An energy consumption optimization control method and system for a display panel production line carbon neutralization process

Abstract

The application discloses a kind of energy consumption optimization control method and system of display panel production line carbon neutralization process, comprising: obtaining each process equipment real-time energy consumption, energy structure and process quality data;Based on green energy and fossil energy power distribution ratio calculation dynamic carbon emission factor, establish carbon neutralization target constraint model;Establish the multi-process nonlinear coupling model between clean room air conditioner and etching, lithography, thin film deposition equipment;With carbon neutralization as constraint, with total energy consumption, total carbon emission and comprehensive yield loss as minimization target to build multi-objective optimization function, wherein yield loss is based on model prediction uncertainty and carries out probability calculation;Solve and issue optimization instruction;Through carbon emission deviation correction constraint boundary, yield deviation adjusts optimization weight, forms double closed-loop feedback control.The application realizes the active optimization of energy consumption and carbon emission of display panel production line under the carbon neutralization target, significantly reduces energy consumption and carbon emission while guaranteeing process yield.

Description

Technical Field

[0001] This invention relates to the field of display panel manufacturing technology, and more specifically, to an energy consumption optimization and control method and system for the carbon neutralization process in a display panel production line. Background Technology

[0002] Driven by global carbon neutrality goals, the display panel manufacturing industry faces enormous pressure regarding carbon emissions. Display panel production lines encompass multiple processes, including arraying, cell assembly, and module assembly, involving energy-intensive equipment such as photolithography, etching, thin film deposition, and cleaning, as well as cleanroom environmental control systems. The overall energy density of these production lines is extremely high. Currently, the industry proposes achieving carbon neutrality through green electricity procurement, carbon footprint tracking, and carbon offsetting mechanisms. However, these methods are highly dependent on external resources and lack effective means to proactively optimize and control the energy consumption and carbon emissions of equipment in each process within the production line.

[0003] Existing technologies already include solutions for energy efficiency management in display manufacturing. For example, Chinese invention patent CN119376348A, entitled "Machinery Energy-Saving Management System and Method" and including a manufacturing machine containing this system, discloses an energy-saving management system and method. This system includes an energy consumption measurement module, a product sensing module, a control module, and a display unit. This prior art monitors the energy consumption of the manufacturing machine through the energy consumption measurement module, detects the number of processed products per unit output through the product sensing module, and the control module generates energy consumption data and carbon emission data per unit of product, which are then displayed on the display unit. Essentially, this solution is a tool for monitoring and displaying energy consumption and carbon emissions for a single machine. Its control module only performs data conversion and display logic and lacks the function of optimizing equipment operating parameters in reverse, constrained by carbon neutrality goals. Specifically, it has the following shortcomings:

[0004] 1. The system has not established production line-level carbon emission target constraints, and cannot generate control commands based on carbon neutrality targets to drive the dynamic adjustment of parameters of each equipment;

[0005] 2. The carbon emission data is generated based solely on energy consumption conversion, without taking into account the real-time energy structure, making it difficult to reflect the true carbon emission level of the production line;

[0006] 3. The system is an open-loop monitoring architecture and lacks a closed-loop mechanism to feed back the actual carbon emissions to the control decision-making process, which prevents the production line from actively moving towards the carbon neutrality target.

[0007] Therefore, how to construct a closed-loop optimization control mechanism constrained by the carbon neutrality target, and dynamically adjust the operating parameters of each process equipment according to the real-time carbon emission feedback signal, so as to achieve proactive optimization of energy consumption and carbon emissions within the production line, is a key technical problem that needs to be solved in this field. Summary of the Invention

[0008] To address the shortcomings of existing technologies, the present invention aims to provide an energy consumption optimization and control method and system for the carbon neutrality process in display panel production lines, thereby solving the technical problem of the lack of closed-loop reverse optimization of equipment operating parameters with carbon neutrality as the target constraint.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A method for optimizing and controlling energy consumption in a carbon neutralization process for a display panel production line, comprising the following steps:

[0011] Step S1: Obtain real-time energy consumption data, production capacity cycle data, real-time energy structure data, and process quality correlation data for each process equipment in the display panel production line; wherein, each process equipment includes at least photolithography equipment, etching equipment, thin film deposition equipment, and cleanroom air conditioning system; the process quality correlation data includes at least the exposure energy deviation value of the photolithography process, the radio frequency power fluctuation value of the etching process, the film thickness uniformity of the thin film deposition process, and the supply air temperature difference and particulate matter concentration of the cleanroom air conditioning system;

[0012] In step S1, the real-time energy structure data is used to indicate the power distribution ratio of green energy and fossil energy in the current power consumption of the production line;

[0013] Step S2: Based on the real-time energy structure data, according to the power distribution ratio of green energy and fossil energy in the current period, the carbon emission factors of green energy and fossil energy are weighted and calculated to obtain a dynamic carbon emission factor. The real-time energy consumption data of each process equipment is multiplied by the dynamic carbon emission factor to convert it into real-time carbon emission data, and a production line-level carbon neutrality target constraint model is established. The carbon neutrality target constraint model includes an upper limit value of total carbon emissions in natural days, and the upper limit value of total carbon emissions can be dynamically boundary-corrected under dual closed-loop feedback control.

[0014] In step S2, the real-time energy consumption data of each process equipment is converted into real-time carbon emission data. Specifically, the real-time energy consumption of each process equipment is multiplied by a dynamic carbon emission factor. The dynamic carbon emission factor is obtained by weighting the green energy carbon emission factor and the fossil energy carbon emission factor according to the power allocation ratio.

[0015] Step S3: Establish a multi-process coupling model for the display panel production line. The multi-process coupling model includes: a first nonlinear coupling relationship between the supply air temperature of the cleanroom air conditioning system and the radio frequency power of the etching process equipment on the chip linewidth consistency; a second nonlinear coupling relationship between the fresh air ratio of the cleanroom air conditioning system and the stability of the exposure energy of the photolithography process equipment on the photolithography defect rate; and a third nonlinear coupling relationship between the heating temperature of the thin film deposition process equipment and the supply air temperature of the cleanroom air conditioning system on the thin film deposition rate.

[0016] Step S4: Using the carbon neutrality target constraint model as the constraint condition, the production line capacity target as the boundary condition, and the feasible domain of the operating parameters of each process equipment in the multi-process coupling model as the process constraint condition, a nonlinear multi-objective optimization objective function is constructed. The nonlinear multi-objective optimization objective function minimizes the total energy consumption of the production line, the total carbon emissions of the production line, and the overall yield loss. The overall yield loss is calculated as follows: Based on the first, second, and third nonlinear coupling relationships, the predicted mean and prediction uncertainty measure of each relationship's output value are obtained. Based on each output value and its corresponding preset threshold, the probability of linewidth consistency exceeding the tolerance, the probability of lithography defect rate exceeding the standard, and the probability of deposition rate deviation exceeding the tolerance are calculated based on the predicted mean and prediction uncertainty measure. Then, based on the assumption that the yield stages of each process are independent, the overall yield loss is calculated as: 1 - (1 - probability of linewidth consistency exceeding the tolerance) × (1 - probability of lithography defect rate exceeding the standard) × (1 - probability of deposition rate deviation exceeding the tolerance).

[0017] Step S5: Solve the nonlinear multi-objective optimization objective function to obtain the set of operating parameter optimization instructions for each process equipment; the set of operating parameter optimization instructions includes at least the optimized values ​​of the supply air temperature of the cleanroom air conditioning system, the optimized value of the fresh air ratio, the optimized value of the exposure energy of the photolithography equipment, the optimized value of the radio frequency power of the etching equipment, and the optimized value of the heating temperature of the thin film deposition equipment.

[0018] Step S6: Send the set of operating parameter optimization instructions to the controller of the corresponding process equipment to adjust the operating status of the equipment;

[0019] Step S7: Obtain real-time carbon emission data of each process equipment after adjustment and the process quality correlation data; calculate the carbon emission deviation between the adjusted actual carbon emission and the upper limit of total carbon emission in the carbon neutrality target constraint model; calculate the yield deviation between the adjusted actual yield data and the target yield; use the carbon emission deviation as a feedback signal to perform boundary correction on the upper limit of total carbon emission in the carbon neutrality target constraint model; use the yield deviation as a feedback signal to dynamically adjust the weight of minimizing comprehensive yield loss in the nonlinear multi-objective optimization objective function; the corrected and adjusted model and objective function are used in the next optimization cycle to form a double closed-loop feedback control.

[0020] Further, in step S3, the first nonlinear coupling relationship is characterized by the following mapping function established by historical operating data: CD = f1(Tsup, PRF), where CD is the chip linewidth consistency deviation, Tsup is the supply air temperature of the cleanroom air conditioning system, and PRF is the radio frequency power of the etching process equipment.

[0021] The second nonlinear coupling relationship is characterized by the following mapping function: Defect = f2(Rfresh, Exp), where Defect is the lithography defect rate, Rfresh is the fresh air ratio of the cleanroom air conditioning system, and Exp is the exposure energy of the lithography equipment.

[0022] The third nonlinear coupling relationship is characterized by the following mapping function: DR = f3(Theat, Tsup), where DR is the thin film deposition rate deviation and Theat is the heating temperature of the thin film deposition equipment.

[0023] Furthermore, the mapping functions f1, f2, and f3 are each independently neural network models obtained by system identification based on the historical operating data of the display panel production line.

[0024] Further, in step S5, the nonlinear multi-objective optimization objective function is solved by using a decomposition-based multi-objective evolutionary algorithm or a non-dominated sorting genetic algorithm with an elitist strategy to obtain the Pareto front solution set; from the Pareto front solution set, the final set of operating parameter optimization instructions is determined according to the preset carbon neutrality constraint priority rule.

[0025] This invention also provides an energy consumption optimization and control system for the carbon neutralization process in a display panel production line, comprising:

[0026] The data acquisition module is used to acquire real-time energy consumption data, production capacity cycle data, real-time energy structure data, and process quality correlation data of equipment in each process of the display panel production line; wherein, the equipment in each process includes at least photolithography equipment, etching equipment, thin film deposition equipment, and cleanroom air conditioning system; the process quality correlation data includes at least the exposure energy deviation value of photolithography process, the radio frequency power fluctuation value of etching process, the film thickness uniformity of thin film deposition process, and the supply air temperature difference and particulate matter concentration of cleanroom air conditioning system;

[0027] A real-time carbon emission calculation module, connected to the data acquisition module, is used to calculate a dynamic carbon emission factor by weighting the green energy carbon emission factor and the fossil energy carbon emission factor according to the power distribution ratio of green energy and fossil energy in the current period. It then multiplies the real-time energy consumption data of each process equipment by the dynamic carbon emission factor to convert it into real-time carbon emission data, and establishes a production line-level carbon neutrality target constraint model. The carbon neutrality target constraint model includes an upper limit value for total carbon emissions per natural day, and this upper limit value can be dynamically corrected under dual closed-loop feedback control.

[0028] A multi-process coupling model module, connected to the data acquisition module, is used to store and run the multi-process coupling model of the display panel production line. The multi-process coupling model includes: a first nonlinear coupling relationship between the supply air temperature of the cleanroom air conditioning system and the RF power of the etching process equipment on chip linewidth consistency; a second nonlinear coupling relationship between the fresh air ratio of the cleanroom air conditioning system and the stability of the exposure energy of the photolithography process equipment on the photolithography defect rate; and a third nonlinear coupling relationship between the heating temperature of the thin film deposition process equipment and the supply air temperature of the cleanroom air conditioning system on the thin film deposition rate.

[0029] A nonlinear multi-objective optimization solution module, connected to the real-time carbon emission calculation module and the multi-process coupling model module, is used to construct a nonlinear multi-objective optimization objective function with the carbon neutrality target constraint model as the constraint, production line capacity achievement as the boundary condition, and the feasible region of the operating parameters of each process equipment in the multi-process coupling model as the process constraint. The module then solves the nonlinear multi-objective optimization objective function and outputs an optimization instruction set for the operating parameters of each process equipment. The nonlinear multi-objective optimization objective function aims to minimize total production line energy consumption, total production line carbon emissions, and overall yield loss. The overall yield loss is calculated as follows: Based on the first, second, and third nonlinear coupling relationships, the predicted mean and prediction uncertainty measure of each relationship output value are obtained. Based on each output value and its corresponding preset threshold, the probability of linewidth consistency exceeding the tolerance, the probability of lithography defect rate exceeding the standard, and the probability of deposition rate deviation exceeding the tolerance are calculated according to the predicted mean and prediction uncertainty measure. Then, based on the assumption that the yield links of each process are independent, the overall yield loss is calculated as 1 - (1 - probability of linewidth consistency exceeding the tolerance) × (1 - probability of lithography defect rate exceeding the standard) × (1 - probability of deposition rate deviation exceeding the standard).

[0030] The instruction execution module, connected to the nonlinear multi-objective optimization solution module, is used to send the set of operating parameter optimization instructions to the controller of the corresponding process equipment to adjust the operating status of the equipment.

[0031] Furthermore, a dual-loop feedback module, connected to the instruction execution module, the real-time carbon emission calculation module, and the nonlinear multi-objective optimization solution module, is used to acquire real-time carbon emission data of each process equipment after adjustment and the process quality correlation data; calculate the carbon emission deviation between the adjusted actual carbon emission and the upper limit of total carbon emissions in the carbon neutrality target constraint model; and calculate the yield deviation between the adjusted actual yield data and the target yield. The carbon emission deviation is used as a feedback signal to perform boundary correction on the upper limit of total carbon emissions in the carbon neutrality target constraint model. The yield deviation is also used as a feedback signal to dynamically adjust the weight for minimizing the comprehensive yield loss in the nonlinear multi-objective optimization objective function.

[0032] Furthermore, the nonlinear multi-objective optimization solution module uses a decomposition-based multi-objective evolutionary algorithm or a non-dominated sorting genetic algorithm with an elitist strategy to obtain a Pareto front solution set; and determines the final set of operating parameter optimization instructions from the Pareto front solution set according to a preset carbon neutrality constraint priority rule.

[0033] The present invention also provides a computer-readable storage medium storing computer program instructions thereon, characterized in that, when the computer program instructions are executed by a processor, the energy consumption optimization control method for the carbon neutralization process of the above-mentioned display panel production line is implemented.

[0034] By adopting the above technical solution, the beneficial effects of the present invention are as follows:

[0035] 1. Based on the power allocation ratio of real-time green energy and fossil energy, the dynamic carbon emission factor is calculated, and the energy consumption data is converted into real-time carbon emission data to realize dynamic carbon emission accounting. This overcomes the problem of static and lagging carbon emission factors in traditional methods, making the carbon emission assessment of the production line more accurate and real-time.

[0036] 2. This invention establishes for the first time a nonlinear coupling relationship model between the cleanroom air conditioning system and core process equipment such as etching, photolithography, and thin film deposition, revealing the coupling influence mechanism of environmental parameters such as supply air temperature and fresh air ratio on process quality (linewidth consistency, photolithography defect rate, deposition rate), and providing a model basis for multi-process collaborative optimization.

[0037] 3. In calculating overall yield loss, a probability-based method for estimating deviation is constructed using the predicted mean and variance output by models such as Gaussian process regression. This method not only focuses on the model's predicted values ​​but also quantifies the reliability of the model's predictions, guiding the optimization algorithm to tend to select process parameter regions where the model is more confident and the quality is more stable, thus significantly improving the robustness of optimization decisions.

[0038] 4. The carbon emission closed loop dynamically adjusts the constraint boundary by monitoring the deviation between actual carbon emissions and the target upper limit in real time; the yield closed loop dynamically adjusts the weight of yield loss in the optimization objective function by monitoring the deviation between actual yield and target yield; a dual-loop feedback control mechanism is constructed. The two closed loops act on different levels of the optimization model (constraint boundary and target weight), and the differentiated design enables the system to achieve a dynamic balance between carbon neutrality constraints and yield assurance.

[0039] 5. Through the comprehensive application of the above methods, the total energy consumption and total carbon emissions were reduced by approximately 10% in actual production line testing, while the overall yield loss increased only slightly. This demonstrates that the method significantly reduces energy consumption and carbon emissions while ensuring a basically stable yield. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the module structure of the energy consumption optimization control system in an embodiment of the present invention.

[0041] Figure 2 This is a schematic diagram of the data flow for dual closed-loop feedback control in an embodiment of the present invention.

[0042] Figure 3 This is a schematic diagram of the input-output relationship of a multi-process coupling model. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of the present invention clearer, the following description will be provided in conjunction with the accompanying drawings. Figures 1 to 3 The present invention will be further described in detail below with reference to specific embodiments. It should be noted that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0044] Example 1

[0045] This embodiment provides an energy consumption optimization control method for a carbon neutrality process in a display panel production line. The method is applied to a TFT-LCD (Thin Film Transistor Liquid Crystal Display) manufacturing line, which includes array, cell assembly, and module assembly processes. The main process equipment includes photolithography equipment, etching equipment, thin film deposition equipment, and a cleanroom air conditioning system. The control method in this embodiment is executed through an edge computing server on the production line, which communicates with the production line's SCADA system, MES system, energy management system, and equipment controllers.

[0046] Step S1, Data Acquisition

[0047] The edge computing server obtains real-time energy consumption data of each process equipment from the production line SCADA system through the OPCUA protocol. Specifically, this includes the real-time electrical power Pl(t) of the lithography equipment, the real-time electrical power Pe(t) of the etching equipment, the real-time electrical power Pd(t) of the thin film deposition equipment, and the real-time electrical power Pa(t) of the cleanroom air conditioning system, in kW.

[0048] The production capacity cycle data is obtained through the MES system, including the current production cycle Tc (unit: seconds / piece) and the cumulative output N for the day.

[0049] The energy management system acquires real-time energy structure data, which indicates the power distribution ratio of green energy to fossil fuels in the current electricity consumption of the production line. In this embodiment, the energy management system is interfaced with the power grid dispatching system to acquire the green energy power Pg(t) and fossil fuel power Pf(t) for the current time period in real time. The sum of the two equals the total power consumption of the production line.

[0050] Process quality-related data is obtained through SCADA and process quality inspection systems, specifically including: the exposure energy deviation value ΔE of the photolithography process (the difference between the actual exposure energy and the set value, in mJ / cm). 2 The following parameters are considered: RF power fluctuation ΔPRF (the percentage deviation between real-time RF power and set power) for etching processes; film thickness uniformity δ (the ratio of standard deviation of film thickness within the wafer to average film thickness, expressed as a percentage) for thin film deposition processes; and supply air temperature difference ΔTsup (the difference between actual supply air temperature and target supply air temperature, in °C) and particulate matter concentration Cp (particles / m³) for cleanroom air conditioning systems. 3 (For particles larger than 0.3μm).

[0051] The above data is collected once per minute, which meets the needs of real-time control of the production line.

[0052] Step S2: Establishing dynamic carbon emission accounting and carbon neutrality target constraints

[0053] This step first converts the real-time energy consumption data of each process equipment into real-time carbon emission data based on real-time energy structure data.

[0054] Specifically, a green energy carbon emission factor EFg and a fossil energy carbon emission factor EFf are defined. In this embodiment, EFg is set to 0.02 kg CO2 / kWh (considering the carbon emissions allocated throughout the entire life cycle of green energy production equipment), and EFf is set to 0.85 kg CO2 / kWh (the baseline emission factor for the Chinese regional power grid). Based on the power allocation ratio obtained in step S1, the dynamic carbon emission factor EF(t) is calculated:

[0055] EF(t)=[Pg(t)·EFg+Pf(t)·EFf] / [Pg(t)+Pf(t)]

[0056] The real-time carbon emission rate Ci(t) of each process equipment (unit: kgCO) 2 / h) is:

[0057] Ci(t) = Pi(t) × EF(t), where i represents the equipment for photolithography, etching, thin film deposition, and cleanroom air conditioning.

[0058] The total real-time carbon emission rate of the production line is C(t) = ΣCi(t).

[0059] Based on this, a production line-level carbon neutrality target constraint model is established. In this embodiment, the carbon neutrality target constraint model sets an upper limit value of total carbon emissions, Cmax_daily, on a daily basis. This upper limit value is decomposed and set according to the annual carbon neutrality plan of the production line. For example, if the annual carbon emission quota is 100,000 tons, it is allocated to approximately 273.97 tons per day. Considering production fluctuations, a certain margin is allowed, and the final value is set at 250 tons of CO2 per day.

[0060] This constraint is expressed as: , where T is the prediction period of the day's running time plus the remaining time.

[0061] Step S3: Establish a multi-process coupling model

[0062] This step establishes a multi-process coupling model for the display panel production line. This model is stored and run by the multi-process coupling model module in the edge computing server. The model contains the following three non-linear coupling relationships:

[0063] The first nonlinear coupling relationship: the influence of the supply air temperature Tsup of the cleanroom air conditioning system and the radio frequency power PRF of the etching process equipment on the chip linewidth consistency CD.

[0064] In this embodiment, the relationship is obtained by systematically identifying the historical operating data of the production line over the past 12 months. The historical data collection conditions are as follows: under normal production conditions, the cleanroom supply air temperature Tsup (range 22.0±2.0℃), etching equipment RF power PRF (range 1500W-2500W), and the corresponding chip linewidth consistency CD (expressed as linewidth standard deviation σ, unit nm) for each batch are recorded.

[0065] The system identification was performed using a Gaussian process regression (GPR) model, and the mapping function was obtained:

[0066] CD=f1(Tsup,PRF)=k1·(Tsup-Tsup_opt) 2 +k2·(PRF-PRF_opt) 2 +k3·(Tsup-Tsup_opt)·(PRF-PRF_opt)+CD0

[0067] Where Tsup_opt is the optimal supply air temperature (22.5℃ in this embodiment), PRF_opt is the optimal RF power (2000W in this embodiment), CD0 is the optimal linewidth consistency (0.85nm in this embodiment), and k1, k2, and k3 are the identified coupling coefficients. In this embodiment, training is performed using a training set (approximately 5000 sets of data) and validation is performed using a test set (approximately 1000 sets of data), with k1 set to 0.12nm / ℃. 2 k2 is taken as 3.5 × 10 -6 nm / W 2 With k3 set to -0.008 nm / (℃·W), the model fit R² reached 0.93.

[0068] This mapping function reveals the coupling effect between supply air temperature and RF power: when the supply air temperature deviates from the optimal value, even if the RF power remains optimal, the linewidth consistency will deteriorate; and when both deviate simultaneously, there is a cross-coupling term that exacerbates the effect.

[0069] It should be noted that the Gaussian process regression (GPR) model, while outputting the predicted CD value μ_CD, can also output the variance σ_CD of that predicted value. 2This variance quantifies the model's prediction uncertainty under the current input (Tsup, PRF). Prediction uncertainty stems from factors such as the sparsity of the model's training data and measurement noise. For the f3 model built using a neural network, its prediction variance can also be estimated using techniques such as Monte Carlo Dropout or Deep Ensemble. This prediction uncertainty will be used to subsequently calculate the linewidth consistency out-of-error probability, making the estimation of Y_loss more scientific and robust.

[0070] The second nonlinear coupling relationship: the influence of the fresh air ratio Rfresh of the cleanroom air conditioning system and the exposure energy Exp of the lithography equipment on the lithography defect rate.

[0071] Using a similar historical data system identification method, lithography defect rate data (expressed as defects per 10,000 pieces) were collected under different fresh air ratio settings (range 15%-35%) and exposure energy settings (range ±5%). A mapping function was established using a support vector regression (SVR) model.

[0072] Defect=f2(Rfresh,Exp)=α·exp(β·|Rfresh-Rfresh_opt|)+γ·(Exp-Exp_opt) 2 +δ·|Rfresh-Rfresh_opt|·|Exp-Exp_opt|+Defect0

[0073] In this embodiment, Rfresh_opt is the optimal fresh air ratio of 25%, and Exp_opt is the optimal exposure energy of 28.5 mJ / cm². 2 With Defect0 as the base defect rate of 3.2 per 10,000 pieces, and α, β, γ, and δ as identification coefficients, the model R² reaches 0.89.

[0074] The mapping function shows that while a lower fresh air ratio is beneficial for energy saving, it can lead to changes in the positive pressure and airflow organization of the cleanroom, affecting the microenvironment of the lithography area and increasing the lithography defect rate. When the exposure energy also deviates from the optimal value, the two produce a synergistic deterioration effect.

[0075] The third nonlinear coupling relationship: the influence of the heating temperature (Theat) of the thin film deposition equipment and the supply air temperature (Tsup) of the cleanroom air conditioning system on the thin film deposition rate (DR).

[0076] A neural network model was used to establish the mapping function: DR = f3(Theat, Tsup). This neural network is a three-layer fully connected feedforward network with two neurons in the input layer (Theat, Tsup), eight neurons in the hidden layer (using the ReLU activation function), and one neuron in the output layer (DR). Approximately 8000 training datasets were used, and the Adam optimizer was employed, achieving a mean squared error (MSE) of 1.2 × 10⁻⁻⁶. 4 .

[0077] The mapping function indicates that the thin film deposition process is affected not only by its own heating temperature but also indirectly by the cleanroom supply air temperature's influence on the deposition chamber's ambient temperature field; these two factors are nonlinearly coupled. When the cleanroom supply air temperature increases due to energy-saving requirements, the heating temperature of the thin film deposition equipment needs to be adjusted accordingly to maintain the stability of the deposition rate.

[0078] It should be noted that the specific forms of the above mapping functions f1, f2, and f3 are the specific implementation methods adopted in this embodiment. In practical applications, other neural network models can be used, as long as the nonlinear coupling relationship can be established based on the historical operation data of the production line.

[0079] Step S4: Construct a nonlinear multi-objective optimization objective function.

[0080] This step uses the carbon neutrality target constraint model as the constraint condition, the production line capacity target as the boundary condition, and the feasible domain of the equipment operating parameters of each process in the multi-process coupling model in step S3 as the process constraint condition to construct a nonlinear multi-objective optimization objective function.

[0081] The decision variables are as follows:

[0082] The supply air temperature Tsup of the cleanroom air conditioning system has a feasible range of [21.0, 24.0] (unit: °C).

[0083] The feasible range for the fresh air ratio (Rfresh) of a cleanroom air conditioning system is [15%, 35%].

[0084] Exponent energy Exp for photolithography equipment, feasible region [27.0, 30.0] (unit: mJ / cm²) 2 );

[0085] Radio frequency power (PRF) of etching process equipment, feasible range [1500, 2500] (unit: W);

[0086] The heating temperature of the thin film deposition equipment, feasible range [350, 420] (unit: °C).

[0087] Production capacity target boundary conditions: Within the current optimization cycle, the optimized combination of equipment operating parameters must ensure that the production line capacity is not less than 95% of the target capacity Qtarget, i.e., Q≥0.95·Qtarget. The production capacity Q is calculated using the production cycle time data obtained in step S1: Q=3600 / Tc (unit: pieces / hour).

[0088] Three minimization objectives:

[0089] Objective 1: Total energy consumption of the production line E_total = (P_l + P_e + P_d + P_a)·Δt, where P_l, P_e, P_d, and P_a are the estimated power of the lithography, etching, deposition, and air conditioning equipment under optimized parameters, respectively, and Δt is the optimization cycle length.

[0090] The relationship between the power of each device and the decision variables is established through the device power model: For example, the functional relationship between the power Pa of the cleanroom air conditioning system and the supply air temperature Tsup and the fresh air ratio Rfresh is Pa=g(Tsup,Rfresh), which is obtained through regression analysis of the historical operating data of the air conditioning system; the power Pe of the etching device is approximately linearly related to the radio frequency power PRF Pe=η·PRF+Pe_base, where η is the power transfer efficiency and Pe_base is the base standby power.

[0091] As an example, the functional relationship between the power P_a of a cleanroom air conditioning system and the supply air temperature Tsup and the fresh air ratio Rfresh, P_a=g(Tsup,Rfresh), can be approximated by the following multiple linear regression model:

[0092] P_a=a0+a1·(T_target-Tsup)+a2·Rfresh+a3·(T_target-Tsup)·Rfresh

[0093] Where T_target is the rated supply air temperature setpoint for the cleanroom (e.g., 22℃); a0 is the base power (e.g., 85kW); a1 is the cooling / heating power coefficient (e.g., 4.2kW / ℃); a2 is the fresh air handling power coefficient (e.g., 0.6kW / %); and a3 is the coupling term coefficient (e.g., 0.08kW / (℃·%)). These coefficients were obtained by fitting historical operating data of the air conditioning system.

[0094] An example of a linear relationship model between etching equipment power P_e and radio frequency power PRF is: P_e = η·PRF + Pe_base, where η is the power transfer efficiency (0.75 in this embodiment), and Pe_base is the base standby power (3.5kW in this embodiment). Power models for lithography equipment and thin film deposition equipment can be established in a similar manner.

[0095] Objective 2: Total carbon emissions from the production line C_total = E_total·EF(t), where EF(t) is the dynamic carbon emission factor calculated in step S2.

[0096] Objective 3: The overall yield loss Y_loss is calculated from the joint probability density function of the interaction between various parameters in the process quality correlation data and the nonlinear coupling relationship in the multi-process coupling model, which affects the production line yield. In this embodiment, the overall yield loss Y_loss is specifically defined as:

[0097] Y_loss=1-[1-P(CD>CD_threshold)]·[1-P(Defect>Defect_threshold)]·[1-P(|DR|>DR_threshold)]

[0098] CD_threshold is the linewidth consistency threshold (1.5nm in this embodiment), Defect_threshold is the lithography defect rate threshold (10 per 10,000 wafers in this embodiment), and DR_threshold is the deposition rate deviation threshold (5% in this embodiment).

[0099] The above-mentioned out-of-range probabilities are not simply compared with the model predictions, but are calculated based on probability propagation according to the uncertainty of the model predictions, as follows:

[0100] (1) Calculate the probability of linewidth consistency deviation P(CD>CD_threshold):

[0101] First, using the mapping function f1 (e.g., Gaussian process regression model) established in step S3, the predicted distribution of chip linewidth consistency deviation CD is obtained based on the current decision variable combination (Tsup, PRF): CD ~ N(μ_CD, σ_CD) 2 Where μ_CD is the predicted mean of the f1 output, and σ_CD is the predicted standard deviation of the f1 output (representing the combination of model uncertainty and inherent noise).

[0102] The probability of linewidth consistency exceeding tolerance is:

[0103] P(CD>CD_threshold)=1-Φ((CD_threshold-μ_CD) / σ_CD)

[0104] Where Φ is the cumulative distribution function of the standard normal distribution.

[0105] (2) Calculate the probability of exceeding the lithography defect rate limit P(Defect>Defect_threshold) and the probability of exceeding the deposition rate deviation P(|DR|>DR_threshold):

[0106] Using the exact same method as above, the predicted distributions Defect~N(μ_Defect, σ_Defect) are based on the outputs of mapping functions f2 and f3, respectively. 2 ) and DR~N(μ_DR,σ_DR 2 ), calculate the corresponding out-of-tolerance probability:

[0107] P(Defect>Defect_threshold)=1-Φ((Defect_threshold-μ_Defect) / σ_Defect)

[0108] P(|DR|>DR_threshold)=1-[Φ((DR_threshold-μ_DR) / σ_DR)-Φ((-DR_threshold-μ_DR) / σ_DR)]

[0109] The rationale behind this method lies in its acknowledgment of the inherent prediction errors in the process model and its quantification of this uncertainty as a probability distribution. When the model lacks confidence in predicting a certain region (i.e., σ is large), the calculated out-of-range probability also increases accordingly, guiding the optimization algorithm to favor process parameter regions where the model is more confident and of higher quality. Compared to existing methods that treat model predictions as deterministic values, this method aligns better with the principles of statistical process control in industrial production, significantly improving the robustness and reliability of optimization decisions. For black-box models that cannot directly output prediction variance (such as ordinary neural networks), mature techniques such as Monte Carlo dropout or guided aggregation (Bagging) can be used to approximate and estimate their prediction uncertainty.

[0110] The nonlinear multi-objective optimization objective function is constructed as follows:

[0111] minF(Tsup,Rfresh,Exp,PRF,Theat)={E_total,C_total,Y_loss}

[0112] Constraints:

[0113] (1) Carbon neutrality target constraint: C_total≤Cmax_daily;

[0114] (2) Capacity target boundary: Q≥0.95·Qtarget;

[0115] (3) Process constraints: Each decision variable is within its feasible region.

[0116] Step S5, Multi-objective optimization solution

[0117] The nonlinear multi-objective optimization objective function constructed in step S4 is solved by using a decomposition-based multi-objective evolutionary algorithm (MOEA / D). This algorithm decomposes the multi-objective optimization problem into multiple single-objective sub-problems and searches for the Pareto front simultaneously through population evolution.

[0118] In this embodiment, the parameters of MOEA / D are set as follows: population size is 100, neighborhood size is 20, crossover probability is 0.9, mutation probability is 0.1, and maximum number of iterations is 500 generations.

[0119] During the solution process, the fitness of each individual is evaluated using the values ​​of the three objective functions defined in step S4. After iterative convergence, the algorithm obtains a Pareto front solution set, where each solution represents a combination of equipment operating parameters that achieves a non-dominated equilibrium among the three objectives.

[0120] After obtaining the Pareto front solution set, the final set of operating parameter optimization instructions is determined according to a pre-defined priority rule for carbon neutrality constraints. This priority rule is defined as follows:

[0121] Rule 1: Prioritize ensuring the carbon neutrality objective constraint is met, and exclude solutions where C_total exceeds Cmax_daily;

[0122] Rule 2: Among the solutions that satisfy Rule 1, select the solution with the lowest E_total as the optimal solution for the current optimization cycle. If there are multiple solutions with similar energy efficiency, further select the solution with the smallest Y_loss.

[0123] This yields the set of operating parameter optimization instructions for the current optimization cycle, specifically including: the supply air temperature optimization value Tsup of the cleanroom air conditioning system, the fresh air ratio optimization value Rfresh, the exposure energy optimization value Exp of the lithography equipment, the radio frequency power optimization value PRF of the etching equipment, and the heating temperature optimization value Theat of the thin film deposition equipment.

[0124] In a specific optimization example of this embodiment, the production line operating parameters before optimization were: Tsup=22.0℃, Rfresh=28%, Exp=28.5mJ / cm². 2 With PRF = 2000W and Theatre = 380℃, E_total = 1250kWh / h, C_total = 956kgCO2 / h (EF(t) = 0.764kgCO2 / kWh), and Y_loss = 0.035, the optimized operating parameters are: Tsup = 23.2℃, Rfresh = 22%, and Exp = 28.3mJ / cm². 2The parameters were: PRF = 1950 W, Theatre = 386 °C, E_total = 1020 kWh / h, C_total = 779 kg CO2 / h, and Y_loss = 0.038. After optimization, total energy consumption and total carbon emissions decreased, while the overall yield loss increased only slightly, indicating that this method can significantly reduce energy consumption and carbon emissions while ensuring a relatively stable yield.

[0125] Step S6, Instruction Issuance and Execution

[0126] The operating parameter optimization instruction set obtained in step S5 is sent to the controller of the corresponding process equipment via the OPCUA protocol to adjust the equipment operating status. Specifically, this includes:

[0127] Send Tsup and Rfresh to the PLC controller of the cleanroom air conditioning system to adjust the supply air temperature setpoint and the air valve opening;

[0128] The Exp value is sent to the exposure control system of the lithography equipment to adjust the exposure energy setting.

[0129] The PRF is sent to the RF power controller of the etching equipment to adjust the RF output power;

[0130] The Heat is sent to the heating temperature control system of the thin film deposition equipment to adjust the heater power.

[0131] After receiving the instruction, each equipment controller completes the parameter adjustment within the preset gradual transition time (usually 30 seconds to 2 minutes, depending on the equipment response characteristics) to avoid sudden changes affecting process stability.

[0132] Step S7, Dual Closed-Loop Feedback Control

[0133] After the operating parameters are issued and executed, real-time carbon emission data and process quality-related data of each process equipment are continuously acquired after adjustment.

[0134] Carbon emission closed-loop control loop: Calculate the carbon emission deviation ΔC(t) between the adjusted actual carbon emissions and the upper limit of total carbon emissions in the carbon neutrality target constraint model.

[0135] ΔC(t)=C_actual(t)-Cmax_daily_remaining(t)

[0136] Where C_actual(t) is the cumulative actual carbon emissions monitoring value at the current moment, and Cmax_daily_remaining(t) is the upper limit of carbon emissions allowed for the remaining period of the day.

[0137] The carbon emission deviation value ΔC(t) is used as a feedback signal to adjust the boundary of the carbon neutrality target constraint model in step S4. Specifically, when ΔC(t) > 0 (i.e., the actual carbon emissions have exceeded the expected proportion), the carbon emission constraint boundary for the next optimization cycle is tightened, reducing Cmax_daily by 5%; when ΔC(t) < 0 and the deviation exceeds a preset threshold (e.g., -10%), the constraint boundary is appropriately relaxed, increasing Cmax_daily by 3%, so as to create more optimization space for improving yield when there is a surplus in carbon emissions.

[0138] Yield closed-loop control loop: Calculate the yield deviation ΔY(t) between the adjusted actual yield data and the target yield.

[0139] ΔY(t) = Y_target - Y_actual(t)

[0140] Where Y_target is the target yield (95% in this embodiment), and Y_actual(t) is the measured overall yield of the current optimization cycle.

[0141] The yield deviation value ΔY(t) is used as a feedback signal to dynamically adjust the weight of minimizing the overall yield loss in step S4. Specifically, when ΔY(t) > 0 and exceeds the preset tolerance range (i.e., the actual yield is lower than the target yield and the deviation exceeds 0.5%), the weight coefficient of the overall yield loss Y_loss in the objective function is increased from the default 1.0 to 1.5, so that the next optimization cycle is more biased towards ensuring yield; when ΔY(t) stabilizes within the tolerance range, the weight coefficient is restored to the default value.

[0142] The aforementioned carbon emission deviation and yield deviation values ​​are simultaneously input into the nonlinear multi-objective optimization objective function in step S4, forming a dual-loop feedback control mechanism. The two loops act on different parts of the optimization model: the carbon emission loop influences the feasible region of the solution by correcting the constraint boundaries, while the yield loop influences the search direction of the solution by adjusting the objective weights. This differentiated design enables the system to achieve a dynamic balance between carbon neutrality constraints and yield assurance, reaching an overall optimal operating state.

[0143] In this embodiment, the dual closed-loop feedback control runs continuously in the form of rolling optimization. Each optimization cycle is 15 minutes, that is, a complete set of steps S1 to S7 is executed every 15 minutes, so as to achieve continuous closed-loop optimization of the carbon neutralization process of the display panel production line.

[0144] Example 2

[0145] This embodiment provides an energy consumption optimization control system for the carbon neutralization process of a display panel production line, which corresponds completely to the control method described in Embodiment 1. The correspondence between the modules in this system and the steps in Embodiment 1 is as follows:

[0146] The data acquisition module is used to perform the data acquisition function in step S1 of Embodiment 1. It interfaces with the SCADA system, MES system and energy management system through the OPCUA protocol to collect real-time energy consumption data, production capacity cycle data, real-time energy structure data and process quality related data.

[0147] The real-time carbon emission accounting module is used to perform the dynamic carbon emission accounting and carbon neutrality target constraint establishment function in step S2 of Embodiment 1. It includes a dynamic carbon emission factor calculation unit and a carbon neutrality target constraint management unit.

[0148] The multi-process coupling model module is used to perform the multi-process coupling model storage and operation function in step S3 of Embodiment 1. It internally stores the specific model parameters of three nonlinear coupling relationship mapping functions f1, f2, and f3 identified through the historical operation data system.

[0149] The nonlinear multi-objective optimization solution module is used to perform the optimization objective function construction and solution functions in steps S4 to S5 of Embodiment 1. In this embodiment, the module adopts a decomposition-based multi-objective evolutionary algorithm (MOEA / D), but a non-dominated sorting genetic algorithm with an elitist strategy (NSGA-II) can also be used. After the algorithm obtains the Pareto front solution set, the final set of running parameter optimization instructions is determined according to the preset carbon neutrality constraint priority rules (Rule 1: exclude solutions that exceed Cmax_daily; Rule 2: select the solution with the lowest E_total among the remaining solutions).

[0150] The instruction execution module is used to execute the instruction issuance function in step S6 of Embodiment 1, and to issue the optimized instruction set to each device controller through the OPCUA protocol.

[0151] The dual closed-loop feedback module is used to execute the dual closed-loop feedback control function in step S7 of Embodiment 1. It includes a carbon emission deviation value calculation unit, a yield deviation value calculation unit, a constraint boundary correction unit, and a weight dynamic adjustment unit.

[0152] Example 3

[0153] This embodiment provides a computer-readable storage medium storing computer program instructions. When executed by a processor, these computer program instructions implement all the steps of the energy consumption optimization control method for the carbon neutralization process of the display panel production line as described in Embodiment 1.

[0154] In this embodiment, the computer-readable storage medium is a solid-state drive (SSD) deployed in the production line edge computing server. The processor is an Intel Xeon series server CPU, and the memory is 64GB DDR4. Of course, the computer-readable storage medium can also be a USB flash drive, external hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, or any other medium capable of storing program code.

[0155] The computer program stored in this storage medium is written in Python, wherein:

[0156] The data acquisition module implements OPCUA client communication through the opcua-asyncio library;

[0157] The mapping function f1 is implemented using GaussianProcessRegressor from the scikit-learn library, f2 is implemented using SVR, and f3 is implemented using PyTorch to build a neural network model.

[0158] The multi-objective optimization solution is implemented using the MOEA / D algorithm from the pymoo library;

[0159] The gradient information of the optimization objective is estimated using the forward difference method.

[0160] When executing this computer program, those skilled in the art only need to configure the interface addresses of the program with the production line SCADA system, MES system, and energy management system, and import historical operating data of the production line to train the mapping function of the multi-process coupling model, so that it can be deployed and run in the display panel production line to realize the energy consumption optimization closed-loop control of the production line-level carbon neutrality process.

[0161] The control method of this invention is automatic control through a controller. The control circuit of the controller can be implemented by simple programming by those skilled in the art. The power supply is also common knowledge in the art. Furthermore, since this invention is mainly used to protect mechanical devices, the control method and circuit connection will not be explained in detail here.

[0162] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any ordinary changes and substitutions made by those skilled in the art within the scope of the technical solution of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing and controlling energy consumption in a carbon neutralization process for a display panel production line, characterized in that, Includes the following steps: Step S1: Obtain real-time energy consumption data, production capacity cycle data, real-time energy structure data, and process quality correlation data for each process equipment in the display panel production line; wherein, each process equipment includes at least photolithography equipment, etching equipment, thin film deposition equipment, and cleanroom air conditioning system; the process quality correlation data includes at least the exposure energy deviation value of the photolithography process, the radio frequency power fluctuation value of the etching process, the film thickness uniformity of the thin film deposition process, and the supply air temperature difference and particulate matter concentration of the cleanroom air conditioning system; Step S2: Based on the real-time energy structure data, according to the power distribution ratio of green energy and fossil energy in the current period, the carbon emission factors of green energy and fossil energy are weighted and calculated to obtain a dynamic carbon emission factor. The real-time energy consumption data of each process equipment is multiplied by the dynamic carbon emission factor to convert it into real-time carbon emission data, and a production line-level carbon neutrality target constraint model is established. The carbon neutrality target constraint model includes an upper limit value of total carbon emissions in natural days, and the upper limit value of total carbon emissions can be dynamically boundary-corrected under dual closed-loop feedback control. Step S3: Establish a multi-process coupling model for the display panel production line. The multi-process coupling model includes: a first nonlinear coupling relationship between the supply air temperature of the cleanroom air conditioning system and the radio frequency power of the etching process equipment on the chip linewidth consistency; a second nonlinear coupling relationship between the fresh air ratio of the cleanroom air conditioning system and the stability of the exposure energy of the photolithography process equipment on the photolithography defect rate; and a third nonlinear coupling relationship between the heating temperature of the thin film deposition process equipment and the supply air temperature of the cleanroom air conditioning system on the thin film deposition rate. Step S4: Using the carbon neutrality target constraint model as the constraint condition, the production line capacity target as the boundary condition, and the feasible domain of the operating parameters of each process equipment in the multi-process coupling model as the process constraint condition, a nonlinear multi-objective optimization objective function is constructed. The nonlinear multi-objective optimization objective function minimizes the total energy consumption of the production line, the total carbon emissions of the production line, and the overall yield loss. The overall yield loss is calculated as follows: Based on the first, second, and third nonlinear coupling relationships, the predicted mean and prediction uncertainty measure of each relationship's output value are obtained. Based on each output value and its corresponding preset threshold, the probability of linewidth consistency exceeding the tolerance, the probability of lithography defect rate exceeding the standard, and the probability of deposition rate deviation exceeding the tolerance are calculated based on the predicted mean and prediction uncertainty measure. Then, based on the assumption that the yield stages of each process are independent, the overall yield loss is calculated as: 1 - (1 - probability of linewidth consistency exceeding the tolerance) × (1 - probability of lithography defect rate exceeding the standard) × (1 - probability of deposition rate deviation exceeding the tolerance). Step S5: Solve the nonlinear multi-objective optimization objective function to obtain the set of operating parameter optimization instructions for each process equipment; the set of operating parameter optimization instructions includes at least the optimized values ​​of the supply air temperature of the cleanroom air conditioning system, the optimized value of the fresh air ratio, the optimized value of the exposure energy of the photolithography equipment, the optimized value of the radio frequency power of the etching equipment, and the optimized value of the heating temperature of the thin film deposition equipment. Step S6: Send the set of operating parameter optimization instructions to the controller of the corresponding process equipment to adjust the operating status of the equipment; Step S7: Obtain real-time carbon emission data of each process equipment after adjustment and the process quality correlation data; calculate the carbon emission deviation between the adjusted actual carbon emission and the upper limit of total carbon emission in the carbon neutrality target constraint model; calculate the yield deviation between the adjusted actual yield data and the target yield; use the carbon emission deviation as a feedback signal to perform boundary correction on the upper limit of total carbon emission in the carbon neutrality target constraint model; use the yield deviation as a feedback signal to dynamically adjust the weight of minimizing comprehensive yield loss in the nonlinear multi-objective optimization objective function; the corrected and adjusted model and objective function are used in the next optimization cycle to form a double closed-loop feedback control.

2. The energy consumption optimization and control method for the carbon neutralization process in a display panel production line according to claim 1, characterized in that, In step S1, the real-time energy structure data is used to indicate the power distribution ratio of green energy and fossil energy in the current power consumption of the production line; in step S2, the real-time energy consumption data of each process equipment is converted into real-time carbon emission data, specifically by multiplying the real-time energy consumption of each process equipment by a dynamic carbon emission factor, which is obtained by weighting the green energy carbon emission factor and the fossil energy carbon emission factor according to the power distribution ratio.

3. The energy consumption optimization and control method for a carbon neutralization process in a display panel production line according to claim 1, characterized in that, In step S3, the first nonlinear coupling relationship is characterized by the following mapping function established by historical operating data: CD = f1(Tsup,PRF), where CD is the chip linewidth consistency deviation, Tsup is the supply air temperature of the cleanroom air conditioning system, and PRF is the radio frequency power of the etching process equipment. The second nonlinear coupling relationship is characterized by the following mapping function: Defect = f2(Rfresh, Exp), where Defect is the lithography defect rate, Rfresh is the fresh air ratio of the cleanroom air conditioning system, and Exp is the exposure energy of the lithography equipment. The third nonlinear coupling relationship is characterized by the following mapping function: DR = f3(Theat, Tsup), where DR is the thin film deposition rate deviation and Theat is the heating temperature of the thin film deposition equipment.

4. The energy consumption optimization and control method for the carbon neutralization process in a display panel production line according to claim 3, characterized in that, The mapping functions f1, f2, and f3 are each independently neural network models obtained by system identification based on the historical operating data of the display panel production line.

5. The energy consumption optimization and control method for a carbon neutralization process in a display panel production line according to claim 1, characterized in that, In step S5, the nonlinear multi-objective optimization objective function is solved by using a decomposition-based multi-objective evolutionary algorithm or a non-dominated sorting genetic algorithm with an elitist strategy to obtain the Pareto front solution set; from the Pareto front solution set, the final set of operating parameter optimization instructions is determined according to the preset carbon neutrality constraint priority rule.

6. An energy consumption optimization control system for a carbon neutralization process in a display panel production line, characterized in that, include: The data acquisition module is used to acquire real-time energy consumption data, production capacity cycle data, real-time energy structure data, and process quality correlation data of equipment in each process of the display panel production line; wherein, the equipment in each process includes at least photolithography equipment, etching equipment, thin film deposition equipment, and cleanroom air conditioning system; the process quality correlation data includes at least the exposure energy deviation value of photolithography process, the radio frequency power fluctuation value of etching process, the film thickness uniformity of thin film deposition process, and the supply air temperature difference and particulate matter concentration of cleanroom air conditioning system; A real-time carbon emission calculation module, connected to the data acquisition module, is used to calculate a dynamic carbon emission factor by weighting the green energy carbon emission factor and the fossil energy carbon emission factor according to the power distribution ratio of green energy and fossil energy in the current period. It then multiplies the real-time energy consumption data of each process equipment by the dynamic carbon emission factor to convert it into real-time carbon emission data, and establishes a production line-level carbon neutrality target constraint model. The carbon neutrality target constraint model includes an upper limit value for total carbon emissions per natural day, and this upper limit value can be dynamically corrected under dual closed-loop feedback control. A multi-process coupling model module, connected to the data acquisition module, is used to store and run the multi-process coupling model of the display panel production line. The multi-process coupling model includes: a first nonlinear coupling relationship between the supply air temperature of the cleanroom air conditioning system and the RF power of the etching process equipment on chip linewidth consistency; a second nonlinear coupling relationship between the fresh air ratio of the cleanroom air conditioning system and the stability of the exposure energy of the photolithography process equipment on the photolithography defect rate; and a third nonlinear coupling relationship between the heating temperature of the thin film deposition process equipment and the supply air temperature of the cleanroom air conditioning system on the thin film deposition rate. A nonlinear multi-objective optimization solution module, connected to the real-time carbon emission calculation module and the multi-process coupling model module, is used to construct a nonlinear multi-objective optimization objective function with the carbon neutrality target constraint model as the constraint, production line capacity achievement as the boundary condition, and the feasible region of the operating parameters of each process equipment in the multi-process coupling model as the process constraint. The module then solves the nonlinear multi-objective optimization objective function and outputs an optimization instruction set for the operating parameters of each process equipment. The nonlinear multi-objective optimization objective function aims to minimize total production line energy consumption, total production line carbon emissions, and overall yield loss. The overall yield loss is calculated as follows: Based on the first, second, and third nonlinear coupling relationships, the predicted mean and prediction uncertainty measure of each relationship output value are obtained. Based on each output value and its corresponding preset threshold, the probability of linewidth consistency exceeding the tolerance, the probability of lithography defect rate exceeding the standard, and the probability of deposition rate deviation exceeding the tolerance are calculated according to the predicted mean and prediction uncertainty measure. Then, based on the assumption that the yield links of each process are independent, the overall yield loss is calculated as 1 - (1 - probability of linewidth consistency exceeding the tolerance) × (1 - probability of lithography defect rate exceeding the standard) × (1 - probability of deposition rate deviation exceeding the standard). The instruction execution module, connected to the nonlinear multi-objective optimization solution module, is used to send the set of operating parameter optimization instructions to the controller of the corresponding process equipment to adjust the operating status of the equipment. Furthermore, a dual-loop feedback module, connected to the instruction execution module, the real-time carbon emission calculation module, and the nonlinear multi-objective optimization solution module, is used to acquire real-time carbon emission data of each process equipment after adjustment and the process quality correlation data; calculate the carbon emission deviation between the adjusted actual carbon emission and the upper limit of total carbon emissions in the carbon neutrality target constraint model; and calculate the yield deviation between the adjusted actual yield data and the target yield. The carbon emission deviation is used as a feedback signal to perform boundary correction on the upper limit of total carbon emissions in the carbon neutrality target constraint model. The yield deviation is also used as a feedback signal to dynamically adjust the weight for minimizing the comprehensive yield loss in the nonlinear multi-objective optimization objective function.

7. The energy consumption optimization control system for the carbon neutralization process of a display panel production line according to claim 6, characterized in that, The nonlinear multi-objective optimization solution module uses a decomposition-based multi-objective evolutionary algorithm or a non-dominated sorting genetic algorithm with an elitist strategy to obtain a Pareto front solution set; and determines the final set of operating parameter optimization instructions from the Pareto front solution set according to a preset carbon neutrality constraint priority rule.

8. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the energy consumption optimization control method for the carbon neutralization process of the display panel production line as described in any one of claims 1 to 5.

Images