Vacuum induction heating process closed-loop control system applying industrial mechanism model

Abstract

The invention relates to the technical field of industrial process control, and discloses a vacuum induction heating process closed-loop control system applying an industrial mechanism model, which comprises the following steps: acquiring the input active power of an induction coil through a power monitoring unit, and measuring the total external radiation heat flux of the outer wall of a vacuum chamber by a heat flow sensing unit; the embedded decision-making module calculates the thermal-electric response ratio and the time derivative thereof in real time, judges the phase change critical state of the material based on the derivative change and generates a regulation and control instruction, through dynamic capture of the thermal-electric energy balance relation, recognition of the phase change critical point of the material is achieved, and by combining working condition pre-judgment of the vibration spectrum entropy and current ripple characteristic arbitration, the phase change critical state of the material is judged. An anti-interference multi-verification mechanism is constructed, and the robustness of the control system under the preset working condition is improved.

Description

Technical Field

[0001] This invention relates to a closed-loop control system for a vacuum induction heating process that applies an industrial mechanism model, belonging to the field of industrial process control technology. Background Technology

[0002] Currently, this field mainly adopts a decoupled control strategy based on temperature field monitoring, which indirectly regulates the phase transition process through multi-sensor fusion and physical field simulation models. Such methods can maintain basic control under normal operating conditions, but they face three technical bottlenecks in high-precision material processing: First, relying on macroscopic temperature parameters to infer microscopic phase transitions has a physical delay, leading to frequent overheating or cold shut defects; Second, under the decoupled architecture, a single disturbance such as power grid fluctuations can be amplified into system oscillations through the electromagnetic-thermal field coupling path, which is difficult to suppress by traditional PID control; Third, the prediction model relying on offline training needs to be frequently recalibrated as the operating conditions change, resulting in high maintenance costs.

[0003] Despite industry attempts to optimize systems using high-frequency sampling or composite control algorithms, these efforts have exacerbated system complexity and failed to overcome the inherent limitations of decoupling methods: the indirect correlation between temperature signals and the nature of phase transitions, the systemic fragility of disturbance propagation, and the model's poor adaptability to dynamic conditions. Fundamentally, the bottleneck of existing technologies stems from the indirectness of their control signals. For example, in thermocouple-based temperature measurement, the response is inevitably limited by the physical process of heat conduction from the melt core to the measuring point. The inherent thermal resistance and thermal capacity of this process are the physical basis for signal delay. Furthermore, when the grid voltage fluctuates, this disturbance is directly transmitted to the input power, thus affecting the temperature field. Traditional PID control, due to its hysteresis compensation characteristics, struggles to effectively suppress this systemic oscillation transmitted via the electromagnetic-thermal coupling path. Therefore, there is an urgent need in this field for a direct sensing method that can bypass thermal conduction delays and inherently resist common-mode disturbances. Summary of the Invention

[0004] This invention provides a closed-loop control system for a vacuum induction heating process using an industrial mechanism model. Its main purpose is to solve the problems of insufficient phase change control accuracy and lack of system robustness caused by traditional decoupled control strategies due to sensing lag, disturbance amplification, and model maintenance costs.

[0005] To achieve the above objectives, the present invention provides a closed-loop control system for a vacuum induction heating process based on an industrial mechanism model, comprising: A power monitoring unit is configured to acquire the input active power of the induction coil in real time; A heat transfer sensor unit is configured to measure the total external radiative heat flux outside the vacuum chamber in a non-contact manner. The heat transfer sensor unit is installed on the outer wall of the vacuum chamber and its monitoring area is directly facing the heating and melting zone. An embedded decision module is configured to: calculate the thermoelectric response ratio in real time based on the input active power and the total external radiant heat flux, where the thermoelectric response ratio is the ratio of the total external radiant heat flux to the input active power; calculate the time first derivative of the thermoelectric response ratio; and determine whether the material has entered the critical state of phase transition based on the comparison result of the time first derivative of the thermoelectric response ratio with zero, and generate a power modulation command, wherein the condition for determining that the material has entered the critical state of phase transition is that the time first derivative is less than or equal to a preset determination threshold. A power execution unit is configured to switch the input active power to a preset insulation power value in response to a power modulation command.

[0006] Preferably, the embedded decision module is further configured to: determine that the system is in a stable heating stage if the time first derivative of the thermo-electric response ratio is continuously greater than zero; the preset judgment threshold is a value greater than zero and less than a first predetermined small value.

[0007] Preferably, the system further includes a vibration sensing unit configured to acquire the vibration spectrum characteristics of the vacuum induction heating system at different operating stages and calculate the vibration entropy based on the vibration spectrum characteristics.

[0008] Preferably, the vibration sensing unit includes: a piezoelectric vibration sensor adsorbed on the support structure of the vacuum chamber for acquiring vibration signals; and a vibration signal processing unit configured to perform a fast Fourier transform on the vibration signal and calculate the vibration entropy of the vibration signal.

[0009] Preferably, the system further includes a working condition pre-classifier, configured as follows: when the vibration entropy is lower than the first vibration entropy judgment threshold, the system is determined to be in a spectral ordered state, and the weight of the embedded decision module in making decisions based on the total external radiant heat flux is increased; and when the vibration entropy is higher than the second vibration entropy judgment threshold, the system is determined to be in a spectral chaotic state, and the weight of the embedded decision module in making decisions based on the total external radiant heat flux is reduced.

[0010] Preferably, the system further includes a current ripple analysis unit, configured to analyze the predetermined phase transition distortion characteristics of the high-frequency current ripple of the current feedback signal by sampling the existing current feedback signal in the power controller at high speed, so as to determine the impedance change caused by the material phase transition.

[0011] Preferably, the system further includes a decision arbitrator configured such that: when the embedded decision module generates a power modulation command, the decision arbitrator is activated; within a preset arbitration time window after activation, if the current ripple analysis unit simultaneously identifies a predetermined phase transition distortion characteristic, the decision arbitrator confirms the power modulation command and allows its execution; and if the current ripple analysis unit fails to identify the predetermined phase transition distortion characteristic, the decision arbitrator rejects the power modulation command, temporarily locks the power, and simultaneously sends a sensor source conflict alarm to the system.

[0012] Preferably, the system also includes a decision threshold dynamic self-compensation module, which is configured to: when the rate of change of vibration entropy exceeds a preset event change judgment threshold, dynamically adjust the decision threshold used to determine the critical state of phase transition based on the integral value of the time change rate of the thermo-electric response ratio.

[0013] Preferably, the decision threshold dynamic self-compensation module is configured to dynamically adjust the decision threshold according to the following formula: in, It is the initial decision threshold set to determine the critical state of phase transition; This is an adjustable compensation coefficient; It is a phase change inertia characterization factor calculated based on the integral of the time first derivative of the thermo-electric response ratio.

[0014] Preferably, the system further includes an online melt viscosity inversion module, configured to: superimpose a micro-perturbation AC signal with a fixed amplitude and frequency onto the input active power of the induction coil; and invert the signal or data characterizing the change in melt viscosity in real time by analyzing the amplitude of the response signal in the total external radiant heat flux that is in the same frequency as the micro-perturbation signal.

[0015] Meanwhile, in this invention, the material phase transition process is transformed from a microscopic change of state that is difficult to observe directly into a macroscopic energy balance change that can be accurately captured. Its physical essence is: during the stable heating phase, the active power input to the system... This is mainly converted into an increase in the internal energy of the material and an increase in external radiative heat flux. With the increase of , the thermo-electric response ratio at this time derivative It remains a small positive value; however, when the material enters the critical state of phase transition, it absorbs a large amount of latent heat of phase transition. This means that the continuously input electrical energy is mainly used to overcome intermolecular forces rather than to raise the material temperature. Therefore, the external radiative heat flux... The growth rate will be relative to the nearly constant input power. Significant slowdown or even stagnation, leading to By rapidly reducing and crossing the zero point, this invention achieves direct and instantaneous identification of the phase transition critical point by dynamically capturing the sudden change in system energy balance caused by the absorption of latent heat of phase transition, thus fundamentally avoiding the heat conduction delay problem of traditional temperature measurement methods.

[0016] In addition, when external power grid fluctuations and other factors cause input power When changes occur, due to the rapid response characteristics of induction heating, heat is dissipated externally. It will also change rapidly in the same direction, when calculating the ratio. At this time, such common-mode interference is largely canceled out mathematically, thus making the decision basis... It is insensitive to such external disturbances, which greatly improves the robustness of the control system.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. By dynamically capturing the abrupt inflection point of energy balance through the time derivative of the thermo-electric response ratio, the system directly senses the thermo-electromagnetic coupling imbalance triggered by the phase change within the material. The dual verification mechanism of vibration entropy predicting operating conditions and current ripple distortion characteristics helps reduce the impact of environmental noise on critical state identification, thereby helping to reduce or suppress overheating or cold shut defects. The design of the input power to external heat dissipation ratio minimizes common-mode interference from grid fluctuations. The values ​​are naturally canceled out during calculation. When the vibration entropy detects a chaotic state in the spectrum, the system automatically reduces the weight of the heat flow signal and activates the current ripple arbitration, forming a cross-physical domain anti-disturbance logic closed loop without the need for additional compensation algorithms.

[0018] 2. The vibration entropy change rate trigger threshold dynamic compensation module quantifies phase transition inertia based on the integral value of the thermo-electric response ratio derivative. The nonlinear coupling of this factor with the decision threshold enables the system to autonomously adjust the tolerance of the critical state criterion to adapt to the diffuse phase transition process of high-entropy alloys, which helps to reduce or suppress control oscillations caused by fixed thresholds.

[0019] 3. The heat flow sensing unit is reused to capture the same frequency thermal response excited by micro-disturbances. Through the multi-domain coupling effect of electromagnetic-thermal-rheological, the melt viscosity, which cannot be directly observed, is converted into a quantifiable signal. This mechanism enables closed-loop optimization of processes such as refining agent addition and electromagnetic stirring based on real-time physical property data. The power supply current feedback signal is reused to realize ripple arbitration. The vibration entropy of the support structure is used to predict the working condition. Multiple protections are built with a sensing cost 10% lower than that of traditional solutions. The control core only requires lightweight derivative calculation and logic judgment, so that high-precision phase change control can be deployed in industrial-grade embedded systems. Attached Figure Description

[0020] Figure 1This is a schematic diagram of the structure and signal interaction process of the closed-loop control system for the vacuum induction heating process of the present invention; Figure 2 This is a curve comparing the changes in current ripple amplitude and phase transition characteristic indicators, as well as the phase transition identification threshold, according to the present invention. Figure 3 This is a timing diagram of the dual verification mechanism for phase transition detection and the power modulation process of the present invention.

[0021] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail with reference to the accompanying drawings.

[0023] The vacuum induction heating process closed-loop control system disclosed in this invention, which applies an industrial mechanism model, has an overall architecture in which a core control loop is constructed using a power monitoring unit, a heat flow sensing unit, an embedded decision module, and a power execution unit. It is supplemented by modules such as a vibration sensing unit, a current ripple analysis unit, and a decision arbitrator to form a multi-verification and adaptive mechanism. The operating logic of this system begins with the synchronous monitoring of the system's electrical input and heat output. The resulting data stream is fed into the embedded decision module for core mechanism calculation and material critical state discrimination, thereby generating a power modulation command, which is implemented by the power execution unit. This decision-making process is continuously subject to parallel channel arbitration and dynamic correction from vibration spectrum analysis and power supply current ripple characteristics, thus constructing a closed-loop control system with multi-physical domain information fusion, disturbance resistance, and adaptive capabilities.

[0024] In the actual process of vacuum induction heating, traditional control strategies that rely on temperature measurement methods such as thermocouples to infer material phase changes often suffer from a lag in determining the critical point of phase change due to the physical distance between the measuring point and the core region of the melt and the inherent delay in heat conduction. This can lead to process defects such as overheating or cold shut-off. To address this challenge, the core control procedure of this system begins with the dynamic capture of the energy balance relationship. Specifically, the power monitoring unit continuously acquires the input active power of the induction coil in real time. Meanwhile, a non-contact heat flow sensor unit, with its monitoring area directly facing the heated and molten zone, is installed on the outer wall of the vacuum chamber to simultaneously measure the total external radiative heat flux. These two raw data streams are fed into the embedded decision module, which first calculates the thermal-electric response ratio in real time based on the input. It is defined as the ratio of total external radiative heat flux to input active power, i.e. The module then further calculates the time first derivative of the thermo-electric response ratio. Before the material enters the solid-liquid phase transition zone, its temperature and external radiant heat flux steadily increase with continuous energy injection. If the temperature remains consistently above a preset, slightly higher reference value (above zero), the system determines that it is in a stable heating phase. However, when the material begins to melt and absorb a large amount of latent heat of phase change, the rate of increase in its external radiant heat flux will significantly slow down or even stop compared to the nearly constant input power, leading to… The system rapidly decreases the time derivative. Once the system detects that the first derivative of the time is less than or equal to a calibrated preset threshold, it determines that the material has entered the critical phase transition state and immediately generates a power modulation command. The power execution unit precisely switches the input active power to a preset heat preservation power value to maintain the molten state. By directly linking the dynamic ratio change of input electrical energy to output heat energy, this mechanism achieves essential identification and instantaneous response to the critical phase transition point from the underlying physical mechanism of energy conservation. Furthermore, in complex industrial environments, external disturbances such as power grid fluctuations, spatial electromagnetic noise, or equipment mechanical vibrations can contaminate heat transfer sensing signals, leading to misjudgments by decision-making mechanisms that rely solely on the heat-electric response ratio. Therefore, the system also integrates a working condition prediction mechanism based on vibration spectrum characteristics to assess the reliability of the decision-making basis. This mechanism uses a piezoelectric vibration sensor adsorbed on the vacuum chamber support structure to collect data under different working conditions. The vibration signal is immediately sent to the vibration signal processing unit for Fast Fourier Transform (FFT) to calculate the vibration entropy, which characterizes the degree of order or chaos in the system's operation. A pre-classifier adjusts the decision weights based on this vibration entropy value. The specific procedure is as follows: a first and a second vibration entropy judgment threshold are preset. When the calculated vibration entropy is lower than the first threshold, the system is judged to be in a stable working state with an ordered spectrum. At this time, the total external radiated heat flux data obtained by the heat flow sensing unit enables the system to have adaptive adjustment capabilities. When the embedded decision module executes the decision, its corresponding weight is increased. Conversely, when the vibration entropy is higher than the second threshold, the system is judged to be in a non-steady state or a strongly disturbed state with a chaotic spectrum. The decision weight of the heat flow signal is reduced accordingly. In this way, the system obtains the ability to dynamically assess the reliability of the core sensing source based on the overall physical state of the equipment, thereby initially filtering out some disturbances at the decision source.

[0025] To enhance anti-interference capabilities, the system further introduces a cross-validation step based on electrical characteristics to arbitrate the phase transition judgment conclusion obtained from the thermo-electric model. This step aims to address extreme operating conditions where the heat flow sensor itself malfunctions or its signal weight, even after being reduced, is still insufficient to avoid misjudgments. The system's embedded current ripple analysis unit, through high-speed sampling of the inherent current feedback signal in the power controller, focuses on analyzing the preset distortion pattern of the ripple in the high-frequency band of this signal. This design utilizes the physical phenomenon that the permeability, resistivity, and other physical properties of materials undergo abrupt changes during phase transitions, which in turn affect the impedance of the induced heating load, ultimately leaving identifiable predetermined phase transition distortion characteristics in the power supply current ripple. A decision arbitrator is configured within the embedded decision module based on... The thermoelectric model is activated synchronously when generating power modulation commands. Within a preset arbitration time window of nanoseconds to milliseconds after activation, the arbitrator will verify whether the current ripple analysis unit has synchronously identified the aforementioned predetermined phase transition distortion characteristics. If the conclusions of the two sources are consistent, the decision arbitrator confirms the power modulation command and allows it to be sent to the power execution unit. If the current ripple analysis unit fails to identify the corresponding characteristics within the window period, the decision arbitrator determines that there is a source conflict, executes the command rejection, temporarily locks the current power output, and sends a sensor source conflict alarm to the system. This cross-physical domain dual verification mechanism provides a cross-verification mechanism for the final confirmation of the phase transition critical point, improving the decision robustness of the control system under preset operating conditions.

[0026] Furthermore, considering the differences in the kinetic characteristics of phase transformation processes, such as dispersion degree and duration, among different types and even different batches of metallic materials, using a fixed judgment threshold may lead to insufficient system adaptability. Therefore, this system also includes a dynamic self-compensation module for the decision threshold, enabling the system to have adaptive adjustment capabilities. This module continuously monitors the vibration entropy and its rate of change calculated by the vibration sensing unit. When this rate of change exceeds a preset event change judgment threshold used to characterize significant changes in process state, such as furnace charge collapse or violent boiling, the module is triggered. After triggering, the module will calculate the time first derivative based on the thermo-electric response ratio. The integral values ​​over a period of time are used to calculate a phase transition inertia characterization factor that can quantify the severity and duration of the phase transition process. Subsequently, the module dynamically adjusts the decision threshold for determining the critical state of phase transition based on the preset nonlinear compensation formula: threshold = original threshold × (1 + α⋅φ). Here, the original threshold is the initial decision threshold set for the reference material. This is an adjustable compensation coefficient that can be calibrated offline or learned online, used to specify the compensation strength by inertial characteristics of the phase transition process. The dynamic feedback to the decision threshold setting stage allows the system to autonomously adjust its tolerance for judging the critical state, thereby flexibly adapting to the phase transition dynamics of different materials and avoiding control oscillations that may result from the mismatch between the fixed threshold and the dynamic process.

[0027] Finally, in some refining processes with extreme requirements for melt quality, melt viscosity, as a key but difficult-to-measure state parameter, restricts the optimal control of the process. This system provides an extended function for online melt viscosity inversion to address this measurability bottleneck. This function is implemented by an online melt viscosity inversion module, whose workflow is as follows: a micro-perturbation AC signal with a fixed amplitude and frequency is actively superimposed on the input active power of the induction coil. This micro-power perturbation is transmitted to the melt via electromagnetic-thermal coupling, and due to differences in melt viscosity, it induces varying degrees of heat dissipation and flow response, which is reflected in subtle fluctuations in the external radiated heat flux of the vacuum chamber. The system reuses a high-sensitivity heat flow sensing unit and uses precise phase-locked amplification and other signal processing algorithms to obtain the total external radiated heat flux. In this study, the amplitude of the response signal with the same frequency as the applied micro-perturbation signal is specifically analyzed and extracted. Since this response amplitude is directly related to the flow and heat transfer characteristics inside the melt, and the latter is controlled by the melt viscosity, analyzing the real-time changes in the amplitude of this synchronous response signal allows for the derivation of a signal or data characterizing the melt viscosity change. This mechanism, through an excitation-response mode, transforms the core physical properties of the melt, which cannot be directly observed, into a quantifiable external thermal signal. This provides a new technical path for achieving advanced closed-loop control strategies such as efficient addition of refining agents and precise optimization of electromagnetic stirring intensity based on real-time physical property data. Simultaneously, the phase change inertia characterization factor... It is a physical quantity used to quantify the kinetic characteristics of phase transition processes; it is obtained by expressing the first time derivative of the thermo-electric response ratio. The integral is obtained by integration within a preset time window, and the starting point of this integration is usually the moment when the system determines that it has entered a state close to a phase transition (i.e., The moment when the continuous downward trend begins to emerge (the endpoint of the integral is the moment when the system triggers the dynamic self-compensation module for the decision threshold). The physical significance lies in the fact that for materials with a smooth phase transition process and a long duration (such as the dispersed phase transition of certain high-entropy alloys), its Negative values ​​will persist for a long time, leading to a decrease in the cumulative integral value. The absolute value is relatively large; however, for materials with violent phase transition processes and clear inflection points, The absolute value is relatively small, therefore, The value directly reflects the inertia of the phase transition, providing a quantitative and clearly defined physical basis for the tolerance of the system's adaptive adjustment of the decision threshold. These are all extended implementation methods known to those skilled in the art.

[0028] Example 1: In a vacuum induction melting operation of a high-purity aerospace high-entropy alloy, the closed-loop control system was executing a stable heating program. At this time, a high-frequency voltage transient occurred in the external power grid, and simultaneously, due to the complex alloy composition, an unexpected and violent outgassing and collapse event occurred in the furnace near the melting point. The superposition of these two disturbances of very different natures directly contaminated the input active power obtained by the power monitoring unit. The stability of the data, and the total external radiative heat flux measured by the heat transfer sensor unit. This causes impactful disturbances, plunging the overall physical state of the system into chaos; under this combined extreme condition, relying solely on the contaminated... and Calculated thermo-electric response ratio and its time first derivative Making decisions carries the risk of judgment failure; the system's multi-layer verification mechanism responds immediately. First, the piezoelectric vibration sensor adsorbed on the vacuum chamber support structure captures the structural vibration caused by the collapse of the furnace charge. The vibration entropy calculated by the vibration signal processing unit instantly exceeds the preset second vibration entropy judgment threshold. Based on this, the operating condition pre-classifier determines that the system has entered a spectral chaotic state and immediately reduces the amount of heat flux based on total external radiation in the embedded decision module. By assigning weights to decisions, this action effectively isolates the transmission of heat flow signal distortion caused by physical impacts to the main decision logic.

[0029] Although the weight of the heat flux signal has been reduced, the decision path based on the thermo-electric response ratio is still not entirely reliable due to the instability of the input power signal. At this point, the decision arbitrator checks the conclusions of the current ripple analysis unit from an independent physical channel. This unit, by analyzing the high-frequency characteristics of the power controller's current feedback signal, identifies an electrical fingerprint caused by a step change in the material's permeability and resistivity at the phase transition point, and notes low sensitivity to external physical vibrations and low-frequency fluctuations in the power grid. This electrical fingerprint indicates that the material has reached its melting critical point. The decision arbitrator then confirms the phase transition judgment and allows the execution of the power modulation command, switching the power to the preset insulation power value. This control scheme does not attempt to filter noise from the contaminated signal to reconstruct the ideal input. Instead, it transforms the control problem from obtaining a pure signal to making reliable judgments when multiple signals may be distorted simultaneously through a multi-source and cross-physical domain decision-making arbitration system. By basing decision-making on electrical signal characteristics that reflect the material's intrinsic physical properties, and using macroscopic state parameters such as vibration entropy to dynamically evaluate the instantaneous reliability of different sources, the system completes the determination and control of the phase transition critical point under conditions of dual disturbances at both the energy input and heat output ends. The ingots finally obtained from this batch of high-entropy alloys, after testing, showed that their internal grains were uniform and fine, without metallurgical defects such as cold shuts or inclusions caused by local overheating or insufficient melting due to underheating or undertemperature. All performance indicators met the established standards for aerospace applications.

[0030] Example 2: To objectively quantify the performance of the closed-loop control system of the present invention in identifying the critical state of material phase transition, and to define its response difference compared with the traditional temperature measurement control method, the following comparative experiment was established. The experiment aims to simultaneously record the timing of the determination of the critical point of phase transition by the thermo-electric response ratio method of the present invention and the high-precision thermocouple temperature measurement method, which is the industry benchmark, in the same melting process under controlled conditions.

[0031] The experiment was conducted on a 50-kilowatt unit with a vacuum level of [missing information]. The experiment was conducted in a Pascal-level laboratory-standard vacuum induction furnace. The experimental material consisted of 1.5 kg of high-temperature nickel-based alloy Inconel 718 bars placed in a yttrium oxide ceramic crucible. For direct control, the experimental platform integrated the complete control and sensing units of this invention, including a power monitoring unit, a non-contact heat transfer sensing unit, a piezoelectric vibration sensor, and a current ripple analysis unit. A separate conventional temperature measurement system was also deployed, employing an S-type platinum-rhodium / platinum thermocouple closely attached to the center of the crucible's outer wall for temperature acquisition. The data acquisition and system decision-making calculation cycles for all sensors were uniformly set. The setting of this cycle aims to balance data real-time performance with system processing load. Its setting procedure is based on the following principles: the data acquisition cycle should be less than the thermal inertia time constant of the system under test to ensure distortion-free capture of process dynamics. In this experiment, the system thermal inertia time constant under the preset furnace and load combination was measured to be approximately 3.5 seconds through a pre-test step response test. Therefore, 30 milliseconds was selected as the data acquisition and calculation cycle. This value satisfies engineering constraints and provides a margin for signal processing. After the experiment starts, the system heats the furnace charge with a constant input active power and simultaneously records the thermo-electric response ratio calculated by the control system of this invention. and its time first derivative The temperature values ​​measured by the S-type thermocouple showed that all parameters changed steadily during the initial solid-state heating phase; after approximately 360 seconds, the core decision parameters... The temperature showed a continuous downward trend and reached the preset critical state determination threshold for phase change for the first time at 372 seconds. At this time, the system determined that the material had entered the phase change process. The temperature reading of the thermocouple, which served as a reference, continued to rise almost linearly during this period until 385 seconds, when the slope of its temperature rise curve showed a gradual change that could be identified for the first time due to the latent heat absorbed by the material. Table 1 shows some of the process data captured within this critical time window.

[0032] Table 1: Dynamic Comparison of Core Decision Parameters and Temperature Measurements.

[0033]

[0034] See Table 1 for parameters. The response to phase transition events stems from the fact that the absorption of latent heat of phase transition by the material directly and instantaneously alters the macroscopic energy balance between the system's input electrical power and output thermal radiation. This change is a direct reflection of the material's internal state and is therefore unaffected by thermal conduction delays. In contrast, thermocouple temperature measurement relies on the physical conduction of heat from the melt core through the crucible wall to the thermocouple's sensing end. The inherent thermal resistance and thermal capacity of this process are the fundamental reasons for its response lag. In this experiment, the system of this invention made a phase transition judgment at 372 seconds, while the thermocouple did not show a clear response until 385 seconds, resulting in a 13-second time difference. The experimental data shows that the judgment mechanism based on the thermo-electric response ratio time derivative can identify the critical phase transition state of the material earlier than the traditional contact temperature measurement method. The response time advance it provides reserves the necessary process window for the control system to perform subsequent precise operations such as power adjustment. This data directly verifies the role of this mechanism in improving the timeliness of phase transition process control.

[0035] Example 3: This example combines Figures 1 to 3 This paper describes a closed-loop control system for a vacuum induction heating process that applies an industrial mechanism model, such as... Figure 1 As shown, the core of the system is built around a vacuum induction heating furnace, which is connected to a power monitoring unit and a heat flow sensing unit, respectively, to acquire the input active power in real time. And measuring the total external radiative heat flux. The two data paths mentioned above are simultaneously input to the embedded decision module, which calculates the thermoelectric response ratio. and its time derivative This is used to determine whether the material is in a critical phase transition state. After identifying the critical state, the embedded decision module performs a multi-source signal cross-validation process through a decision arbitrator. If the verification is successful, the power execution unit switches the heating power to a preset heat preservation power value to maintain the material in a suitable state. To enhance the system's operating condition perception capability, the vibration sensing unit collects the system's physical state and calculates the vibration entropy to assess the operating condition; this information is also input into the decision threshold dynamic self-compensation module, which is based on the phase transition inertia factor. To adjust the threshold and adapt to different material properties, the system also includes a current ripple analysis unit that uses existing feedback signals to identify phase transition slight change characteristics for cross-validation with embedded decision results. In addition, to achieve online sensing of melt viscosity, an online melt viscosity inversion module is also included. By superimposing micro-perturbation signals into the input power and performing superposition and response analysis, feature values ​​related to melt rheology are extracted and further fed back into the decision system.

[0036] like Figure 2As shown, the current ripple amplitude is the measured data curve corresponding to the left vertical axis, in amperes (A), representing the high-frequency ripple amplitude of the power supply feedback current in the induction heating system, which serves as one of the electrical fingerprint signals for judging material phase transitions. The phase transition characteristic index, shown on the right vertical axis, is an auxiliary analysis parameter that reflects the intensity of distortion modes related to material phase transition behavior extracted from the current ripple signal. The phase transition identification threshold is a system-preset decision benchmark value used to identify whether the material has reached the critical state of phase transition, and is used to compare and judge with the current ripple and its characteristic index. The horizontal axis in the figure is time (seconds), representing the time evolution of the entire heating process. As can be seen from the figure, the current ripple amplitude shows a significant upward slope at approximately 372 seconds, reaching its peak at around 378 seconds, and then slowly decreasing. At the same time, the phase transition characteristic index also rises rapidly in a similar period and exceeds the phase transition identification threshold, verifying that the system successfully identifies the occurrence of a phase transition event.

[0037] like Figure 3 As shown, firstly, the embedded decision module determines the phase transition based on the thermoelectric model, generates a power modulation command, and activates the dual verification mechanism for phase transition judgment in the decision arbitrator. Subsequently, the decision arbitrator opens an arbitration time window and simultaneously requests the current ripple analysis unit to provide a current feedback signal. During this window, the current ripple analysis unit performs high-speed sampling analysis to identify high-frequency ripple characteristics. If a phase transition characteristic is identified, the identification result is sent to the decision arbitrator; if no phase transition characteristic is found, the result is also fed back. The decision arbitrator compares the judgments from the two sources. If the conclusions of the two sources are consistent, the power modulation command is confirmed and sent to the power execution unit, switching to the heat preservation power and completing the power adjustment. If the current ripple analysis unit does not identify a phase transition characteristic, a source conflict is determined, the decision arbitrator rejects the power modulation command, locks the current power, and simultaneously sends a sensor conflict alarm. The power controller maintains the existing power.

[0038] Example 4: Before applying the closed-loop control system of this invention to the precision melting of an amorphous metal alloy with dispersed phase transformation characteristics, the core parameters in its advanced control module need to be calibrated to ensure that the system's decision logic matches the physical properties of the preset material. The phase transformation process of this type of amorphous alloy occurs gradually within a temperature range rather than exhibiting a clear temperature inflection point. This poses a challenge to the effectiveness of the dynamic self-compensation module for the decision threshold in this system. Simultaneously, the high temperature sensitivity of its melt viscosity also requires precise setting of the parameters of the online melt viscosity inversion module. The calibration process first determines the parameters of the online melt viscosity inversion module... The purpose of the micro-perturbation signal parameters of the module is to select the excitation frequency that can generate the highest signal-to-noise ratio thermal response. The specific procedure is as follows: In a reference melting of the amorphous alloy, after it is completely melted and in a stable liquid state, multiple micro-perturbation AC signals with constant amplitude within a preset frequency range are sequentially superimposed on a constant base heat preservation power. The response signal amplitude with the same frequency as each frequency perturbation is recorded synchronously using a heat flow sensing unit and a lock-in amplification algorithm. By analyzing the relationship between perturbation frequency and response amplitude, the frequency point with the largest response amplitude and the most stable signal fluctuation is selected as the curing working frequency of the module for this material.

[0039] Subsequently, the key thresholds in the dynamic self-compensation module for decision thresholds are calibrated. This step begins with setting a preset event change judgment threshold to trigger the module. This threshold is used to identify sudden changes in vibration entropy caused by physical events such as furnace charge collapse. In a benchmark smelting batch, the system continuously records the vibration entropy and its rate of change over time during the process from solid-state heating to complete melting. By statistically analyzing the vibration entropy change rate data during the stable heating and stable melting stages, the mean value is calculated. with standard deviation The preset event change judgment threshold is set to the sum of the mean and a preset integer multiple of the standard deviation, i.e. Among them, the confidence coefficient The value of this threshold is used to ensure that it is higher than the background fluctuation under normal operating conditions; based on this, the core parameter of this module, namely the compensation coefficient in the formula "corrected threshold = original threshold", is adjusted. Calibration is performed, and this coefficient determines the system's compensation strength for phase transition inertia; the calibration process employs a series of comparative melting experiments. The first step involves... Set to zero. At this point, the dynamic compensation function is disabled. A baseline melting process is performed, and the phase transition inertia characterization factor during the phase transition is recorded. The cumulative value and the final ingot quality; the second step is to gradually increase the value by a preset increment step. The value of the compensation coefficient was determined and the ingot was repeatedly melted. After each melting, the amorphous formation degree and microstructure uniformity of the ingot were quantitatively scored through metallographic analysis and density measurement. Finally, the compensation coefficient was determined. - Correlation analysis was performed on the ingot quality score data to select the data that consistently achieved the highest quality score. The values, serving as solidification parameters for this amorphous alloy, are determined through the aforementioned calibration process. Each module's core parameters are assigned specific values ​​related to the preset material's physical properties for subsequent melting control. Simultaneously, compensation coefficients... It is a dimensionless parameter used to adjust the dynamic compensation intensity. Its value is usually a positive value, and the specific value needs to be determined through offline calibration experiments based on the preset characteristics of the smelting material and equipment. A typical calibration process is as described in this embodiment: First, Setting it to zero disables the compensation function and performs a baseline melting operation; subsequently, the value is gradually increased in preset increments (e.g., 0.1). The values ​​are selected and the melting process is repeated multiple times. After each melting, the quality of the final ingot, such as microstructure uniformity and defect rate, is quantitatively scored using metallographic analysis, density measurement, and other methods. Finally, the value corresponding to the highest quality score that can be consistently obtained is selected. The value serves as the solidified working parameter for this preset working condition.

[0040] Example 5: When the closed-loop control system of the present invention is first deployed on a new melting equipment or used to process alloy materials without historical data, a set of pre-calibration procedures needs to be executed to establish the electrical fingerprint of the preset material. This procedure aims to provide the current ripple analysis unit with predetermined phase transformation distortion characteristics for arbitration decision-making. The process is as follows: Select the new alloy sample, and in a device equipped with a high-precision differential scanning calorimeter or a high-temperature dilatometer, determine the temperature points at the start and end of its phase transformation. At the same time, the same batch of samples are synchronously melted on an induction heating power supply of the same model as the main system, and the high-frequency current feedback signal of the entire process is recorded by the current ripple analysis unit. By aligning the precise phase transformation time points measured by the calorimeter or dilatometer with the recorded current ripple data on the time axis and performing correlation analysis, preset harmonic amplitude changes, phase shifts, or nonlinear distortion modes that can be clearly distinguished from the phase transformation behavior of the material are separated. This mode is solidified as the predetermined phase transformation distortion characteristics of the material and stored in the system's internal parameter library for subsequent use.

[0041] After completing the material electrical fingerprint calibration, it is also necessary to perform on-site baseline calibration of the coupling characteristics between the entire control system and the preset melting furnace to avoid systematic deviations introduced by differences in the thermal insulation performance and coil efficiency of different furnace bodies. This calibration procedure uses a standard reference with known and stable physical properties, which is placed in the furnace and subjected to a standard power-up program covering the commonly used operating temperature range. During this process, the system continuously records the input active power. With total external radiative heat flux The corresponding data is used to generate a baseline characteristic curve representing the thermal-electrical response ratio of the preset device's energy transfer efficiency. Based on this, the system normalizes and corrects all decision thresholds based on the thermo-electric response ratio, such as the original threshold used to determine the critical state of phase transition, in order to compensate for the inherent thermal characteristics of the preset hardware platform.

[0042] Example 6: Before the closed-loop control system of this invention is integrated with the smelting furnace hardware and put into production operation, a standardized system deployment pre-calibration and self-test procedure needs to be executed to ensure that the parameters of each sensing and decision-making unit match the preset physical environment. This procedure begins with a pre-run self-test sequence after the system is powered on. This sequence confirms that all signal sources of the power monitoring unit, heat flow sensing unit, vibration sensing unit, and current ripple analysis unit are online and their static outputs are within the preset normal range. Subsequently, the system is based on the thermo-electric response ratio time first derivative. The main decision-making logic remains on standby until input active power is detected. With total external radiative heat flux Once the power reaches and stabilizes at over 90% of the initial heating power platform for a duration exceeding one preset thermal stabilization cycle, the main decision logic is enabled. This mechanism is used to avoid false triggering caused by power ramp-up and transient heat flow during the initial stage of system startup.

[0043] After self-testing, the calibration of sensor physical parameters begins. For the heat flow sensor unit, a laser collimation tool is used to align the central axis of its monitoring field of view with the geometric center of the molten zone inside the crucible. The distance between the sensor and the outer wall of the vacuum chamber is adjusted so that the projected area of ​​the measuring field cone on the surface of the molten zone covers 70% to 80% of the crucible's inner diameter area. Next, the vibration entropy judgment threshold of the working condition pre-classifier is set using data-driven methods. This process is achieved by performing a complete melting cycle on a standard reference material. The system records the vibration entropy data throughout the process. The maximum vibration entropy value recorded during the stable solid-phase heating stage is multiplied by a preset safety factor greater than one and set as the first vibration entropy judgment threshold. Simultaneously, the minimum vibration entropy value recorded during known physical processes such as the initial melting stage or furnace charge collapse is multiplied by a preset safety factor less than one and set as the second vibration entropy judgment threshold.

[0044] Subsequently, based on data from the same reference melting, key parameters of the core decision-making logic were calibrated. The initial threshold used for determining the critical state of phase transition was set as follows: by post-analysis or by referring to an established electrical fingerprint, the phase transition initiation point of this reference melting was located, and then the pre-defined time window before and after that initiation point was analyzed. The curve is obtained by multiplying the peak value of the absolute value of the negative slope of the curve during the phase transition by a preset scaling factor between zero and one. The result is set as the original threshold of the system. The preset arbitration time window required by the decision arbitrator is determined according to the inherent operation cycle of the current ripple analysis unit. The duration is set as a preset multiple of the time required for the analysis unit to complete one high-frequency data acquisition, Fourier transform and feature matching. This setting is intended to cover the verification return time of electrical fingerprints. All of these are extended implementation methods known to those skilled in the art.

[0045] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A closed-loop control system for a vacuum induction heating process using an industrial mechanism model, characterized in that, include: A power monitoring unit is configured to acquire the input active power of the induction coil in real time; A heat transfer sensor unit is configured to measure the total external radiative heat flux outside the vacuum chamber in a non-contact manner. The heat transfer sensor unit is installed on the outer wall of the vacuum chamber and its monitoring area is directly facing the heating and melting zone. An embedded decision module is configured to calculate the heat-electric response ratio in real time based on the input active power and the total external radiative heat flux. The heat-electric response ratio is the ratio of the total external radiative heat flux to the input active power. Calculate the time first derivative of the thermo-electric response ratio; And based on the comparison between the first time derivative of the thermo-electric response ratio and zero, it is determined whether the material has entered the critical state of phase transition and a power modulation command is generated. The condition for determining that the material has entered the critical state of phase transition is that the first time derivative is less than or equal to a preset determination threshold. A power execution unit is configured to switch the input active power to a preset insulation power value in response to a power modulation command.

2. The closed-loop control system for a vacuum induction heating process using an industrial mechanism model as described in claim 1, characterized in that, The embedded decision module is also configured to: if the time first derivative of the thermo-electric response ratio is continuously greater than zero, then the system is determined to be in a stable heating stage; the preset judgment threshold is a value greater than zero and less than a first predetermined small value.

3. The closed-loop control system for a vacuum induction heating process using an industrial mechanism model as described in claim 1, characterized in that, The system also includes a vibration sensing unit configured to acquire the vibration spectrum characteristics of the vacuum induction heating system at different operating stages and calculate the vibration entropy based on the vibration spectrum characteristics.

4. A closed-loop control system for a vacuum induction heating process using an industrial mechanism model as described in claim 3, characterized in that, The vibration sensing unit includes: a piezoelectric vibration sensor, which is adsorbed on the support structure of the vacuum chamber for collecting vibration signals; and a vibration signal processing unit configured to perform a fast Fourier transform on the vibration signal and calculate the vibration entropy of the vibration signal.

5. A closed-loop control system for a vacuum induction heating process using an industrial mechanism model as described in claim 4, characterized in that, The system also includes a working condition pre-classifier, configured as follows: when the vibration entropy is lower than the first vibration entropy judgment threshold, the system is determined to be in a spectral ordered state, and the embedded decision module increases the weight of its decision based on the total external radiant heat flux; and when the vibration entropy is higher than the second vibration entropy judgment threshold, the system is determined to be in a spectral chaotic state, and the embedded decision module decreases the weight of its decision based on the total external radiant heat flux.

6. A closed-loop control system for a vacuum induction heating process using an industrial mechanism model according to claim 1, characterized in that, The system also includes a current ripple analysis unit, configured to analyze the predetermined phase transition distortion characteristics of the high-frequency current ripple of the current feedback signal by sampling the existing current feedback signal in the power controller at high speed, so as to determine the impedance change caused by the material phase transition.

7. A closed-loop control system for a vacuum induction heating process using an industrial mechanism model as described in claim 6, characterized in that, The system also includes a decision arbitrator, configured such that: when the embedded decision module generates a power modulation command, the decision arbitrator is activated; within the preset arbitration time window after activation, if the current ripple analysis unit simultaneously identifies a predetermined phase transition distortion characteristic, the decision arbitrator confirms the power modulation command and allows it to be executed. If the current ripple analysis unit fails to identify the predetermined phase transition distortion characteristics, the decision arbitrator will reject the power modulation command, temporarily lock the power, and issue a sensor source conflict alarm to the system.

8. A closed-loop control system for a vacuum induction heating process using an industrial mechanism model according to claim 5, characterized in that, The system also includes a dynamic self-compensation module for decision thresholds. This module is configured to dynamically adjust the decision threshold for determining the critical state of phase transition based on the integral value of the time change rate of the thermo-electric response ratio when the rate of change of vibration entropy exceeds the preset event change judgment threshold.

9. A closed-loop control system for a vacuum induction heating process using an industrial mechanism model as described in claim 8, characterized in that, The decision threshold dynamic self-compensation module is configured to dynamically adjust the decision threshold according to the following formula: in, It is the initial decision threshold set to determine the critical state of phase transition; This is an adjustable compensation coefficient; It is a phase change inertia characterization factor calculated based on the integral of the time first derivative of the thermo-electric response ratio.

10. A closed-loop control system for a vacuum induction heating process using an industrial mechanism model according to claim 1, characterized in that, The system also includes an online melt viscosity inversion module, configured to superimpose a micro-perturbation AC signal with a fixed amplitude and frequency onto the input active power of the induction coil; Furthermore, by analyzing the amplitude of the response signal that is at the same frequency as the micro-disturbance signal in the total external radiative heat flux, the signal or data characterizing the change in melt viscosity can be retrieved in real time.

Images