CN-122021359-A - Group intelligence-based film aluminizing process parameter optimization method and system

CN122021359ACN 122021359 ACN122021359 ACN 122021359ACN-122021359-A

Abstract

The invention relates to the field of process parameter optimization, in particular to a method and a system for optimizing film aluminizing process parameters based on group intelligence, wherein the method comprises the steps of obtaining the process parameters of a film aluminizing process; the method comprises the steps of calculating a health state factor of an evaporation boat, calculating the fluctuation degree of the health state factor, searching an adjusting scheme comprising an input power adjusting amount and a film running adjusting amount based on an intensity pareto evolution algorithm, wherein the objective functions of the adjusting scheme comprise a first objective function and a second objective function, and selecting and executing the adjusting scheme from the obtained pareto optimal front edge to optimize technological parameters. The invention realizes intelligent protection of ageing equipment and optimizing of technological parameters, and improves the automation level and robustness of the production process.

Inventors

YANG FENG
Deng Quandong
HE YINGSHENG

Assignees

中山市金海包装科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (10)

1. The film aluminizing technological parameter optimization method based on group intelligence is characterized by comprising the following steps: Acquiring technological parameters of a film aluminizing process, including input power, film running speed and sheet resistance; calculating the fluctuation degree of the health state factors, wherein the fluctuation degree is a normalized value of standard deviation of all the health state factors in a sliding window set along the history direction; The target function of the adjustment scheme comprises a first target function, a second target function and a process parameter, wherein the first target function is the absolute value of the difference value between the predicted square resistance and the target square resistance corresponding to the adjustment scheme, the second target function is positively correlated with the absolute value of the input power adjustment quantity and negatively correlated with the health state factor, the second target function is positively correlated with the absolute value of the film running adjustment quantity and the fluctuation degree, and the adjustment scheme is selected from the obtained pareto optimal front edge and is executed to optimize the process parameter.
2. The method for optimizing parameters of a thin film aluminizing process based on population intelligence according to claim 1, wherein calculating the health state factor of the evaporation boat comprises calculating the actual evaporation efficiency of the evaporation boat at the current moment, wherein the expression is: In the formula, Indicating time of day The actual evaporation rate of the evaporation boat; Indicating time of day The sheet resistance of the film; Indicating time of day The film is fast; The method comprises the steps of obtaining real-time input power at the current moment, substituting the real-time input power into a preset reference evaporation rate function to obtain a theoretical reference value, and taking the ratio of the actual evaporation efficiency at the current moment to the theoretical reference value as a health state factor.
3. The method for optimizing the thin film aluminizing process parameters based on population intelligence according to claim 1 is characterized by comprising the steps of constructing a sliding window along a history direction, obtaining all health state factors in the window, calculating standard deviations of all health state factors in the sliding window, and carrying out linear normalization processing on the standard deviations to obtain the fluctuation degree.
4. The method for optimizing the film aluminizing process parameters based on population intelligence according to claim 1, wherein the first objective function comprises a prediction sheet resistance obtained based on a prediction model, the prediction model is specifically a multi-layer perceptron model, an input layer of the multi-layer perceptron model receives current input power, current film running speed, input power adjustment quantity and film running speed adjustment quantity, the multi-layer perceptron model comprises at least one hidden layer, the hidden layers all adopt ReLU as an activation function, and an output layer comprises a neuron and outputs the prediction sheet resistance.
5. The method for optimizing parameters of a thin film aluminizing process based on population intelligence according to claim 1, wherein in the process of searching an adjustment scheme comprising an input power adjustment amount and a thin film running speed adjustment amount based on an intensity pareto evolution algorithm, the self-adaptive mutation probability of an individual after crossing is calculated, and the calculation method is as follows: In the formula, Is the current time Adaptive mutation probability of (a); is a preset minimum mutation probability; Is a preset maximum mutation probability; Indicating the degree of fluctuation at time t.
6. The optimization method for thin film aluminizing process parameters based on population intelligence as recited in claim 5, wherein the minimum mutation probability is The value range of (C) is [0.005,0.02], and the maximum mutation probability is shown in the specification The value range of (3) is [0.15,0.3].
7. The method for optimizing parameters of a thin film aluminizing process based on population intelligence as recited in claim 1, wherein the selecting an adjustment scheme from the obtained pareto optimal fronts comprises for each adjustment scheme in the pareto optimal fronts The method comprises the steps of applying disturbance in a set range to obtain a plurality of disturbed first objective function values, calculating variances of the disturbed first objective function values, and selecting an adjustment scheme which is smaller than a set threshold and is located in an inflection point area from the pareto optimal front edge and serves as a final adjustment scheme.
8. The method for optimizing parameters of a thin film aluminizing process based on population intelligence as recited in claim 1, wherein the expression of the second objective function is: In the formula, Representing a second objective function; An adjustment amount indicating the input power of the evaporation boat; Indicating the adjustment amount of the film running speed; Indicating time of day The health state factor of the evaporation boat; Is a preset small constant for preventing denominator from being zero; Indicating the degree of fluctuation at time t.
9. The optimization method of thin film aluminizing process parameters based on population intelligence according to claim 1, wherein the optimization method further comprises linear normalization processing of input power, thin film running speed and sheet resistance.
10. A group intelligence based thin film aluminizing process parameter optimization system, characterized by comprising a processor and a memory, the memory storing computer program instructions which, when executed by the processor, implement the group intelligence based thin film aluminizing process parameter optimization method according to any one of claims 1-9.

Description

Group intelligence-based film aluminizing process parameter optimization method and system Technical Field The invention relates to the field of process parameter optimization, in particular to a method and a system for optimizing thin film aluminizing process parameters based on group intelligence. Background In the film vacuum aluminizing process, such as producing capacitor film or packaging film, maintaining the stability of the sheet resistance of the aluminized layer is a core quality requirement. However, the actual production process is a multi-objective, multi-constraint complex system, such as minimizing deviation of measured value from target value of sheet resistance of aluminized layer, maximizing running speed of thin film, and minimizing input power of key executing components such as evaporation boat. In an ideal steady state, an off-line optimization algorithm, such as the intensity pareto evolution algorithm, can be used to find an optimal set of process parameter combinations, i.e., to achieve an optimal balance between the above three targets, referred to as the pareto optimal front in a multi-target optimization. However, in the actual continuous production, taking the vacuum aluminizing process as an example, the heating efficiency of the core heating element evaporation boat is gradually reduced due to long-time high-temperature evaporation, slag bonding of aluminum liquid and physical loss. This efficiency decay is an unavoidable, slowly occurring progressive failure. When progressive faults occur, the original off-line calibrated optimal balance point (namely the pareto optimal front) is not static any more, but is changed continuously along with the deepening of the faults. The existing strength pareto evolution algorithm is usually used for solving a static optimization problem, is usually triggered passively after the product quality (such as sheet resistance) has deviated seriously, and searches the whole solution space again, so that the fact that the pareto front moves cannot be perceived, the moving direction of the pareto front cannot be predicted, the reconfiguration process is seriously lagged, and a large number of inferior products are generated. Disclosure of Invention The invention provides a method and a system for optimizing parameters of a thin film aluminizing process based on group intelligence, aiming at solving the problem that the prior art cannot effectively process the dynamic movement of the optimal front edge of pareto when dealing with progressive faults. In a first aspect, the invention provides a group-intelligence-based film aluminizing process parameter optimization method, which adopts the following technical scheme: A film aluminizing technological parameter optimization method based on group intelligence comprises the following steps: Acquiring technological parameters of a film aluminizing process, including input power, film running speed and sheet resistance; calculating the fluctuation degree of the health state factors, wherein the fluctuation degree is a normalized value of standard deviation of all the health state factors in a sliding window set along the history direction; The target function of the adjustment scheme comprises a first target function, a second target function and a process parameter, wherein the first target function is the absolute value of the difference value between the predicted square resistance and the target square resistance corresponding to the adjustment scheme, the second target function is positively correlated with the absolute value of the input power adjustment quantity and negatively correlated with the health state factor, the second target function is positively correlated with the absolute value of the film running adjustment quantity and the fluctuation degree, and the adjustment scheme is selected from the obtained pareto optimal front edge and is executed to optimize the process parameter. Compared with the prior art which generally adopts a static optimization model, the device can only passively respond after progressive fault, thereby causing adjustment lag and increasing defective products, according to the invention, the health state factors and the fluctuation degree of the evaporation boat are introduced and calculated, and the two dynamic indexes reflecting the real-time health and stability of the equipment are integrated into the objective function of the intensity pareto evolution algorithm. The optimization process can actively sense the state attenuation of the equipment due to loss, and track the continuously drifting optimal technological parameter combination, namely the pareto optimal front edge in real time, so that the prospective dynamic optimization in the fault occurrence process is realized, the problem of technological adjustment lag caused by equipment aging is reduced, and the product yield and the robustness in the production process are improved. Preferably, calculating the