Adaptive Control System and Method for Energy-Saving Parameters in Circuit Board Metal Melting

Abstract

This invention belongs to the field of circuit board metal melting control technology. It discloses an adaptive control system and method for energy-saving parameters in circuit board metal melting. The method includes: real-time monitoring and fluctuation analysis of the heating circuit voltage collected per unit time during the circuit board metal melting process to obtain the voltage fluctuation level; processing the melting state data collected per unit time during the circuit board metal melting process to obtain a melting state characteristic data set and identifying the current melting stage; when the voltage fluctuation level is greater than or equal to a preset voltage fluctuation level threshold, generating an energy-saving strategy control identifier by combining the current melting stage with the predicted melting stage for the next unit time; and dynamically adjusting the energy-saving control parameters based on the voltage fluctuation level, the melting state characteristic data set, and the energy-saving strategy control identifier. This application enhances the intelligent adaptive control level of energy-saving parameters in circuit board metal melting, enabling energy-saving control to shift from passive execution to proactive response.

Description

Technical Field

[0001] This invention relates to the field of circuit board metal smelting control technology, and more specifically, to an adaptive control system and method for energy-saving parameters in circuit board metal smelting. Background Technology

[0002] Circuit board metal smelting and recycling technology, as a key link in achieving efficient resource utilization of electronic waste, is gradually being widely applied in various precision recycling scenarios. To achieve high-purity separation of metal elements and energy consumption control, current mainstream circuit board smelting systems generally employ electromagnetic heating, pulse heating, or frequency conversion heating, supplemented by temperature feedback and energy consumption monitoring modules for basic adjustments to meet the initial requirements of process temperature control and energy efficiency indicators. Simultaneously, some systems have introduced staged heating strategies or energy-saving control logic based on process rules, attempting to optimize overall power consumption while meeting smelting requirements.

[0003] However, in actual operation, the smelting environment is often affected by multiple disturbances such as unstable power supply, external interference fluctuations, and differences in the thermal properties of circuit board raw materials, resulting in delayed temperature control response and drastic energy consumption fluctuations in the smelting zone. Existing energy-saving control strategies mostly rely on static configuration parameters or fixed threshold rules, lacking the ability to continuously perceive and analyze the dynamic evolution trend of the smelting process, and failing to consider the linkage mechanism between voltage fluctuations, a key external variable, and energy-saving control. Furthermore, energy-saving parameters are usually set to pre-calibrated fixed values, lacking a real-time coordinated adjustment mechanism based on the smelting stage and voltage fluctuation level. This leads to problems such as untimely response, overly coarse strategies, and insufficient adjustment precision in energy-saving control, making it difficult to meet the actual needs for intelligent and flexible energy-saving control under complex operating conditions.

[0004] In view of this, the present invention proposes an adaptive control system and method for energy-saving parameters in circuit board metal smelting to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and achieve the above objectives, the present invention provides the following technical solution: an adaptive control method for energy-saving parameters in circuit board metal smelting, comprising:

[0006] Real-time monitoring and fluctuation analysis of the heating circuit voltage collected per unit time during the circuit board metal smelting process are performed to obtain the voltage fluctuation level;

[0007] The melting state data collected per unit time during the circuit board metal melting process are processed to obtain a set of melting state characteristic data and identify the current melting stage.

[0008] When the voltage fluctuation level is greater than or equal to the preset voltage fluctuation level threshold, an energy-saving strategy control identifier is generated by combining the current smelting stage with the predicted smelting stage for the next unit time.

[0009] Energy-saving control parameters are dynamically adjusted based on voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifiers.

[0010] Furthermore, methods for dynamically adjusting energy-saving control parameters based on voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifiers include:

[0011] The energy-saving strategy control identifier is matched with the pre-built control identifier numerical conversion table to obtain the digitized energy-saving strategy control identifier, which is recorded as the energy-saving strategy control identifier value.

[0012] The voltage fluctuation level, smelting state characteristic data set, and energy-saving strategy control identifier value are input into the energy-saving parameter setting model to obtain the energy-saving adjustment parameter set; the energy-saving adjustment parameter set includes the maximum heating power, pulse heating cycle, energizing time duty cycle, temperature response rate upper limit, and heat power adjustment delay time within the stage.

[0013] A set of corresponding control commands is generated based on the set of energy-saving adjustment parameters; the control commands in the set of control commands are distributed to the corresponding actuators for execution, thereby completing the dynamic adjustment of the energy-saving control parameters.

[0014] Furthermore, the method for generating the energy-saving strategy control label includes:

[0015] Pre-constructed energy-saving control stage set;

[0016] The set of smelting state characteristic data and the current smelting stage are input into the smelting stage prediction model to obtain the predicted smelting stage for the next unit of time.

[0017] If the current smelting stage exists in the energy-saving control stage set, then the current smelting stage is an energy-saving control stage; if the predicted smelting stage for the next unit time exists in the energy-saving control stage set, then the predicted smelting stage for the next unit time is an energy-saving control stage.

[0018] If both the current smelting stage and the predicted smelting stage for the next unit of time are energy-saving control stages, then the energy-saving strategy control flag will be set to the maintenance adjustment state flag.

[0019] If the current smelting stage is an energy-saving control stage, and the predicted smelting stage for the next unit of time is a non-energy-saving control stage, then the energy-saving strategy control flag will be set to the flag indicating that the adjustment state is about to end.

[0020] If the current smelting stage is a non-energy-saving control stage, and the predicted smelting stage for the next unit of time is an energy-saving control stage, then the energy-saving strategy control flag will be set to the advance preparation adjustment status flag.

[0021] If both the current smelting stage and the predicted smelting stage for the next unit of time are non-energy-saving control stages, then the energy-saving strategy control flag will be set to the no-adjustment status flag.

[0022] Furthermore, the method for obtaining the set of smelting state characteristic data includes:

[0023] Acquire smelting state data collected per unit time, including smelting temperature, metal conductivity, heating power, metal flow rate in the smelting zone, and metal liquid level.

[0024] Divide the unit time into H time points, construct a set of melting temperatures for the H time points, construct a set of metal conductivity for the H time points, construct a set of heating power for the H time points, construct a set of metal flow rates for the H melting zones for the H time points, and construct a set of metal liquid level heights for the H time points.

[0025] The following calculations were performed based on the following sets of melting temperatures: melting temperature change rate, melting temperature change rate, metal conductivity change rate, heating power change rate, metal flow velocity change amplitude, and molten metal level height.

[0026] The set of melting state characteristic data is constructed by combining the sets of melting temperature change rate, metal conductivity change rate, heating power change rate, metal flow rate change amplitude, and liquid level height change amplitude.

[0027] Furthermore, current methods for identifying the smelting stage include:

[0028] Input the set of smelting state characteristic data into the smelting stage diagnostic model to obtain the current smelting stage.

[0029] Furthermore, the method for obtaining the voltage fluctuation level includes:

[0030] Divide the unit time into H time points, and construct the initial heating circuit voltage set from the heating circuit voltages at the H time points;

[0031] A preset set of sliding window pairs; the set of sliding window pairs includes G sliding window pairs, the first digit of the sliding window pair is the sequence number, and the second digit of the sliding window pair is the length of the sliding window;

[0032] The initial heating circuit voltage set is divided based on the sliding window binary set to obtain G window sliding voltage sets;

[0033] Feature extraction is performed on each of the G window sliding voltage sets to obtain G voltage diagnostic data sets;

[0034] A comprehensive voltage diagnosis dataset is constructed from G sets of voltage diagnosis data; the comprehensive voltage diagnosis dataset is then input into the comprehensive voltage diagnosis model to obtain the corresponding comprehensive voltage diagnosis score.

[0035] The voltage comprehensive diagnostic score is matched with the pre-constructed diagnostic score-fluctuation level mapping table to obtain the corresponding voltage fluctuation level.

[0036] Furthermore, the methods for obtaining the set of G window sliding voltages include:

[0037] S101: Let the initial value of g be 1, and the range of g is from 1 to G;

[0038] S102: Obtain the g-th sliding window binary from the set of sliding window binary; obtain the sliding window length corresponding to the g-th sliding window binary; extract the voltages sequentially from the initial heating circuit voltage set in units of sliding window length to form the g-th window sliding voltage set;

[0039] S103: Let g = g + 1. If g is less than or equal to G, return to S102 to continue execution. If g is greater than G, end the current process.

[0040] Furthermore, the methods for obtaining the G voltage diagnostic data sets include:

[0041] S201: Let the initial value of g be 1, and the range of g is from 1 to G;

[0042] S202: Obtain the set of sliding voltages for the g-th window; denote the number of window sliding voltage subsets in the set of sliding voltages for the g-th window as... The number of heating circuit voltages in the window sliding voltage subset is equal to the corresponding sliding window length.

[0043] S203: Let the loop index variable... The initial value is 1. The value range is 1 to ;

[0044] S204: Obtain the first... in the set of sliding voltages of the g-th window. A window-sliding voltage subset; calculate the steady-state voltage reference value, transient disturbance response amplitude, segment fluctuation discrete intensity, and peak disturbance sensitivity of the window-sliding voltage subset;

[0045] Add the steady-state voltage reference value to the steady-state voltage reference value set; add the transient disturbance response amplitude to the transient disturbance response amplitude set; add the segment fluctuation discrete intensity to the segment fluctuation discrete intensity set; add the peak disturbance sensitivity to the peak disturbance sensitivity set;

[0046] S205: Order ,like Less than or equal to If so, return to S204 and continue execution; Greater than Then, the voltage diagnostic data set is constructed from the voltage steady-state reference value set, the transient disturbance response amplitude set, the segment fluctuation discrete intensity set, and the peak disturbance sensitivity set into the g-th window sliding voltage set, and g = g + 1 is set. If g is less than or equal to G, the process returns to S202 to continue execution. If g is greater than G, the current process ends.

[0047] Furthermore, the training method for the smelting stage diagnostic model includes:

[0048] A pre-constructed smelting stage diagnostic dataset is constructed, which includes P sets of smelting stage diagnostic data and the smelting stages corresponding to the P sets of smelting stage diagnostic data, where P is a positive integer; the smelting stage diagnostic data includes a set of smelting state feature data; the smelting stage diagnostic dataset is divided into a training set and a validation set, the training set is used for learning the parameters of the smelting stage diagnostic model, and the validation set is used for real-time monitoring of the generalization performance and overfitting degree of the smelting stage diagnostic model.

[0049] A deep neural network with a multilayer perceptron as its core is used as the diagnostic model for the melting stage. The diagnostic data for the melting stage is standardized and vectorized before being input into the deep neural network, which consists of an input layer, hidden layers, and an output layer. Each hidden layer uses a nonlinear activation function to extract features, and the output layer uses a softmax activation function to obtain the probability distribution corresponding to each melting stage. Finally, the melting stage corresponding to the highest probability is taken as the prediction result of the melting stage diagnostic model. During training, the cross-entropy loss function is used as the optimization objective, and a gradient descent-type optimization algorithm is used to update the network weights. An early stopping strategy is set: when the prediction accuracy on the validation set reaches or exceeds a preset threshold, the melting stage diagnostic model is determined to have converged and training is terminated.

[0050] An adaptive control system for energy-saving parameters in circuit board metal smelting, used to implement the adaptive control method for energy-saving parameters in circuit board metal smelting, includes:

[0051] The voltage fluctuation processing module is used to monitor and analyze the voltage fluctuation of the heating circuit collected per unit time during the circuit board metal melting process in real time, and obtain the voltage fluctuation level.

[0052] The melting status processing module is used to process the melting status data collected per unit time during the circuit board metal melting process, obtain a set of melting status feature data, and identify the current melting stage.

[0053] The comprehensive diagnostic module is used to generate an energy-saving strategy control identifier when the voltage fluctuation level is greater than or equal to the preset voltage fluctuation level threshold, by combining the current smelting stage with the predicted smelting stage for the next unit time.

[0054] The energy-saving strategy adjustment module dynamically adjusts energy-saving control parameters based on voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifiers.

[0055] Compared with existing technologies, the technical effects and advantages of the adaptive control system and method for energy-saving parameters in circuit board metal smelting of the present invention are as follows:

[0056] The adaptive control system and method for energy-saving parameters in circuit board metal melting provided in this application, through the construction of a multi-module collaborative structure including voltage acquisition, fluctuation level assessment, melting state identification, energy-saving strategy control identifier generation, and energy-saving adjustment parameter setting, can realize a dynamic energy-saving adjustment response mechanism based on voltage fluctuation level and melting stage state. The system first divides the heating circuit voltage into multiple scales using a sliding window, extracting composite voltage features including steady-state voltage reference value, transient disturbance response amplitude, segment fluctuation dispersion intensity, and peak disturbance sensitivity, constructing a comprehensive voltage diagnostic score and quantifying it as a fluctuation level; then, combining multi-dimensional process parameters such as melting temperature, metal conductivity, heating power, metal flow rate in the melting zone, and metal liquid level height per unit time, it extracts melting state features and identifies the current and predicted melting stages; based on this, it constructs an energy-saving strategy control identifier and its numerical mapping relationship for energy-saving stage control strategies, forming a refined energy-saving strategy control judgment logic. Ultimately, the system combines the current voltage fluctuation level with the smelting state characteristic data set to obtain a set of energy-saving adjustment parameters. Through the generation of control commands and linkage with the actuator, the parameters are dynamically adjusted, achieving full-process, closed-loop, and dynamic adjustment of energy-saving control parameters.

[0057] Compared to existing technologies, this invention overcomes the limitations of existing energy-saving strategies that rely on static configuration or preset switching based on a single environmental variable. It can accurately calculate and flexibly configure optimal energy-saving parameters according to the actual evolution of the melting process and real-time disturbances in the external voltage environment, thereby effectively improving the accuracy and response speed of energy-saving regulation under unstable operating conditions. Furthermore, the "control identifier-driven - model-guided - parameter closed-loop adjustment" mode of this application shifts energy-saving control from passive execution to proactive prediction and forward-looking response. This not only improves energy efficiency and power control safety but also enhances the system's intelligence and adaptability. It represents a significant improvement over existing technologies in terms of process control continuity, adjustment strategy precision, and energy consumption control flexibility. While ensuring the stability of circuit board metal melting, it achieves superior energy consumption control, demonstrating significant engineering application value. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of the adaptive control system for energy-saving parameters in circuit board metal smelting according to Embodiment 1 of the present invention;

[0059] Figure 2 This is a schematic diagram of the adaptive control system for energy-saving parameters in circuit board metal smelting according to Embodiment 2 of the present invention;

[0060] Figure 3 This is a flowchart of the adaptive control method for energy-saving parameters in circuit board metal smelting according to Embodiment 3 of the present invention;

[0061] Figure 4 A flowchart illustrating the method for obtaining voltage fluctuation levels;

[0062] Figure 5 A flowchart illustrating the method for comprehensively acquiring smelting state characteristic data sets and smelting stages;

[0063] Figure 6 A flowchart illustrating the method for dynamically adjusting energy-saving control parameters. Detailed Implementation

[0064] The technical solutions of the embodiments of the present invention will be described in detail, clearly, and completely below with reference to the accompanying drawings. It should be particularly noted that the specific embodiments described below are only for better illustrating and explaining the technical solutions of the present invention, and are intended to enable those skilled in the art to better understand and implement the present invention, and should not be construed as limiting the scope of protection of the present invention. Without departing from the spirit and substance of the present invention, those skilled in the art can modify, adjust, or make equivalent substitutions based on the content disclosed in the present invention, and these should all be considered within the scope of protection of the present invention.

[0065] Example 1:

[0066] Please see Figure 1 As shown, this embodiment discloses an adaptive control system for energy-saving parameters in circuit board metal smelting, including a voltage fluctuation processing module, a smelting status processing module, a comprehensive diagnostic module, and an energy-saving strategy adjustment module. Each module is connected via wired and / or wireless means to achieve data transmission.

[0067] The voltage fluctuation processing module is used to monitor and analyze the voltage fluctuation of the heating circuit collected per unit time during the circuit board metal smelting process in real time, and obtain the voltage fluctuation level.

[0068] like Figure 4 As shown, the method for obtaining the voltage fluctuation level includes:

[0069] Divide the unit time into H time points, and construct the initial heating circuit voltage set from the heating circuit voltages at the H time points;

[0070] A preset set of sliding window pairs; the set of sliding window pairs includes G sliding window pairs, the first digit of the sliding window pair is the sequence number, and the second digit of the sliding window pair is the length of the sliding window;

[0071] The initial heating circuit voltage set is divided based on the sliding window binary set to obtain G window sliding voltage sets;

[0072] Feature extraction is performed on each of the G window sliding voltage sets to obtain G voltage diagnostic data sets;

[0073] A comprehensive voltage diagnosis dataset is constructed from G sets of voltage diagnosis data; the comprehensive voltage diagnosis dataset is then input into the comprehensive voltage diagnosis model to obtain the corresponding comprehensive voltage diagnosis score.

[0074] The voltage comprehensive diagnostic score is matched with the pre-constructed diagnostic score-fluctuation level mapping table to obtain the corresponding voltage fluctuation level.

[0075] Methods for obtaining the set of G window sliding voltages include:

[0076] S101: Let the initial value of g be 1, and the range of g is from 1 to G;

[0077] S102: Obtain the g-th sliding window binary from the set of sliding window binary; obtain the sliding window length corresponding to the g-th sliding window binary; extract the voltages sequentially from the initial heating circuit voltage set in units of sliding window length to form the g-th window sliding voltage set;

[0078] S103: Let g = g + 1. If g is less than or equal to G, return to S102 to continue execution. If g is greater than G, end the current process.

[0079] It should be noted that, since voltage fluctuations during the smelting process have nonlinear, multi-scale, and interval characteristics, a single fixed window is prone to missing local abrupt changes or periodic variations. By constructing a multi-scale sliding window, short-term peak fluctuations and medium-term trend fluctuation information can be extracted simultaneously, improving the completeness and response sensitivity of fluctuation modeling.

[0080] Methods for obtaining G sets of voltage diagnostic data include:

[0081] S201: Let the initial value of g be 1, and the range of g is from 1 to G;

[0082] S202: Obtain the set of sliding voltages for the g-th window; denote the number of window sliding voltage subsets in the set of sliding voltages for the g-th window as... The number of heating circuit voltages in the window sliding voltage subset is equal to the corresponding sliding window length.

[0083] S203: Let the loop index variable... The initial value is 1. The value range is 1 to ;

[0084] S204: Obtain the first... in the set of sliding voltages of the g-th window. A window-sliding voltage subset; calculate the steady-state voltage reference value, transient disturbance response amplitude, segment fluctuation discrete intensity, and peak disturbance sensitivity of the window-sliding voltage subset;

[0085] Add the steady-state voltage reference value to the steady-state voltage reference value set; add the transient disturbance response amplitude to the transient disturbance response amplitude set; add the segment fluctuation discrete intensity to the segment fluctuation discrete intensity set; add the peak disturbance sensitivity to the peak disturbance sensitivity set;

[0086] S205: Order ,like Less than or equal to If so, return to S204 and continue execution; Greater than Then, the voltage diagnostic data set is constructed from the voltage steady-state reference value set, the transient disturbance response amplitude set, the segment fluctuation discrete intensity set, and the peak disturbance sensitivity set into the g-th window sliding voltage set, and g = g + 1 is set. If g is less than or equal to G, the process returns to S202 to continue execution. If g is greater than G, the current process ends.

[0087] The method for calculating the steady-state voltage reference value includes:

[0088] ;

[0089] in, For the set of sliding voltages in the g-th window, the first... The steady-state voltage reference value of a subset of sliding voltages within a window. Let g be the length of the sliding window corresponding to the g-th sliding voltage set. For the first The voltage of the i-th heating circuit in the window sliding voltage subset, where i is the index variable of the summation formula.

[0090] The method for calculating the transient interference response amplitude includes:

[0091] ;

[0092] in, For the set of sliding voltages in the g-th window, the first... Transient disturbance response amplitude of a subset of sliding voltage windows For the first The maximum heating circuit voltage in the window sliding voltage subset. For the first The minimum heating circuit voltage in the subset of sliding voltage windows.

[0093] The method for calculating the discrete intensity of the segment fluctuation includes:

[0094] ;

[0095] in, For the set of sliding voltages in the g-th window, the first... The discrete intensity of segment fluctuations in a subset of sliding voltages within a window.

[0096] The method for calculating the spike interference sensitivity includes:

[0097] ;

[0098] in, For the set of sliding voltages in the g-th window, the first... Sensitivity to spike interference of a subset of sliding voltages within a window.

[0099] It should be noted that the steady-state voltage reference value is used to characterize the concentration level of voltage fluctuations in the heating circuit per unit time, reflecting the steady-state reference range of the system under the current voltage disturbance; the transient disturbance response amplitude is used to measure the maximum disturbance amplitude difference of the voltage signal within the sliding window, reflecting the upper and lower boundary range of the system voltage deviation under short-term disturbances; the segment fluctuation dispersion intensity is used to describe the overall dispersion of the voltage point within the current window relative to the steady-state reference value, reflecting the intensity level of voltage fluctuations; the spike disturbance sensitivity is used to determine whether there are sharp change points in the window data, thereby determining whether the voltage is amplified sensitive to high-frequency disturbances. The larger the spike disturbance sensitivity value, the more likely there are spikes far from the mean in the voltage distribution, indicating that the system has a strong amplified response to high-frequency disturbances or transient anomalies under this time window.

[0100] The training method for the voltage comprehensive diagnostic model includes:

[0101] A voltage diagnostic dataset is pre-constructed, which includes E sets of voltage diagnostic data and corresponding voltage comprehensive diagnostic scores for the E sets of voltage diagnostic data, where E is a positive integer; the voltage diagnostic data includes a voltage comprehensive diagnostic dataset; the voltage diagnostic dataset is divided into a training set and a validation set, the training set is used for learning the parameters of the voltage comprehensive diagnostic model, and the validation set is used for real-time monitoring of the generalization performance and overfitting of the voltage comprehensive diagnostic model;

[0102] A deep neural network with a multilayer perceptron as its core is used as the voltage comprehensive diagnostic model. The voltage diagnostic data is standardized and vectorized before being input into the deep neural network, which consists of an input layer, hidden layers, and an output layer. Each hidden layer uses a nonlinear activation function to extract features, and the output layer uses a softmax activation function to obtain the probability distribution corresponding to each voltage comprehensive diagnostic score. Finally, the voltage comprehensive diagnostic score corresponding to the highest probability is taken as the prediction result of the voltage comprehensive diagnostic model. During training, the cross-entropy loss function is used as the optimization objective, and a gradient descent-type optimization algorithm is used to update the network weights. An early stopping strategy is set: when the prediction accuracy on the validation set reaches or exceeds a preset threshold, the voltage comprehensive diagnostic model is determined to have converged and training is terminated.

[0103] It should be noted that the voltage comprehensive diagnostic score is used to comprehensively and quantitatively evaluate the voltage fluctuation of the heating circuit within a unit time period, reflecting the current voltage stability of the system and providing a basis for subsequent energy-saving strategy selection. The value range of the voltage comprehensive diagnostic score is set according to the standardization mechanism of the voltage comprehensive diagnostic model. Preferably, the value range of the voltage comprehensive diagnostic score is defined as follows: To achieve a direct conversion between the scoring results and voltage fluctuation levels, an example of the diagnostic scoring-fluctuation level mapping table is shown in Table 1:

[0104] Table 1 Diagnostic Score-Fluctuation Level Mapping Table

[0105]

[0106] It should be noted that during the metal smelting process of circuit boards, the smelting heating equipment typically outputs pulse power according to a set temperature control curve to achieve phased temperature control, such as the initial melting and heating stage, the isothermal melting stage, and the cooling and holding stage. Furthermore, the temperature control curve is highly coupled with energy-saving strategy parameters (such as pulse width, duty cycle, and feedback cycle). The smelting heating equipment is highly dependent on the power supply voltage. If the voltage fluctuates, the heat output of the smelting heating equipment will exhibit a non-linear shift, leading to abnormal temperature fluctuations in the smelting zone. When voltage fluctuations cause abnormal temperature fluctuations in the smelting zone, it will subsequently affect multiple dimensions, including heater power output, thermal field stability, control algorithm accuracy, and system resource load. For example, localized overheated areas will continue to dissipate heat, causing system heat loss; unmelted areas will be forced to undergo repeated heating, lengthening the smelting cycle; thus, the energy-saving strategy's control rhythm will fail, meaning the energy-saving strategy cannot maintain its optimal execution state. This not only increases total energy consumption but also compromises the consistency of the smelting process's quality.

[0107] Therefore, this application creatively introduces a voltage fluctuation level identification and strategy dynamic switching mechanism to effectively overcome the problem of increased energy consumption and decreased energy efficiency caused by heating circuit voltage during the circuit board metal melting process.

[0108] The melting state processing module is used to process the melting state data collected per unit time during the circuit board metal melting process, obtain a set of melting state characteristic data, and identify the current melting stage. The melting stage refers to the physical evolution stage of the circuit board metal material during the transformation from solid to liquid under high temperature. For example, the melting stage includes the initial melting heating stage, the isothermal melting stage, the impurity removal and heat preservation stage, and the cooling and crystallization stage.

[0109] The method for obtaining the set of smelting state characteristic data includes:

[0110] Acquire smelting state data collected per unit time, including smelting temperature, metal conductivity, heating power, metal flow rate in the smelting zone, and metal liquid level.

[0111] Divide the unit time into H time points, construct a set of melting temperatures for the H time points, construct a set of metal conductivity for the H time points, construct a set of heating power for the H time points, construct a set of metal flow rates for the H melting zones for the H time points, and construct a set of metal liquid level heights for the H time points.

[0112] The following calculations were performed based on the following sets of melting temperatures: melting temperature change rate, melting temperature change rate, metal conductivity change rate, heating power change rate, metal flow velocity change amplitude, and molten metal level height.

[0113] The set of melting state characteristic data is constructed by combining the sets of melting temperature change rate, metal conductivity change rate, heating power change rate, metal flow rate change amplitude, and liquid level height change amplitude.

[0114] Current methods for identifying the smelting stage include:

[0115] Input the set of smelting state characteristic data into the smelting stage diagnostic model to obtain the current smelting stage.

[0116] It should be noted that the flowchart for the comprehensive acquisition method of smelting state characteristic data set and smelting stage is as follows: Figure 5 As shown. The melting temperature is obtained by thermocouples or non-contact infrared temperature sensors installed inside or on the wall of the melting zone. It reflects the current thermal field intensity of the melting zone and is an important basis for determining the initial melting, stable melting, or cooling stages. The metal conductivity is collected by a conductivity detection unit installed in the electromagnetic excitation circuit or electrode resistance measurement circuit. It can characterize the phase change behavior of the metal material, especially showing significant jump characteristics in the solid-liquid transition critical range, and is a key physical indicator for identifying the melting state. The heating power is obtained by calculating the real-time voltage-current product of the heating system control port, or directly provided by the power control unit, and is used to reflect the current... The system's heat input regulation strategy serves as an auxiliary parameter for determining the heating, isothermal, or heat preservation stages. The metal flow velocity in the melting zone is collected by electromagnetic velocity sensors or laser Doppler velocimeters installed at the melting channel or molten pool outlet. This directly reflects whether the molten metal is in a stable flow dynamic, exhibiting rapid response and high reliability. It is an important indicator for determining the isothermal melting stage or the impurity removal and heat preservation stage. The metal liquid level height is obtained by laser displacement sensors or ultrasonic level gauges vertically installed above the melting zone. Its fluctuation level per unit time can indirectly reflect the activity level of internal metal disturbance, which helps in identifying the dynamic impurity removal stage or the stirring effect stage.

[0117] By acquiring the smelting status data collected per unit time and performing correlation analysis, it is possible to accurately divide the key stages of the smelting process. For example, the smelting stages include the initial melting heating stage, the constant temperature melting stage, the impurity removal and heat preservation stage, and the cooling and crystallization stage, providing a reliable and real-time status input basis for the subsequent energy-saving strategy adjustment module.

[0118] The method for obtaining the set of melting temperature change rates includes:

[0119] Set the loop index variable h, with an initial value of 1 and a value range of 1 to H;

[0120] Traverse the set of smelting temperatures from index h=1 to h=H-1, calculate the rate of change of smelting temperature between the (h+1)th time point and the hth time point in turn, and add the rate of change of smelting temperature to the set of smelting temperature change rates.

[0121] The calculation method for the melting temperature change rate includes:

[0122] ;

[0123] in, The rate of change of melting temperature between the (h+1)th time point and the hth time point. The melting temperature at the (h+1)th time point in the set of melting temperatures. Let h be the melting temperature at the h-th time point in the set of melting temperatures. For the (h+1)th time point, This is the h-th time point.

[0124] The method for obtaining the set of metal conductivity change rates includes:

[0125] Set the loop index variable h, with an initial value of 1 and a value range of 1 to H;

[0126] Traverse the set of metal conductivity from index h=1 to h=H-1, calculate the rate of change of metal conductivity between the (h+1)th time point and the hth time point, and add the rate of change of metal conductivity to the set of metal conductivity change rates.

[0127] The method for calculating the rate of change of the metal conductivity includes:

[0128] ;

[0129] in, The rate of change of the metal conductivity between the (h+1)th time point and the hth time point is... Let H be the metal conductivity at the (h+1)th time point in the set of metal conductivity. Let be the metal conductivity at the h-th time point in the set of metal conductivity.

[0130] The method for obtaining the set of heating power change rates includes:

[0131] Set the loop index variable h, with an initial value of 1 and a value range of 1 to H;

[0132] Traverse the heating power set from index h=1 to h=H-1, calculate the rate of change of heating power between the (h+1)th time point and the hth time point in turn, and add the rate of change of heating power to the set of rate of change of heating power.

[0133] The method for calculating the rate of change of heating power includes:

[0134] ;

[0135] in, The rate of change of heating power between the (h+1)th time point and the hth time point. The heating power at the (h+1)th time point in the heating power set. Let be the heating power at the h-th time point in the heating power set.

[0136] The method for obtaining the set of metal flow velocity variation amplitudes includes:

[0137] Set the loop index variable h, with an initial value of 1 and a value range of 1 to H;

[0138] Traverse the set of metal flow rates in the smelting zone from index h=1 to h=H-1, calculate the change in metal flow rate between the (h+1)th time point and the hth time point in turn, and add the change in metal flow rate to the set of change in metal flow rate.

[0139] The method for calculating the variation range of the metal flow rate includes:

[0140] ;

[0141] in, This represents the change in metal flow velocity between the (h+1)th time point and the hth time point. Let h+1 be the metal flow rate in the smelting zone at the (h+1)th time point in the set of metal flow rates in the smelting zone. Let be the metal flow rate of the smelting zone at the h-th time point in the set of metal flow rates in the smelting zone.

[0142] The method for obtaining the set of liquid level height variation ranges includes:

[0143] Set the loop index variable h, with an initial value of 1 and a value range of 1 to H;

[0144] Traverse the set of liquid metal heights from index h=1 to h=H-1, calculate the change in liquid height between the (h+1)th time point and the hth time point, and add the change in liquid height to the set of change in liquid height.

[0145] The calculation method for the change in liquid level includes:

[0146] ;

[0147] in, This represents the change in liquid level between the (h+1)th time point and the hth time point. This represents the height of the liquid metal at time point h+1 in the set of liquid metal heights. Let be the height of the liquid metal at the h-th time point in the set of liquid metal heights.

[0148] The training method for the diagnostic model of the smelting stage includes:

[0149] A pre-constructed smelting stage diagnostic dataset is constructed, which includes P sets of smelting stage diagnostic data and the smelting stages corresponding to the P sets of smelting stage diagnostic data, where P is a positive integer; the smelting stage diagnostic data is a set of smelting state feature data; the smelting stage diagnostic dataset is divided into a training set and a validation set, the training set is used for learning the parameters of the smelting stage diagnostic model, and the validation set is used for real-time monitoring of the generalization performance and overfitting degree of the smelting stage diagnostic model.

[0150] A deep neural network with a multilayer perceptron as its core is used as the diagnostic model for the melting stage. The diagnostic data for the melting stage is standardized and vectorized before being input into the deep neural network, which consists of an input layer, hidden layers, and an output layer. Each hidden layer uses a nonlinear activation function to extract features, and the output layer uses a softmax activation function to obtain the probability distribution corresponding to each melting stage. Finally, the melting stage corresponding to the highest probability is taken as the prediction result of the melting stage diagnostic model. During training, the cross-entropy loss function is used as the optimization objective, and a gradient descent-type optimization algorithm is used to update the network weights. An early stopping strategy is set: when the prediction accuracy on the validation set reaches or exceeds a preset threshold, the melting stage diagnostic model is determined to have converged and training is terminated.

[0151] It should be noted that, in this embodiment, by constructing sets of melting temperature change rates, metal conductivity change rates, heating power change rates, metal flow rate change amplitudes, and liquid level height change amplitudes, the dynamic evolution behavior of the melting process per unit time can be characterized in multiple dimensions. The change characteristics reflected by each set are highly correlated with the stage transitions in the melting process. Specifically, the melting temperature change rate set can reflect whether the melting system is in the rapid heating, constant temperature maintenance, or cooling stage, and is an important basis for judging the initial melting period, stable melting period, and crystallization period. The metal conductivity change rate set can sensitively capture the solid-liquid phase transition. The sudden change in conductivity is suitable for identifying the melting transition stage and the stable range of phase transformation; the set of heating power change rate directly reflects the current power regulation behavior of the system. High-frequency power fluctuations usually correspond to the unstable range of temperature regulation, while steady-state power output often occurs in the continuous constant temperature control stage; the set of metal flow rate change amplitude is used to reflect the dynamic disturbance characteristics of the melt in the melting chamber. Stable high-amplitude flow usually occurs in the melting and impurity removal stage, while the weakening of fluctuations indicates that the melt is tending to be still or has begun to cool; the set of liquid level change amplitude indirectly characterizes the melt disturbance, bubble release or feeding disturbance, and is an important auxiliary feature for identifying the impurity removal, stirring or material level control process.

[0152] Since the above five sets represent the smelting state from multiple perspectives such as thermodynamic evolution, electrical performance changes, energy consumption response, and fluid disturbance, they have good stage discrimination and physical interpretability. Therefore, they can be used as multi-dimensional feature input sets to build machine learning models. After training on existing stage-labeled datasets, the models can accurately identify the smelting stage corresponding to new input data, thereby establishing a data-driven intelligent stage judgment mechanism and effectively improving the system's adaptive perception and control response capabilities to smelting dynamics.

[0153] The comprehensive diagnostic module is used to generate an energy-saving strategy control indicator when the voltage fluctuation level is greater than or equal to a preset voltage fluctuation level threshold, combining the current smelting stage with the predicted smelting stage for the next unit time. The voltage fluctuation level threshold can be set according to the voltage fluctuation level; for example, referring to the diagnostic score-fluctuation level mapping table in Table 1, the voltage fluctuation level threshold can be set to L3. A higher voltage fluctuation level indicates a greater impact of voltage fluctuations on the energy-saving control strategy.

[0154] The methods for generating energy-saving strategy control labels include:

[0155] Pre-constructed energy-saving control stage set;

[0156] It should be noted that the energy-saving control stage set is used to identify key process stages in the smelting process where energy-saving strategies are suitable for implementation. The construction of this energy-saving control stage set is based on a comprehensive analysis of the physical characteristics and energy consumption distribution of the smelting process, determining the types of stages where power input can be optimized and adjusted without affecting smelting quality and production line stability. Specifically, by collecting historical smelting process temperature curves, power input trajectories, and smelting status records, combined with statistical analysis and expert experience, periods with high energy consumption elasticity, high power load stability, or strong thermal inertia response in each smelting stage are identified. The smelting stages corresponding to these periods are then classified as energy-saving control stages, thus constructing the energy-saving control stage set.

[0157] For example, the energy-saving control stage set includes the initial melting heating stage, where the metal has just begun to heat up but has not yet melted extensively, exhibiting high thermal inertia and a long adjustment window; the isothermal melting stage, where the metal is in a stable molten state, and power can be adjusted to save energy through duty cycle; and the impurity removal and heat preservation stage, where the temperature is maintained at a relatively constant value, the metal remains fluid, and adjustment tolerance is strong. Stages such as the rapid heating start-up stage, the critical phase transition range, and the crystallization and solidification stage, which require high temperature control accuracy and are prone to process abnormalities due to thermal disturbances, are not included in the energy-saving control stage set. This energy-saving control stage set is used to determine whether the current and predicted melting stages are within the energy-saving control stage, providing a basic decision-making basis for joint diagnosis and adaptive adjustment.

[0158] The set of smelting state characteristic data and the current smelting stage are input into the smelting stage prediction model to obtain the predicted smelting stage for the next unit of time.

[0159] If the current smelting stage exists in the energy-saving control stage set, then the current smelting stage is an energy-saving control stage; if the predicted smelting stage for the next unit time exists in the energy-saving control stage set, then the predicted smelting stage for the next unit time is an energy-saving control stage.

[0160] If both the current smelting stage and the predicted smelting stage for the next unit of time are energy-saving control stages, then the energy-saving strategy control flag will be set to the maintenance adjustment state flag.

[0161] If the current smelting stage is an energy-saving control stage, and the predicted smelting stage for the next unit of time is a non-energy-saving control stage, then the energy-saving strategy control flag will be set to the flag indicating that the adjustment state is about to end.

[0162] If the current smelting stage is a non-energy-saving control stage, and the predicted smelting stage for the next unit of time is an energy-saving control stage, then the energy-saving strategy control flag will be set to the advance preparation adjustment status flag.

[0163] If both the current smelting stage and the predicted smelting stage for the next unit of time are non-energy-saving control stages, then the energy-saving strategy control flag will be set to the no-adjustment status flag.

[0164] The training method for the prediction model of the smelting stage includes:

[0165] A pre-constructed smelting stage prediction dataset is prepared, comprising U sets of smelting stage prediction data and the predicted smelting stage for the next unit time corresponding to the U sets of smelting stage prediction data, where U is a positive integer greater than 0. The smelting stage prediction data includes a set of smelting state feature data and the current smelting stage. The smelting stage prediction dataset is divided into a smelting stage prediction data training set and a smelting stage prediction data validation set. The smelting stage prediction data training set is used for parameter learning of the smelting stage prediction model, and the smelting stage prediction data validation set is used for real-time evaluation of the generalization ability of the smelting stage prediction model.

[0166] During the training of the smelting stage prediction model, a deep neural network structure based on a multilayer perceptron is adopted. The smelting stage prediction data is converted into feature vectors as input. The nonlinear features in the data are extracted through the hidden layer, and finally, the softmax activation function is used in the output layer to generate the probability distribution of the predicted smelting stage for the next unit time. The predicted smelting stage for the next unit time corresponding to the highest probability is output as the final prediction result. The training process aims to minimize the cross-entropy loss function. At the same time, an early stopping strategy is introduced to monitor the performance of the smelting stage prediction data validation set. When the prediction accuracy on the smelting stage prediction data validation set reaches the preset accuracy, it is determined that the smelting stage prediction model has converged, and the training stops.

[0167] It should be noted that the energy-saving strategy control flag generation method provided in this embodiment comprehensively judges whether to maintain, terminate, or prepare for energy-saving strategy adjustment, and outputs a clear energy-saving strategy control flag, thereby establishing a smelting energy-saving control scheduling mechanism for voltage fluctuation conditions. This mechanism introduces four types of control signals: "maintain adjustment state flag," "advance termination adjustment state flag," "preparation for adjustment state flag," and "no adjustment required state flag." It can perform forward-looking energy-saving response planning based on the evolution trend of smelting conditions, avoiding the response delay, over-adjustment, or lack of adjustment problems caused by relying on static threshold judgment or delayed feedback triggering in existing technologies. Especially in the smelting process, due to significant differences in the tolerance for energy consumption adjustment at different stages, direct adjustment of energy-saving parameters without a stage differentiation mechanism can easily lead to molten quality fluctuations or power control failure. This application establishes a cross-cycle, cross-state energy-saving adjustment judgment logic by combining real-time stage identification and stage prediction results, improving the timeliness and accuracy of the system's energy-saving strategy scheduling under complex operating conditions. Compared to existing technologies that rely solely on temperature or voltage for adjustment, this application offers greater intelligence, scalability, and energy-saving control security.

[0168] The energy-saving strategy adjustment module dynamically adjusts energy-saving control parameters based on voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifiers to avoid power fluctuation amplification, energy consumption runaway, and smelting quality fluctuations caused by voltage instability.

[0169] like Figure 6 As shown, the method for dynamically adjusting energy-saving control parameters based on voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifiers includes:

[0170] The energy-saving strategy control identifier is matched with the pre-built control identifier numerical conversion table to obtain the digitized energy-saving strategy control identifier, which is recorded as the energy-saving strategy control identifier value.

[0171] The voltage fluctuation level, smelting state characteristic data set, and energy-saving strategy control identifier value are input into the energy-saving parameter setting model to obtain the energy-saving adjustment parameter set; the energy-saving adjustment parameter set includes the maximum heating power, pulse heating cycle, energizing time duty cycle, temperature response rate upper limit, and heat power adjustment delay time within the stage.

[0172] The system generates a corresponding set of control commands based on the set of energy-saving adjustment parameters; the control commands in the set of control commands are distributed to the corresponding actuators for execution, thereby completing the dynamic adjustment of the energy-saving control parameters.

[0173] An example of the control identifier numerical conversion table is shown in Table 2:

[0174] Table 2 Control Identifier Numerical Conversion Table

[0175]

[0176] It should be noted that the maximum heating power refers to the maximum power that the heating device is allowed to output per unit time. This parameter should be lowered when voltage fluctuation levels increase to avoid amplifying power fluctuations and reducing instantaneous energy consumption peaks. The pulse heating cycle refers to the length of the heater's on-off cycle, such as periodic pulse energization. When it is necessary to suppress frequent system adjustments, the cycle can be extended to slow down the response frequency and improve temperature control stability. The duty cycle refers to the proportion of the energization time within the pulse heating cycle. The duty cycle can finely adjust the actual heat energy input ratio, reduce the average heating power, and achieve energy-saving goals.

[0177] The upper limit of the temperature response rate refers to the maximum allowable heating rate of the heating system per unit time. This upper limit restricts the temperature control system from aggressively raising the temperature under unstable voltage conditions, thereby ensuring smelting quality. The intra-stage heat power adjustment delay time refers to the parameter update response lag time during the execution of the energy-saving strategy. Specifically, for example, when the energy-saving strategy control flag is set to "maintain adjustment state," the intra-stage heat power adjustment delay time allows the control command to take effect quickly, ensuring energy efficiency. When the energy-saving strategy control flag is set to "about to end adjustment state," extending the intra-stage heat power adjustment delay time can prevent frequent triggering of adjustment switching due to external disturbances, which could lead to heating power oscillations or repeated parameter adjustments in the temperature control system.

[0178] The training method for the energy-saving parameter setting model includes:

[0179] A pre-constructed energy-saving parameter setting dataset is provided, comprising JN sets of energy-saving parameter setting data and a set of energy-saving adjustment parameters corresponding to the JN sets of energy-saving parameter setting data, where JN is a positive integer greater than 0. The energy-saving parameter setting data includes voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifier values. The energy-saving parameter setting dataset is divided into an energy-saving parameter setting data training set and an energy-saving parameter setting data validation set. The energy-saving parameter setting data training set is used for parameter learning of the energy-saving parameter setting model, and the energy-saving parameter setting data validation set is used for real-time evaluation of the generalization ability of the energy-saving parameter setting model.

[0180] During the training of the energy-saving parameter setting model, a deep neural network structure based on a multilayer perceptron is adopted. The energy-saving parameter setting data is converted into feature vectors as input, and nonlinear features in the data are extracted through the hidden layer. Finally, the softmax activation function is used in the output layer to generate the probability distribution of the energy-saving adjustment parameter set. The energy-saving adjustment parameter set corresponding to the highest probability is output as the final prediction result. The training process aims to minimize the cross-entropy loss function. At the same time, an early stopping strategy is introduced to monitor the performance of the energy-saving parameter setting data validation set. When the prediction accuracy on the energy-saving parameter setting data validation set reaches the preset accuracy, it is determined that the energy-saving parameter setting model has converged, and the training stops immediately.

[0181] It should be noted that, through the dynamic adjustment mechanism of energy-saving control parameters provided in this embodiment, the system matches the energy-saving strategy control identifier with the pre-built control identifier numerical conversion table to obtain the digitized energy-saving strategy control identifier value, and, together with the current voltage fluctuation level and smelting state characteristic data set, inputs the above multi-source information into the energy-saving parameter setting model, thereby obtaining a set of energy-saving adjustment parameters including maximum heating power, pulse heating cycle, energizing time duty cycle, upper limit of temperature response rate, and heat power adjustment delay time within the stage.

[0182] This application further automatically generates a corresponding set of control commands based on the set of energy-saving adjustment parameters, and distributes each control command to the control actuators connected to the heating system, power management unit, pulse modulator, and response delay adjustment unit for precise issuance, realizing full-process, closed-loop, and dynamic adjustment of energy-saving control parameters. The introduction of this mechanism overcomes the limitations of existing technologies that rely on static configuration or preset switching based on a single environmental variable for energy-saving strategies. It can accurately calculate and flexibly configure the optimal energy-saving parameters according to the actual evolution of the smelting state and real-time disturbances in the external voltage environment, thereby effectively improving the accuracy and response speed of energy-saving adjustment in the smelting process under unstable operating conditions.

[0183] Meanwhile, the "control identifier driven - model guided - parameter closed-loop adjustment" mode of this mechanism enables energy-saving control to shift from passive execution to active prediction and forward response. This not only improves energy efficiency and power control safety, but also enhances the system's intelligence level and adaptability. Compared with existing technologies, it has made significant progress in process control continuity, energy consumption control flexibility and adjustment strategy precision.

[0184] Example 2:

[0185] Please see Figure 2 As shown, this embodiment provides an adaptive control system for energy-saving parameters in circuit board metal smelting, which also includes:

[0186] The sliding window setting module dynamically sets the sliding window binary set based on the voltage fluctuation level and smelting state characteristic data set.

[0187] Methods for dynamically setting the set of sliding window tuples include:

[0188] Input the voltage fluctuation level and smelting state characteristic data set into the sliding window setting model to obtain a dynamically set sliding window binary set.

[0189] The training method for the sliding window setting model includes:

[0190] A pre-constructed sliding window setting dataset is provided, which includes sliding window setting data of group CK and a set of sliding window pairs corresponding to the sliding window setting data of group CK, where CK is a positive integer greater than 0. The sliding window setting data includes a set of voltage fluctuation level and smelting state characteristic data. The sliding window setting dataset is divided into a sliding window setting data training set and a sliding window setting data validation set. The sliding window setting data training set is used for parameter learning of the sliding window setting model, and the sliding window setting data validation set is used for real-time evaluation of the generalization ability of the sliding window setting model.

[0191] During the training of the sliding window model, a random forest model or a support vector machine model is used to convert the sliding window data into feature vectors as input. The nonlinear features in the data are extracted through the hidden layer, and finally, the probability distribution of the sliding window pair set is generated by the softmax activation function in the output layer. The sliding window pair set corresponding to the highest probability is output as the final prediction result. The training process aims to minimize the cross-entropy loss function. At the same time, an early stopping strategy is introduced to monitor the performance of the sliding window data validation set. When the prediction accuracy on the sliding window data validation set reaches the preset accuracy, the sliding window model is determined to have converged, and the training stops.

[0192] It should be noted that by jointly inputting the voltage fluctuation level and smelting state characteristic data set into the sliding window setting model, and dynamically setting the sliding window binary set—that is, dynamically setting the number of sliding windows and their corresponding lengths for multi-scale analysis—the adaptability and diagnostic accuracy of the timing features in the heating circuit voltage data processing can be effectively improved. Specifically, the voltage fluctuation level reflects the stability of the current power supply. The higher the fluctuation level, the more obvious the high-frequency disturbances and spike interference in the voltage sequence. In this case, configuring more short-period, small-window-length division methods helps to capture short-term disturbance characteristics in a timely manner. On the other hand, the smelting state characteristics reflect the dynamic evolution of the current smelting behavior. When entering the thermal steady state, temperature control plateau period, or material change transition stage, the smelting parameters change slowly or show obvious trends. In this case, configuring a longer window length can extract global trend characteristics under stable changes. By jointly modeling the two types of indicators through the sliding window setting model, a highly adaptable sliding window structure can be set as needed, realizing multi-scale decomposition of the voltage sequence and target-oriented feature extraction.

[0193] Compared with the existing technology that uses a fixed number and length of sliding windows, the dynamic sliding window setting mechanism of this application is more flexible and context-aware, which can improve the response sensitivity and fault tolerance of the subsequent voltage comprehensive diagnostic model to abnormal fluctuations, thereby providing more accurate and granular data support for the control decision of energy-saving strategy.

[0194] Example 3:

[0195] Please see Figure 3 As shown, this embodiment provides an adaptive control method for energy-saving parameters in circuit board metal smelting, including:

[0196] Real-time monitoring and fluctuation analysis of the heating circuit voltage collected per unit time during the circuit board metal smelting process are performed to obtain the voltage fluctuation level;

[0197] The melting state data collected per unit time during the circuit board metal melting process are processed to obtain a set of melting state characteristic data and identify the current melting stage.

[0198] When the voltage fluctuation level is greater than or equal to the preset voltage fluctuation level threshold, an energy-saving strategy control identifier is generated by combining the current smelting stage with the predicted smelting stage for the next unit time.

[0199] Energy-saving control parameters are dynamically adjusted based on voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifiers.

[0200] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0201] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An adaptive control method for energy-saving parameters in circuit board metal smelting, characterized in that, include: Real-time monitoring and fluctuation analysis of the heating circuit voltage collected per unit time during the circuit board metal smelting process are performed to obtain the voltage fluctuation level; The method for obtaining the voltage fluctuation level includes: The unit time is divided into H time points, and the heating circuit voltages at these H time points are used to construct an initial heating circuit voltage set. A set of sliding window pairs is preset; the set of sliding window pairs includes G sliding window pairs, where the first digit of each sliding window pair is a sequence number, and the second digit is the sliding window length. Based on the set of sliding window pairs, the initial heating circuit voltage set is divided to obtain G window sliding voltage sets. Feature extraction is performed on each of the G window sliding voltage sets to obtain G voltage diagnostic data sets. The G voltage diagnostic data sets are used to construct a voltage comprehensive diagnostic dataset. The voltage comprehensive diagnostic dataset is input into the voltage comprehensive diagnostic model to obtain the corresponding voltage comprehensive diagnostic score. The voltage comprehensive diagnostic score is matched with a pre-constructed diagnostic score-fluctuation level mapping table to obtain the corresponding voltage fluctuation level. The melting state data collected per unit time during the circuit board metal melting process are processed to obtain a melting state feature data set, and the current melting stage is identified. The method for identifying the current melting stage includes: inputting the melting state feature data set into a melting stage diagnostic model to obtain the current melting stage; the method for obtaining the melting state feature data set includes: Acquire smelting state data collected per unit time, including smelting temperature, metal conductivity, heating power, metal flow rate in the smelting zone, and metal liquid level. Divide the unit time into H time points, construct a set of melting temperatures for the H time points, construct a set of metal conductivity for the H time points, construct a set of heating power for the H time points, construct a set of metal flow rates for the H melting zones for the H time points, and construct a set of metal liquid level heights for the H time points. The following calculations were performed based on the following sets of melting temperatures: melting temperature change rate, melting temperature change rate, metal conductivity change rate, heating power change rate, metal flow velocity change amplitude, and molten metal level height. The sets of melting temperature change rate, metal conductivity change rate, heating power change rate, metal flow rate change amplitude, and liquid level height change amplitude are used to construct a melting state characteristic data set. The training method for the diagnostic model of the smelting stage includes: A pre-constructed smelting stage diagnostic dataset is constructed, which includes P sets of smelting stage diagnostic data and the smelting stages corresponding to the P sets of smelting stage diagnostic data, where P is a positive integer; the smelting stage diagnostic data includes a set of smelting state feature data; the smelting stage diagnostic dataset is divided into a training set and a validation set, the training set is used for learning the parameters of the smelting stage diagnostic model, and the validation set is used for real-time monitoring of the generalization performance and overfitting degree of the smelting stage diagnostic model. A deep neural network with a multilayer perceptron as its core is used as the diagnostic model for the melting stage. The diagnostic data for the melting stage is standardized and vectorized before being input into the deep neural network, which consists of an input layer, hidden layers, and an output layer. Each hidden layer uses a nonlinear activation function to extract features, and the output layer uses a softmax activation function to obtain the probability distribution corresponding to each melting stage. The melting stage corresponding to the highest probability is taken as the prediction result of the melting stage diagnostic model. During training, the cross-entropy loss function is used as the optimization objective, and a gradient descent-type optimization algorithm is used to update the network weights. An early stopping strategy is set: when the prediction accuracy on the validation set reaches or exceeds a preset threshold, the melting stage diagnostic model is determined to have converged and training is terminated. When the voltage fluctuation level is greater than or equal to a preset voltage fluctuation level threshold, an energy-saving strategy control identifier is generated by combining the current smelting stage with the predicted smelting stage for the next unit time. The method for generating the energy-saving strategy control identifier includes: A set of energy-saving control stages is pre-constructed; the set of smelting state characteristic data and the current smelting stage are input into the smelting stage prediction model to obtain the predicted smelting stage for the next unit of time. If the current smelting stage exists in the energy-saving control stage set, then the current smelting stage is an energy-saving control stage; if the predicted smelting stage for the next unit time exists in the energy-saving control stage set, then the predicted smelting stage for the next unit time is an energy-saving control stage. If both the current smelting stage and the predicted smelting stage for the next unit of time are energy-saving control stages, then the energy-saving strategy control flag will be set to the maintenance adjustment state flag. If the current smelting stage is an energy-saving control stage, and the predicted smelting stage for the next unit of time is a non-energy-saving control stage, then the energy-saving strategy control flag will be set to the flag indicating that the adjustment state is about to end. If the current smelting stage is a non-energy-saving control stage, and the predicted smelting stage for the next unit of time is an energy-saving control stage, then the energy-saving strategy control flag will be set to the advance preparation adjustment status flag. If both the current smelting stage and the predicted smelting stage for the next unit of time are non-energy-saving control stages, then the energy-saving strategy control flag will be set to the no-adjustment status flag. Energy-saving control parameters are dynamically adjusted based on voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifiers.

2. The adaptive control method for energy-saving parameters in circuit board metal smelting according to claim 1, characterized in that, Methods for dynamically adjusting energy-saving control parameters based on voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifiers include: The energy-saving strategy control identifier is matched with the pre-built control identifier numerical conversion table to obtain the digitized energy-saving strategy control identifier, which is recorded as the energy-saving strategy control identifier value. The voltage fluctuation level, smelting state characteristic data set, and energy-saving strategy control identifier value are input into the energy-saving parameter setting model to obtain the energy-saving adjustment parameter set; the energy-saving adjustment parameter set includes the maximum heating power, pulse heating cycle, energizing time duty cycle, temperature response rate upper limit, and heat power adjustment delay time within the stage. A set of corresponding control commands is generated based on the set of energy-saving adjustment parameters; the control commands in the set of control commands are distributed to the corresponding actuators for execution, thereby completing the dynamic adjustment of the energy-saving control parameters.

3. The adaptive control method for energy-saving parameters in circuit board metal smelting according to claim 1, characterized in that, Methods for obtaining the set of G window sliding voltages include: S101: Let the initial value of g be 1, and the range of g is from 1 to G; S102: Obtain the g-th sliding window binary from the set of sliding window binary; obtain the sliding window length corresponding to the g-th sliding window binary; extract the voltages sequentially from the initial heating circuit voltage set in units of sliding window length to form the g-th window sliding voltage set; S103: Let g = g + 1. If g is less than or equal to G, return to S102 to continue execution. If g is greater than G, end the current process.

4. The adaptive control method for energy-saving parameters in circuit board metal smelting according to claim 1, characterized in that, Methods for obtaining G sets of voltage diagnostic data include: S201: Let the initial value of g be 1, and the range of g is from 1 to G; S202: Obtain the set of sliding voltages for the g-th window; denote the number of window sliding voltage subsets in the set of sliding voltages for the g-th window as... The number of heating circuit voltages in the window sliding voltage subset is equal to the corresponding sliding window length. S203: Let the loop index variable... The initial value is 1. The value range is 1 to ; S204: Obtain the first... in the set of sliding voltages of the g-th window. A window-sliding voltage subset; calculate the steady-state voltage reference value, transient disturbance response amplitude, segment fluctuation discrete intensity, and peak disturbance sensitivity of the window-sliding voltage subset; Add the steady-state voltage reference value to the steady-state voltage reference value set; add the transient disturbance response amplitude to the transient disturbance response amplitude set; add the segment fluctuation discrete intensity to the segment fluctuation discrete intensity set; add the peak disturbance sensitivity to the peak disturbance sensitivity set; S205: Order ,like Less than or equal to If so, return to S204 and continue execution; Greater than Then, the voltage diagnostic data set is constructed from the voltage steady-state reference value set, the transient disturbance response amplitude set, the segment fluctuation discrete intensity set, and the peak disturbance sensitivity set into the g-th window sliding voltage set, and g = g + 1 is set. If g is less than or equal to G, the process returns to S202 to continue execution. If g is greater than G, the current process ends.

5. An adaptive control system for energy-saving parameters in circuit board metal smelting, used to implement the adaptive control method for energy-saving parameters in circuit board metal smelting as described in any one of claims 1-4, characterized in that, include: The voltage fluctuation processing module is used to monitor and analyze the voltage fluctuation of the heating circuit collected per unit time during the circuit board metal melting process in real time, and obtain the voltage fluctuation level. The melting status processing module is used to process the melting status data collected per unit time during the circuit board metal melting process, obtain a set of melting status feature data, and identify the current melting stage. The comprehensive diagnostic module is used to generate an energy-saving strategy control identifier when the voltage fluctuation level is greater than or equal to the preset voltage fluctuation level threshold, by combining the current smelting stage with the predicted smelting stage for the next unit time. The energy-saving strategy adjustment module dynamically adjusts energy-saving control parameters based on voltage fluctuation levels, smelting state characteristic data sets, and energy-saving strategy control identifiers.

Images