Vacuum heat treatment process continuous control method and system based on process phase identification

Abstract

This invention provides a continuous control method and system for vacuum heat treatment processes based on process phase identification. The method includes: real-time data acquisition; identification of the current process phase based on the data; determination of monitoring data and corresponding thresholds based on the current process phase; calculation of the current warning level based on the deviation between the monitoring data and the thresholds; and calculation of a health index based on the status data of each heating component. A basic decision-making strategy is obtained based on the current process phase, warning level, and decision matrix; the power distribution of each phase heater is dynamically adjusted based on the health index to generate a control command vector; the control command vector is decomposed into specific execution actions and output; and the execution effect is monitored in real-time, feedback data is collected, and the parameters of the decision matrix are updated based on the feedback data. By identifying the current process phase in real-time and using this as a contextual reference to dynamically determine monitoring parameters and thresholds, the shortcomings of existing technologies, such as isolated and flat warning signals and a lack of traceability, are improved.

Description

Technical Field

[0001] This invention relates to the field of industrial process control and automation technology, and in particular to a continuous control method and system for vacuum heat treatment processes based on process phase recognition. Background Technology

[0002] Vacuum heat treatment is a key process for improving the comprehensive performance of metallic materials, and its process control quality directly affects product consistency and yield. Existing vacuum heat treatment control technologies mainly fall into two categories. The first is a passive monitoring system. This type of system collects parameters such as vacuum level and temperature and compares them with preset thresholds, triggering an alarm when an anomaly is detected. Essentially, this type of system is a process status monitoring and fault alarm device. Its warning signals are isolated and flat, lacking hierarchical assessment and traceability capabilities related to the process context. It lacks the ability to autonomously intervene in the process flow, and all anomaly handling relies entirely on manual response. Especially when dealing with dynamic disturbances such as material venting and gradual changes in equipment performance, the system can only issue alarms after the fact, unable to proactively adjust control strategies, making it difficult to effectively prevent process interruptions or product defects. The second type is a fixed-program execution system. Traditional vacuum furnaces mostly use fixed-program control based on time or temperature nodes. This type of system lacks flexibility; once preset conditions cannot be met in time due to interference, the entire process will stagnate or trigger a shutdown alarm. Furthermore, its manual and automatic modes are disconnected; switching modes requires stopping the current process and re-initializing, leading to production interruptions and severely disrupting process continuity.

[0003] In summary, the existing technology has significant shortcomings: it lacks a control system that can understand the process context, classify and autonomously respond to anomalies, and make forward-looking decisions by integrating equipment health status. As a result, the existing system is unable to ensure the continuous and stable execution of the process mainline in complex dynamic environments. Summary of the Invention

[0004] The purpose of this invention is to solve the technical problems in the prior art where vacuum heat treatment control systems cannot understand the process context, lack the ability to autonomously respond to anomalies and make forward-looking decisions on equipment health status, and thus cannot ensure the continuous and stable execution of the main process line in complex dynamic environments.

[0005] A first aspect of the present invention provides a method for continuous control of a vacuum heat treatment process based on process phase identification, comprising the following steps: Real-time data acquisition; identification of the current process phase based on the data; determination of monitoring data and corresponding thresholds based on the current process phase; calculation of the current warning level based on the deviation between the monitoring data and the thresholds; and calculation of the health index based on the status data of each heating component. The basic decision-making strategy is obtained based on the current process phase, the early warning level, and the decision matrix, and the power distribution of each phase heater is dynamically adjusted according to the health index to generate a control command vector. The control command vector is decomposed into specific execution actions and output; at the same time, the execution effect is monitored in real time, feedback data is collected, and the parameters of the decision matrix are updated based on the feedback data.

[0006] Furthermore, identifying the current process phase includes: extracting features from the data, matching the extracted features with a pre-stored process phase feature library, and determining the current process phase.

[0007] Furthermore, the same amount of detection data deviation corresponds to different warning levels under different process phases.

[0008] Furthermore, the warning level includes at least two levels; When the warning level is upgraded, the corresponding conservative sub-phase execution control is switched; When the warning level is upgraded to the highest level, the heating operation is paused, and the system switches to the constant temperature and pressure holding phase to maintain the current temperature.

[0009] Furthermore, the calculation of the health index includes: Obtain the health index of each heating component, and determine the health index range based on the numerical range of the health index.

[0010] Furthermore, when the health index of any heating component is lower than a preset threshold, a protective derating factor is applied to all heating components simultaneously to ensure that the power ratio of each heating zone is the same.

[0011] Furthermore, based on the health index, three intervals are defined, and corresponding power protection measures are implemented: When HI(t) ≥ 0.8, it is considered to be in a healthy state and there are no operational restrictions. When 0.5 ≤ HI(t) < 0.8, it is judged as mild aging, and a derating factor is automatically applied; When HI(t) < 0.5, it is judged as severely aged, an L3 maintenance alarm is generated, the component is marked as to be replaced in automatic mode, and its use is restricted or isolated.

[0012] Furthermore, the acquisition of the basic decision-making strategy includes acquiring the state index: ; Where P represents different process phases, L represents the warning level, and H represents the health index range; Establish a decision matrix, and retrieve the decision parameter vector by querying the decision matrix through the state index: Among them, a s For action type, p s For power distribution reference adjustment, k s For artificial intervention markers, t s This is a flag for switching output modes.

[0013] Furthermore, the dynamic adjustment of the power distribution of each phase heater based on the health index includes: Calculate the basic heating rate coefficient: ; Wherein, the influence function of vacuum change is: ; ΔP ratio Let ΔP be the current rate of change of vacuum. th The threshold for the rate of change of vacuum corresponding to the current phase; The temporary power of each phase heater is normalized to obtain the actual power distribution vector: ; Among them, P act P is the normalized actual power allocation vector; temp P represents the temporary power vector of each phase heater before normalization. temp，i This represents the temporary power value of the heater in phase i.

[0014] A second aspect of the present invention provides a continuous control system for a vacuum heat treatment process based on process phase recognition, employing the continuous control method for a vacuum heat treatment process based on process phase recognition as described in any of the preceding claims, comprising: The sensing and fusion module collects data in real time, identifies the current process phase based on the data, determines the monitoring data and corresponding thresholds based on the current process phase, calculates the current warning level based on the deviation between the monitoring data and the thresholds, and calculates the health index based on the status data of each heating component. The intelligent decision-making module obtains a basic decision-making strategy based on the current process phase, the early warning level, and the decision matrix, and dynamically adjusts the power distribution of each phase heater based on the health index to generate a control command vector. The control execution module is used to decompose the control command vector into specific execution actions and output them, while collecting feedback data and updating the parameters of the decision matrix based on the feedback data.

[0015] Compared to existing technologies, this invention offers at least the following advantages: By identifying the current process phase in real time and using this as a contextual benchmark to dynamically determine monitoring parameters and thresholds, the assessment of the early warning level is deeply correlated with the process progress, overcoming the shortcomings of isolated and flat early warning signals and lack of traceability in existing technologies; by inputting the early warning level and equipment health index into the decision matrix, the system can autonomously generate hierarchical control commands and dynamically adjust the power distribution of heaters in each phase when an anomaly occurs, achieving autonomous response and forward-looking decision-making without relying on manual intervention, effectively avoiding the problem of fixed-program execution systems stalling or shutting down due to rigid conditions not being met; through continuous collection of execution feedback data and real-time updates of decision matrix parameters, the system forms a closed-loop self-learning capability, enabling the control strategy to be continuously optimized according to the dynamic changes in equipment status and process environment, thereby ensuring the continuous and stable execution of the main process line under complex interference conditions such as material venting and gradual changes in equipment performance, and improving product consistency and pass rate. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained as provided without creative effort.

[0017] Figure 1 This is a flowchart of a continuous control method for vacuum heat treatment process based on process phase recognition in one embodiment of the present invention; Figure 2 This is a schematic diagram of a module of a continuous control system for vacuum heat treatment based on process phase recognition in one embodiment of the present invention. Detailed Implementation

[0018] The present invention will now be described in more detail with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. It should be understood that those skilled in the art can modify the invention described herein while still achieving its advantageous effects. Therefore, the following description should be understood as being broadly known to those skilled in the art and is not intended to limit the invention.

[0019] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0020] The invention is described more specifically by way of example in the following paragraphs with reference to the accompanying drawings. The advantages and features of the invention will become clearer as explained below. It should be noted that the drawings are in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the invention.

[0021] For ease of understanding, the following explains some key terms in this embodiment: Real-time acquired data refers to the real-time operating parameters obtained from different types of sensors during the vacuum heat treatment process, such as temperature sensors, vacuum sensors, and current sensors. These data collectively reflect the operating status of the heat treatment furnace and the process environment of the workpiece.

[0022] A process phase refers to a stage in the vacuum heat treatment process that has specific objectives and control logic. Examples include the heating phase, the holding phase, and the cooling phase. Each process phase corresponds to a specific set of process parameter requirements and control strategies.

[0023] Heating components refer to the parts in a vacuum heat treatment furnace that provide heat energy, such as heaters and heating zones. Their operating status directly affects the uniformity and stability of the temperature field inside the furnace.

[0024] The health index is an assessment metric for the current operating status and performance degradation of heating components. This index reflects the reliability and potential failure risk of the heating components.

[0025] A decision matrix is ​​a knowledge base or lookup table that stores the corresponding control strategies under different process phases and warning levels. By querying this matrix, a basic control scheme for the current state can be obtained.

[0026] The basic decision-making strategy refers to the initial control scheme obtained from the decision matrix based on the current process phase and warning level. This strategy provides initial direction for subsequent fine-tuning.

[0027] A control command vector is a set of instructions consisting of multiple detailed control parameters, used to guide the actuator in its operation. Examples include setting heater power and adjusting valve opening.

[0028] Example 1 The first aspect of the present invention is described in reference to... Figure 1 A continuous control method for vacuum heat treatment process based on process phase identification is provided, including the following steps: Data is collected in real time, the current process phase is identified based on the data, monitoring data and corresponding thresholds are determined based on the current process phase, the current warning level is calculated based on the deviation between the monitoring data and the thresholds, and the health index is calculated based on the status data of each heating component.

[0029] Based on the current process phase, the early warning level, and the decision matrix, a basic decision strategy is obtained, and the power distribution of each phase heater is dynamically adjusted according to the health index to generate a control command vector.

[0030] The control command vector is decomposed into specific execution actions and output; at the same time, the execution effect is monitored in real time, feedback data is collected, and the parameters of the decision matrix are updated based on the feedback data.

[0031] Specifically, with "process phase recognition" as the core driving logic, the entire control process is divided into three progressive stages: perception, decision-making, and execution. In the perception stage, the system collects multi-sensor data such as temperature, vacuum level, and current in real time. Through feature matching, it identifies the current process phase (such as rapid evacuation, high vacuum heating, gas release sensitive period, vacuum holding period, etc.) and dynamically determines the parameters to be monitored and their corresponding thresholds based on the current process phase as the context benchmark. Based on the parameter deviation, it outputs a graded early warning level with process relevance. At the same time, the system continuously collects status data such as current, voltage, resistance, and cumulative thermal cycle of each heating component, calculates the real-time health index of each heating component, and provides equipment status basis for subsequent decision-making.

[0032] In the decision-making stage, the system maps the current process phase and early warning level to the built-in decision matrix to obtain the corresponding basic decision strategy. It also uses the health index as a weight to dynamically adjust the power distribution of each phase heater, and finally generates a comprehensive control command vector. In the execution stage, the control command vector is decomposed into specific execution actions and output to each actuator. The system monitors the execution effect in real time and uses the collected feedback data to update the decision matrix parameters in real time. This makes the entire control process form a continuously self-optimizing closed loop, ensuring the continuous and stable execution of the main process line in complex dynamic environments.

[0033] Furthermore, the data includes temperature data, vacuum level data, and current data. Temperature data is one of the most crucial process parameters in vacuum heat treatment, its accuracy directly affecting the heating rate, holding time, and final microstructure of the workpiece. Temperature data is typically acquired in real-time using multiple thermocouples placed inside the furnace or near the workpiece. These thermocouples, such as K-type and S-type, convert temperature signals into electrical signals for analysis by the control system. Continuous monitoring of temperature data allows for precise understanding of the workpiece's heating status, ensuring strict adherence to the process curve. Vacuum level data reflects the pressure environment inside the vacuum furnace. Maintaining a specific vacuum level is crucial in vacuum heat treatment to prevent workpiece oxidation and decarburization, and to ensure efficient heat transfer. Vacuum level data is usually measured by vacuum gauges (such as resistance gauges and ionization gauges), which detect the density of gas molecules inside the furnace in real-time and convert it into an electrical signal. Monitoring vacuum level data allows for timely detection of problems such as vacuum system leaks and decreased pump efficiency, thus ensuring the purity of the processing environment. Current data is used to monitor the operating status and power output of the heating components.

[0034] Furthermore, identifying the current process phase includes: extracting features from the data, matching the extracted features with a pre-stored process phase feature library, and determining the current process phase. By extracting features from multi-sensor data, the raw, complex, and potentially noisy data is transformed into more representative and discriminative features, greatly simplifying the difficulty of phase recognition. Subsequently, these refined features are matched with a pre-established and validated process phase feature library, enabling accurate determination of the current process phase using mature pattern recognition or machine learning algorithms. This recognition method based on feature extraction and feature library matching avoids the limitations of subjective judgment or simple threshold determination of raw data, improving the robustness and accuracy of phase recognition.

[0035] Furthermore, the same amount of detection data deviation corresponds to different warning levels under different process phases.

[0036] For example, during the heating phase of vacuum heat treatment, the tolerance for temperature deviations may be low, and even small deviations may be classified as high-level warnings; while during the holding phase, the tolerance for the same temperature deviation may be higher, and it may only be classified as a low-level warning. This differentiated warning mechanism can more accurately reflect the true risks of the process status and avoid false alarms or omissions.

[0037] Furthermore, the warning level includes at least two levels.

[0038] When the warning level is upgraded, switch to the corresponding conservative sub-phase execution control.

[0039] When the warning level is upgraded to the highest level, the heating operation is paused, and the system switches to the constant temperature and pressure holding phase to maintain the current temperature.

[0040] When the system detects that process parameters deviate from the normal range, causing the warning level to escalate from a lower to a higher level (i.e., when the warning level escalates), the system will no longer execute the original process phase control strategy. Instead, it will immediately switch to a preset conservative sub-phase for control. This conservative sub-phase typically employs milder and safer control parameters than the current normal process phase, such as reducing the heating rate, decreasing heating power, increasing holding time, or adjusting the vacuum level, to slow down the process deterioration and buy time for manual intervention or system self-recovery. For example, if the current phase is a rapid heating phase, the system may switch to a slow heating sub-phase or an isothermal sub-phase after the warning escalation.

[0041] Furthermore, when the warning level escalates to the highest level, it indicates that the process is on the verge of extreme danger or loss of control, requiring immediate and stringent protective measures. At this point, the system will forcibly halt all operations that could exacerbate the risk, particularly suspending heating operations to prevent excessively high temperatures from causing material damage or equipment failure. Simultaneously, the system will switch to a isothermal pressure-holding phase, maintaining the current temperature and vacuum level, and will no longer make any proactive adjustments to process parameters. This strategy aims to stabilize the process at a relatively safe level, prevent further deterioration, and provide operators with maximum response time.

[0042] Furthermore, obtaining the basic decision-making strategy includes: Calculate the status index based on the current process phase and the warning level, and obtain the basic decision strategy by querying the decision matrix through the status index.

[0043] Specifically, a status index is a numerical value or code used to uniquely identify a specific process status. In vacuum heat treatment, process phase and warning level are two key dimensions that determine the current process status.

[0044] State Index: ; Where P represents different process phases, L represents the warning level, and H represents the health index range; Establish a decision matrix, and retrieve the decision parameter vector by querying the decision matrix through the state index: ; Among them, a s For action type, p s For power distribution reference adjustment, k s For artificial intervention markers, t sThis is a flag for switching output modes.

[0045] Furthermore, based on the health index, three intervals are defined, and corresponding power protection measures are implemented: When HI(t) ≥ 0.8, it is considered to be in a healthy state and there are no operational restrictions.

[0046] When 0.5 ≤ HI(t) < 0.8, it is judged as mild aging, and a derating factor is automatically applied.

[0047] When HI(t) < 0.5, it is judged as severely aged, an L3 maintenance alarm is generated, the component is marked as to be replaced in automatic mode, and its use is restricted or isolated.

[0048] Furthermore, the dynamic adjustment of the power distribution of each phase heater based on the health index includes: Calculate the basic heating rate coefficient: ; Wherein, the influence function of vacuum change is: ; ΔP ratio Let ΔP be the current rate of change of vacuum. th The threshold for the rate of change of vacuum corresponding to the current phase; The temporary power of each phase heater is normalized to obtain the actual power distribution vector: ; Among them, P act P is the normalized actual power allocation vector; temp P represents the temporary power vector of each phase heater before normalization. temp，i This represents the temporary power value of the heater in phase i.

[0049] Furthermore, it also includes: In response to the manual mode switching command, save the current process phase and decision matrix state.

[0050] In response to the automatic mode recovery command, execution resumes from the breakpoint of the saved process phase and decision matrix state, without restarting the process.

[0051] Specifically, when operators need to manually intervene in the vacuum heat treatment process, the system receives a manual mode switching command. This command can be triggered via a specific button on the user interface, touchscreen operation, host computer software commands, or a physical switch. Upon receiving this command, the system immediately initiates a state saving process, recording the currently executing process phase and its internal sub-states (e.g., the current heating stage, holding time, target temperature, etc.). Simultaneously, the system also saves the current state of the decision matrix used to generate the control command vector, including all its parameters, weights, and the state of the learning model dynamically updated based on feedback data during operation.

[0052] When the operator completes manual intervention and requests the system to resume automated control, the system receives an automatic mode recovery command. This command can also be triggered via the user interface, host computer software, or a hardware switch. Upon receiving the automatic mode recovery command, the system no longer executes the entire heat treatment process from scratch. Instead, it reads the saved process phase identifiers, sub-states, and decision matrix parameters from the previously saved non-volatile memory. Based on the loaded process phase information, the system precisely restores its internal state machine or control logic to the process stage exactly as it was at the time of saving. For example, if the system was at a specific point in the "holding stage" at the time of saving, it will directly enter that holding stage and continue timing or executing subsequent operations from that point. Simultaneously, the system reloads the loaded decision matrix parameters into the intelligent decision module, enabling it to continue making decisions based on the learning and optimization results before manual intervention. Based on the restored process phase and decision matrix state, the system begins to generate new control command vectors and decomposes them into specific execution actions, outputting them to actuators such as heaters and vacuum pumps, thereby achieving seamless continuity of the process without starting from scratch or performing complex initialization.

[0053] To establish a complete decision matrix: First, we define three dimensions: process phase P, warning level L, and health index range H. In this embodiment, P represents seven different process stages, L represents a four-level warning level, and H represents a three-level health index range.

[0054] State Index S corresponds to a three-state group. Establish a decision matrix. That is, the dimension is 84 rows, each row corresponds to a state index S, and the number of columns is m.

[0055] For any state S, it can be decomposed as: .

[0056] The entire system also includes the following auxiliary matrices and vectors, defining a basic coefficient vector A and B to store the α and β coefficients corresponding to each state index S, respectively. and B The process early warning adjustment matrix Kreg is used to store the adjustment factors under different combinations of process phases and early warning levels.

[0057] Redefining the learning and optimization vector: the rating weight vector Weights are used for process quality, processing efficiency, and health protection. Priority coefficient vector. The weights of each feature are used to store the conflict resolution process. The learning efficiency matrix η is used to store the learning efficiency of each parameter in each state.

[0058] Basic heating rate coefficient: ; Vacuum variation influence function: ; Influence function of heating component health index: ; Where 0.7 represents the starting penalty value of the interval, 0.3 represents the range of change of the penalty value within the interval, and 0.8 represents the critical point of the health index.

[0059] Actual power allocation vector: ; By establishing the above mathematical model, through index mapping and linear parameter calculation, and then through matrix learning algorithm, the execution decision for each state is finally obtained to correspond to the specific control output state. In terms of early prediction, it can intervene in the fault early and avoid serious faults at L3 / L4 level.

[0060] Regarding the health index: The formula for calculating the heating component health index (HI) is as follows: ; Wherein, HI_base(t) is the basic long-term health index; S(t) is the real-time operating stress coefficient; This represents the stress influence weighting coefficient.

[0061] The baseline long-term health index HI_base(t) is: ; Where t is the cumulative operating time of the heating component; E_accum is the time aging characteristic constant; E_L is the cumulative heat energy consumed; and E_L is the energy aging coefficient.

[0062] Real-time stress coefficient S(t): ; Where R(t) is the real-time calculated resistance of the heating component at the current moment. R0 is the initial nominal resistance of the component in a standard cold state (such as room temperature). This is the square term of the rate of change of resistance. It is normal for resistance to increase with increasing temperature, but abnormal increases (such as loose connections or material deterioration) or decreases (partial short circuits) will increase this term, and the square term can amplify the abnormal effects.

[0063] F_I(t): Current fluctuation factor. F_V(t): Voltage imbalance factor (for multiphase heating). Phase_Risk: Process phase risk coefficient. A basic risk value is dynamically assigned by the system based on the currently identified process stage. For example: Vacuuming stage: 0.1 (low risk); Steady heating stage: 0.3 (medium risk); Gas release-sensitive heating stage: 0.8 (high risk); Extreme high-temperature holding stage: 0.9 (extremely high risk).

[0064] The normalized weight coefficients of each item, and satisfying The weights can be adjusted based on component type and primary failure mode. For example, for components prone to connection aging, the weight can be increased. The weight.

[0065] Example 2 This embodiment provides a continuous control system for a vacuum heat treatment process based on process phase recognition, employing the continuous control method for a vacuum heat treatment process based on process phase recognition as described in Embodiment 1. Please refer to [link / reference]. Figure 2 ,include: The sensing and fusion module collects data in real time, identifies the current process phase based on the data, determines the monitoring data and corresponding thresholds based on the current process phase, calculates the current warning level based on the deviation between the monitoring data and the thresholds, and calculates the health index based on the status data of each heating component.

[0066] The intelligent decision-making module obtains a basic decision-making strategy based on the current process phase, the early warning level, and the decision matrix, and dynamically adjusts the power distribution of each phase heater according to the health index to generate a control command vector.

[0067] The control execution module is used to decompose the control command vector into specific execution actions and output them, while collecting feedback data and updating the parameters of the decision matrix based on the feedback data.

[0068] In this embodiment, the control system comprises an innovative closed-loop architecture consisting of three parts: a perception and fusion module, an intelligent decision-making module, and a control execution module. This differs from the traditional open-loop or fixed-program execution mode of "acquisition-alarm". The perception and fusion module, as the system's information entry point, is responsible not only for real-time acquisition of multi-sensor data such as temperature, vacuum level, and current, but also, with process phase identification as its core, outputs two types of key information in real time: first, the current process phase identifier, including rapid evacuation, high vacuum heating, venting sensitive period, and vacuum holding period; second, structured multi-level early warning events. Based on parameter deviation, changing trends, and subsystem correlations, the system generates graded early warning objects ranging from L1 alert to L4 emergency. Each early warning object includes abnormal parameters, level, trend value, and preliminary diagnostic information, providing input basis with process context relevance for subsequent decision-making.

[0069] The intelligent decision-making module, as the core of the system, operates in parallel with two functional units: the adaptive control decision-making unit and the component health management unit. The adaptive control decision-making unit, based on the current process phase identifier and warning level, queries the built-in "phase-warning level-action" decision matrix to generate tiered control commands ranging from parameter fine-tuning to emergency shutdown. Simultaneously, it reads the health index HI output by the component health management unit in real time, using it as a decision weight to dynamically adjust the power distribution of each phase heater. The component health management unit continuously analyzes operating data such as current, voltage, resistance, and cumulative thermal cycles of the heating components, calculating the real-time health index HI and predicted remaining effective life (RUL) through a degradation model, achieving proactive perception of equipment status. The control execution module decomposes the decision commands into specific execution actions, while simultaneously collecting execution feedback data and updating the decision matrix parameters in real time, enabling the entire system to form a continuously self-optimizing closed-loop control capability.

[0070] The present application will now be explained using a specific embodiment: 1. Phase identification and monitoring: The system determines that it is in the "high temperature rise period". Under this process phase, vacuum control has a high priority.

[0071] 2. Sudden Interference and Graded Early Warning: During the temperature rise to 550℃-850℃, a sudden and violent gas release occurs, and the rate of vacuum increase exceeds the limit sharply. The multi-level early warning module comprehensively analyzes the current process phase (sensitive to gas release) and parameter severity to generate an L2 level early warning: "The rate of change of vacuum level is seriously exceeded, and the process stability is threatened."

[0072] 3. Multi-level fusion decision-making and execution: The adaptive control decision-maker receives L2 warnings and the current process phase. It queries the decision matrix and decides to adopt a "vacuum-dominated heating rate regulation" strategy. Health index fusion: The decision-maker simultaneously reads that the B-phase heater HI=0.65. According to preset rules, for components with HI<0.7, a protective derating factor must be applied during power adjustment.

[0073] Comprehensive instruction generation: Decision setter output instructions: ① Smoothly reduce the total heating rate from 3℃ / min to 1.5℃ / min; ② When allocating power, apply an additional 15% derating to the B-phase heater, and apply the same additional derating to the A-phase and C-phase heaters to ensure that the power ratio of each zone is the same and to ensure uniform furnace temperature.

[0074] Control execution: The power of each phase is adjusted smoothly, the overall heating rate is slowed down, and phase B operates in a safer load range.

[0075] 4. Information output: The structured L2 alarm card pops up on the human-machine interface and displays: "Alarm: Vacuum venting | The system has automatically cooled down to 1.5℃ / min and optimized power distribution (phase B is slightly aging, and phases A and C have also been actively derated for protection) | Status: Monitoring".

[0076] 5. Scenario Upgrade and Decision Switching: If the venting continues to worsen and the vacuum level reaches a more dangerous threshold, the early warning module will upgrade the alarm to Level L3.

[0077] The decision-maker initiates a secondary strategy: pause heating and switch to the "constant temperature and pressure" sub-phase to maintain the current temperature.

[0078] The interface alarm has been upgraded to L3, with the following suggestion: "Heating has been paused. Please check the material condition. You can manually confirm and resume heating once it is safe to do so." At this point, the operator can switch to manual mode for detailed inspection. The system saves the current "constant temperature and pressure" decision status.

[0079] Recovery and seamless transition: After operator intervention, switch back to automatic mode. The system reads the saved "constant temperature and pressure" phase and status, automatically detects that the vacuum has been restored, and then resumes from the decision interruption point, smoothly transitioning back to the heating phase to complete the final heating.

[0080] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A continuous control method for vacuum heat treatment process based on process phase recognition, characterized in that, include: Real-time data acquisition; identification of the current process phase based on the data; determination of monitoring data and corresponding thresholds based on the current process phase; calculation of the current warning level based on the deviation between the monitoring data and the thresholds; and calculation of the health index based on the status data of each heating component. The basic decision-making strategy is obtained based on the current process phase, the early warning level, and the decision matrix, and the power distribution of each phase heater is dynamically adjusted according to the health index to generate a control command vector. The control command vector is decomposed into specific execution actions and output; at the same time, the execution effect is monitored in real time, feedback data is collected, and the parameters of the decision matrix are updated based on the feedback data.

2. The continuous control method for vacuum heat treatment process based on process phase recognition as described in claim 1, characterized in that, Identifying the current process phase includes: extracting features from the data, matching the extracted features with a pre-stored process phase feature library, and determining the current process phase.

3. The continuous control method for vacuum heat treatment process based on process phase recognition as described in claim 1, characterized in that, The same amount of deviation in the detection data corresponds to different warning levels under different process phases.

4. The continuous control method for vacuum heat treatment process based on process phase recognition as described in claim 1, characterized in that, The warning levels include at least two levels; When the warning level is upgraded, the corresponding conservative sub-phase execution control is switched; When the warning level is upgraded to the highest level, the heating operation is paused, and the system switches to the constant temperature and pressure holding phase to maintain the current temperature.

5. The continuous control method for vacuum heat treatment process based on process phase recognition as described in claim 1, characterized in that, The calculation of the health index includes: Obtain the health index of each heating component, and determine the health index range based on the numerical range of the health index.

6. The continuous control method for vacuum heat treatment process based on process phase recognition as described in claim 5, characterized in that, When the health index of any heating component is lower than a preset threshold, a protective derating factor is applied to all heating components simultaneously to ensure that the power ratio of each heating zone is the same.

7. The continuous control method for vacuum heat treatment process based on process phase recognition as described in claim 6, characterized in that, Based on the health index, three intervals are defined and corresponding power protection measures are implemented: When HI(t) ≥ 0.8, it is considered to be in a healthy state and there are no operational restrictions. When 0.5 ≤ HI(t) < 0.8, it is judged as mild aging, and a derating factor is automatically applied; When HI(t) < 0.5, it is judged as severely aged, an L3 maintenance alarm is generated, the component is marked as to be replaced in automatic mode, and its use is restricted or isolated.

8. The continuous control method for vacuum heat treatment process based on process phase recognition as described in claim 1, characterized in that, The basic decision-making strategy includes obtaining the state index: ； Where P represents different process phases, L represents the warning level, and H represents the health index range; Establish a decision matrix, and retrieve the decision parameter vector by querying the decision matrix through the state index: ； Among them, a s For action type, p s For power distribution reference adjustment, k s For artificial intervention markers, t s This is a flag for switching output modes.

9. The continuous control method for vacuum heat treatment process based on process phase recognition as described in claim 8, characterized in that, The dynamic adjustment of the power distribution of each phase heater based on the health index includes: Calculate the basic heating rate coefficient: ； Wherein, the influence function of vacuum change is: ； ΔP ratio Let ΔP be the current rate of change of vacuum. th The threshold for the rate of change of vacuum corresponding to the current phase; The temporary power of each phase heater is normalized to obtain the actual power distribution vector: ； Among them, P act P is the normalized actual power allocation vector; temp P represents the temporary power vector of each phase heater before normalization. temp，i This represents the temporary power value of the heater in phase i.

10. A continuous control system for vacuum heat treatment process based on process phase recognition, characterized in that, The continuous control method for vacuum heat treatment process based on process phase recognition as described in any one of claims 1-9 includes: The sensing and fusion module collects data in real time, identifies the current process phase based on the data, determines the monitoring data and corresponding thresholds based on the current process phase, calculates the current warning level based on the deviation between the monitoring data and the thresholds, and calculates the health index based on the status data of each heating component. The intelligent decision-making module obtains a basic decision-making strategy based on the current process phase, the early warning level, and the decision matrix, and dynamically adjusts the power distribution of each phase heater based on the health index to generate a control command vector. The control execution module is used to decompose the control command vector into specific execution actions and output them, while collecting feedback data and updating the parameters of the decision matrix based on the feedback data.

Images