control parameters, avoiding empirical adjustments based solely on single-point temperature, vacuum degree, or liquid level exceeding limits.

[0214] Example 2

[0215] Please see Figure 3 One embodiment of the present invention provides a reduced-pressure concentration control system for cyclic adenosine extraction, the system comprising:

[0216] The basic graph construction module is used to obtain the process connection relationship of the double-effect vacuum concentrator, determine the graph nodes and graph edges based on the process connection relationship, and construct the basic graph of heat and mass transfer.

[0217] The migration map generation module is used to acquire the current measured operating data in each monitoring cycle and generate a thermo-mass migration map based on the current measured operating data and the thermo-mass migration baseline map.

[0218] The prediction map generation module is used to obtain the control parameters of the current monitoring cycle starting from the second monitoring cycle, use the heat mass migration map generated in the previous monitoring cycle as the state map of the previous cycle, and generate a prediction heat mass migration map based on the state map of the previous cycle and the control parameters.

[0219] The trajectory conversion module is used to convert the thermal mass migration map and the corresponding predicted thermal mass migration map within a continuous monitoring period into a material state trajectory.

[0220] The control update module is used to determine the current material state based on the material state trajectory, generate a depressurization concentration control strategy corresponding to the current material state, and use the thermo-mass migration diagram corresponding to the current monitoring cycle as the previous cycle state diagram for the next monitoring cycle.

[0221] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for controlled vacuum concentration in the extraction of cyclic adenosine monophosphate, characterized in that, The method includes: Obtain the process connection relationship of the double-effect vacuum concentrator, determine the graph nodes and graph edges based on the process connection relationship, and construct a basic graph of heat and mass transfer. In each monitoring cycle, the current measured operating data is acquired, and a heat mass transfer map is generated based on the current measured operating data and the heat mass transfer baseline map; Starting from the second monitoring cycle, the control parameters of the current monitoring cycle are obtained, and the heat mass migration map generated in the previous monitoring cycle is used as the state map of the previous cycle. Based on the state map of the previous cycle and the control parameters, a predicted heat mass migration map is generated. The thermal mass state deviation between each thermal mass migration map and the corresponding predicted thermal mass migration map within a continuous monitoring period is converted into a material state trajectory. The current material state is determined based on the material state trajectory, a reduced pressure concentration control strategy corresponding to the current material state is generated, and the thermo-mass migration diagram corresponding to the current monitoring cycle is used as the previous cycle state diagram for the next monitoring cycle.

2. The method for controlled vacuum concentration of cyclic adenosine extract according to claim 1, characterized in that, Based on the aforementioned process connection relationships, graph nodes and edges are determined, and a basic graph of heat and mass transfer is constructed, including: Obtain the material flow path, steam transfer path, condensate discharge path, and circulation return path in the double-effect vacuum concentrator unit; The process locations in the material flow path where material residence, evaporation, heat exchange, and circulation state changes are determined as graph nodes; The material flow relationship, steam transfer relationship, condensate discharge relationship and circulation return relationship between adjacent graph nodes are defined as graph edges; Connect the nodes and edges of the graph according to the connection sequence corresponding to the material flow path, steam transfer path, condensate discharge path and circulation return path to obtain the basic graph of heat and mass transfer.

3. The method for controlled vacuum concentration of cyclic adenosine extract according to claim 1, characterized in that, Generate a heat-mass transfer map based on the current measured operating data and the basic heat-mass transfer map, including: Acquire the operating data of the first effect, the second effect, the condensate, and the circulation data for the current monitoring cycle; The current thermo-mass state value of each node in the diagram is determined based on the first-effect operation data and the second-effect operation data; The current migration state value of each graph edge is determined based on the condensate data and the cycle operation data; The current thermo-mass state value and the current migration state value are mapped to the corresponding graph nodes and graph edges in the thermo-mass migration base graph to obtain the thermo-mass migration graph for the current monitoring period.

4. The method for controlled vacuum concentration of cyclic adenosine extract according to claim 1, characterized in that, Based on the previous cycle state diagram and the control parameters, a predicted thermo-mass migration diagram is generated, including: Extract the previous cycle thermo-mass state value of each graph node and the previous cycle migration state value of each graph edge from the previous cycle state diagram. Based on the control parameters, the previous cycle thermo-mass state value, and the previous cycle migration state value, a migration prediction value for each graph edge is generated. Along the connection direction of each graph edge, the migration prediction value is accumulated to the adjacent graph node to obtain the thermal mass prediction value of each graph node; The predicted thermal mass and the predicted migration values ​​are mapped to the corresponding graph nodes and edges in the basic thermal mass migration graph to obtain the predicted thermal mass migration graph.

5. The method for controlled vacuum concentration of cyclic adenosine extract according to claim 4, characterized in that, Based on the control parameters, the previous cycle thermo-mass state value, and the previous cycle migration state value, a migration prediction value for each graph edge is generated, including: According to the edge type of each graph edge, extract the corresponding control parameter components from the control parameters; Calculate the control increment of the control parameter component relative to the corresponding control parameter component in the previous monitoring cycle; The state response coefficients of each graph edge are generated based on the migration state value of each graph edge in the previous cycle and the thermo-mass state value of its adjacent graph nodes in the previous cycle. The control increment and the state response coefficient are used to correct the migration state value of the corresponding graph edge in the previous cycle, so as to obtain the migration prediction value of each graph edge.

6. The method for controlled vacuum concentration of cyclic adenosine extract according to claim 1, characterized in that, The step of converting the thermal mass migration map deviation between each thermal mass migration map and the corresponding predicted thermal mass migration map within a continuous monitoring period into a material state trajectory includes: Starting from the second monitoring period, the thermal mass migration map and the predicted thermal mass migration map for the same monitoring period will be paired with graph nodes and graph edges. The node deviation is obtained by subtracting the predicted thermal mass value of the graph node from the current thermal mass state value of the paired graph node. The edge deviation is obtained by subtracting the predicted edge migration value from the current migration state value of the paired edge. According to the connection direction of each graph edge, the node deviation and the edge deviation are converted into material status source items of each graph node in the corresponding monitoring period; The material status source items are arranged in order of monitoring cycle to obtain the material status trajectory.

7. The method for controlled vacuum concentration of cyclic adenosine extract according to claim 6, characterized in that, Based on the connection direction of each graph edge, the node deviation and the edge deviation are converted into material status source items for each graph node in the corresponding monitoring period, including: Taking any graph node as the target graph node, the graph edges connected to the target graph node are divided into inbound graph edges and outbound graph edges according to the connection direction. The edge deviations of the incoming graph edges and the outgoing graph edges are summarized separately to obtain the incoming migration deviation and the outgoing migration deviation. The difference between the inbound migration deviation and the outbound migration deviation is taken as the net migration deviation of the target graph node; The net migration deviation is subtracted from the node deviation of the target graph node to obtain the material status source item of the target graph node in the corresponding monitoring period.

8. The method for controlled vacuum concentration of cyclic adenosine extract according to claim 1, characterized in that, Determining the current material state based on the material state trajectory includes: The trajectory components of each graph node within the continuous monitoring period are extracted from the material state trajectory, and the trajectory components are used as the material state source items of the corresponding graph node. For the material state source item corresponding to the same graph node, the sign consistency, duration period number and change slope are calculated to obtain the state duration characteristics of the corresponding graph node. Dominant state nodes are selected from each graph node according to the described state persistence characteristics; The current material state is determined based on the path position of the dominant state node in the thermal mass transfer baseline diagram and the change direction of the material state source item corresponding to the dominant state node.

9. A method for controlled vacuum concentration in cyclic adenosine extraction according to claim 8, characterized in that, The generation of a reduced-pressure concentration control strategy corresponding to the current material state includes: Obtain the dominant state node and material state source item corresponding to the current material state; By tracing back along the graph edges connected to the dominant state node in the basic thermal mass transfer graph, candidate controlled graph edges that have a migration association with the dominant state node are obtained. Based on the path distance, edge type, and historical adjustment status between the candidate controlled graph edge and the dominant state node, the control priority of the candidate controlled graph edge is generated; According to the control priority, the intensity and change slope of the material state source item are converted into the control adjustment amount of each candidate controlled graph edge; The decompression and concentration control strategy is generated based on the control adjustment amount of each candidate controlled graph edge.

10. A reduced-pressure concentration control system for cyclic adenosine extraction, used to implement the reduced-pressure concentration control method for cyclic adenosine extraction as described in any one of claims 1 to 9, characterized in that, The system includes: The basic graph construction module is used to obtain the process connection relationship of the double-effect vacuum concentrator, determine the graph nodes and graph edges based on the process connection relationship, and construct the basic graph of heat and mass transfer. The migration map generation module is used to acquire the current measured operating data in each monitoring cycle and generate a thermo-mass migration map based on the current measured operating data and the thermo-mass migration baseline map. The prediction map generation module is used to obtain the control parameters of the current monitoring cycle starting from the second monitoring cycle, use the heat mass migration map generated in the previous monitoring cycle as the state map of the previous cycle, and generate a prediction heat mass migration map based on the state map of the previous cycle and the control parameters. The trajectory conversion module is used to convert the thermal mass migration map and the corresponding predicted thermal mass migration map within a continuous monitoring period into a material state trajectory. The control update module is used to determine the current material state based on the material state trajectory, generate a depressurization concentration control strategy corresponding to the current material state, and use the thermo-mass migration diagram corresponding to the current monitoring cycle as the previous cycle state diagram for the next monitoring cycle.