Product environment-friendly aluminum alloy production method

Abstract

The invention relates to the technical field of aluminum alloy production, and discloses a product environment-friendly aluminum alloy production method which comprises the following steps of raw material screening and classification, raw material pretreatment, raw material recycling, raw material drying and finished product recycling. The screened and classified secondary aluminum serves as a main raw material to be subjected to pretreatment, efficient smelting, green refining, component fine adjustment, semi-solid casting and chromium-free surface treatment. According to the aluminum alloy green refining algorithm based on reinforcement learning, operation parameters such as the gas flow and the stirring speed serve as action spaces, the inclusion removal rate and the gas content reduction degree serve as reward functions, and through continuous trial and error learning and dynamic parameter adjustment, the algorithm can accurately adapt to the characteristics of different batches of raw materials, the quality fluctuation is reduced, and the refining efficiency is improved. And meanwhile, data are accumulated, a basis is provided for process optimization and equipment improvement, and the progress of an aluminum alloy refining technology is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of aluminum alloy production, and in particular to a method for producing an environmentally friendly aluminum alloy. Background Art

[0002] Currently, the industry's mainstream aluminum alloy refining technology still relies on traditional processes. During the refining process, companies generally use flux refining, which removes impurities by adding fluxes containing chemical components such as fluorine and chlorine. This process produces large amounts of toxic gases and difficult-to-treat waste slag. Regarding energy utilization, they rely heavily on high-energy-consuming equipment and extensive parameter control to complete refining tasks, lacking efficient energy utilization. Although some companies have introduced automated control systems and achieved programmatic parameter setting, these operations are still essentially based on fixed parameter ranges, and environmental protection measures are limited to end-of-pipe treatment, with simple emissions treatment through the installation of purification equipment.

[0003] However, existing technologies struggle to meet the high standards of green manufacturing. Traditional flux refining methods produce toxic gases and waste residues, which, even after final treatment, still pose a potential threat to the ecological environment. Extensive energy utilization leads to high energy consumption, which is inconsistent with the goal of low-carbon production. Fixed parameter control models are unable to adapt to the varying properties of different batches of raw materials, resulting in low refining efficiency and severe resource waste. Furthermore, the lack of dynamic optimization of the production process and deep data mining makes it difficult to achieve a synergistic improvement in environmental protection and production efficiency, hindering the aluminum alloy industry's transition to green and intelligent manufacturing. Therefore, we propose a product-friendly aluminum alloy production method. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides an environmentally friendly aluminum alloy production method, which solves the problem that the fixed parameter control mode in the existing technology cannot adapt to the differences in characteristics of different batches of raw materials, resulting in low refining efficiency and serious waste of resources.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for producing an environmentally friendly aluminum alloy, comprising the following steps:

[0006] S1: Raw material screening and classification steps

[0007] Screening usable materials from various waste aluminum alloy sources, performing detailed classification, and initially removing obvious impurities;

[0008] S2: Raw material pretreatment step

[0009] The screened and classified recycled aluminum is used as the main raw material, and is crushed, sorted and pre-treated to remove impurities. Metal silicon, magnesium ingots and a mixture of rare earth elements are added in precise proportions.

[0010] S3: Efficient Melting Step

[0011] Using electromagnetic induction heating furnace, in the temperature range of 720℃-760℃, with argon gas introduced at a steady flow rate for protection, the raw materials are melted for 1.5-2.5 hours, stirring at appropriate times during the period;

[0012] S4: Green Refining Step

[0013] Apply green self-cleaning physical refining technology for aluminum alloy melt, add environmentally friendly refining agent composed of fluorine-free and chlorine-free inorganic salts and surfactants, and stir thoroughly to make the refining agent evenly dispersed;

[0014] S5: Ingredient fine-tuning step

[0015] Conduct composition testing on the refined aluminum alloy liquid and adjust the alloy composition slightly to the preset standard based on the test results;

[0016] S6: Semi-solid casting step

[0017] The aluminum alloy liquid with fine-tuned composition is cooled to a semi-solid state and injected into the mold for casting in a specific manner while precisely controlling the mold cooling rate;

[0018] S7: Chrome-free surface treatment step

[0019] The aluminum alloy surface is treated with a chromium-free chemical conversion solution using cobalt salt as the main film-forming substance, and then coated with an environmentally friendly water-based paint and cured under specific low-temperature conditions.

[0020] Preferably, in the raw material screening and classification step, the various waste aluminum alloy sources include discarded aluminum alloy doors and windows, automobile parts, electronic equipment casings, and aerospace scrap aluminum materials, and are classified according to the alloy composition, impurity content, and physical form of the aluminum alloy, and obvious impurities including plastic, rubber, and other metal impurities are preliminarily removed.

[0021] Preferably, in the raw material pretreatment step, the crushing adopts a multi-stage crushing method, firstly coarse crushing is performed to crush the raw material into larger particles, and then fine crushing is performed to make it reach an appropriate particle size range; the sorting includes magnetic separation and flotation, and the deep impurity removal further removes tiny impurities through chemical cleaning and high-temperature roasting.

[0022] Preferably, in the efficient smelting step, the argon flow rate is controlled at 5-10 L / min to maintain a slightly positive pressure in the furnace, and the argon is dried and filtered before being introduced to ensure purity; the timely stirring adopts a combination of electromagnetic stirring and mechanical stirring, and the stirring frequency and intensity are dynamically adjusted according to the smelting process.

[0023] Preferably, in the green refining step, the physical refining technology includes introducing rotating argon bubbles into the alloy liquid, the rotation speed and bubble size of the argon bubbles are adjustable, and an electromagnetic stirring device is used, and the intensity and direction of the electromagnetic stirring are changed in real time according to the refining stage; the environmentally friendly refining agent is preheated before addition, and is added gradually and evenly. After addition, stirring is continued for a certain time to allow the refining agent to fully exert its effect.

[0024] Preferably, the refining process in the green refining step includes operating parameters such as gas flow rate, stirring speed, and refining agent addition as action space, and inclusion removal rate and gas content reduction degree indicators as reward functions, and a reinforcement learning model is constructed. The algorithm continuously learns by trial and error in the actual production environment, dynamically adjusts the operating parameters according to the refining effect monitored in real time, and automatically adapts to the differences in characteristics of different batches of aluminum alloy raw materials.

[0025] Preferably, in the semi-solid casting step, the volume fraction of solid phase particles in the alloy liquid in the semi-solid state is 20% to 40%, and the volume fraction of solid phase particles is accurately adjusted by controlling the cooling rate and stirring intensity, wherein the specific injection method is low-pressure injection or extrusion injection, and the appropriate injection method is selected according to the mold shape and size.

[0026] Preferably, a digital twin model of the aluminum alloy semi-solid casting process is established in the semi-solid casting step, and the actual production data is collected in real time by sensors to update the model parameters. The model predictive control algorithm is used to predict the defect risks in the future casting process according to the current production status and preset product quality goals, and the mold temperature distribution and slurry injection speed are optimized in advance. At the same time, the simulation results of the digital twin model are compared and analyzed with the actual production data.

[0027] Preferably, the temperature of the chromium-free chemical conversion solution during treatment is 30°C-40°C, the pH value is 5-6, the treatment time is determined according to the surface condition of the aluminum alloy and the requirements of the conversion film, and the aluminum alloy surface is subjected to degreasing and activation pretreatment operations before treatment.

[0028] Preferably, after the chromium-free surface treatment step, the trained GAN model inputs the image to be detected into the discriminator in the detection stage. If the discriminator determines that the image is significantly different from the normal image, it is marked as having a defect. At the same time, the image of the defective area is input into the generator, and the generator attempts to repair the defect and generates a simulated repaired image. By comparing the original image and the simulated repaired image, the severity and impact range of the defect are evaluated.

[0029] The present invention provides an environmentally friendly aluminum alloy production method. It has the following beneficial effects:

[0030] 1. The present invention adopts a reinforcement learning-based green refining algorithm for aluminum alloys. It uses operating parameters such as gas flow rate and stirring speed as the action space, and the inclusion removal rate and gas content reduction degree as the reward function. Through continuous trial and error learning and dynamic parameter adjustment, the algorithm can accurately adapt to the characteristics of different batches of raw materials, reduce quality fluctuations, and lower the risk of human intervention and errors. At the same time, it accumulates data to provide a basis for process optimization and equipment improvement, and promote the advancement of aluminum alloy refining technology.

[0031] 2. The present invention uses the discriminator in the algorithm to quickly and efficiently detect surface defects of aluminum alloys, reducing the risk of missed detection or false detection. At the same time, it uses the generator to repair the defective area, accurately locate the defect position, and reduce the workload of manual evaluation. Finally, by comparing the original and repaired images, the severity and impact range of the defects are quantitatively evaluated, providing a scientific basis for the formulation of repair plans and quality control, effectively improving product quality, increasing production efficiency, and reducing production costs.

[0032] 3. This invention, through the integration of a digital twin model with a model predictive control algorithm, can accurately predict defect risks, proactively optimize process parameters, reduce casting defect rates, and improve product quality stability. Real-time data acquisition and model updates enable the timely identification and resolution of potential issues, reducing trial and error and adjustment time while also lowering scrap rates, optimizing resource utilization, and effectively improving production efficiency and reducing costs. Furthermore, accumulated production data provides a basis for process improvement and innovation, accelerating technological iteration. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flow chart of the environmentally friendly aluminum alloy production method for this product;

[0034] Figure 2 This is a flow chart of the algorithm for dynamically adjusting operating parameters of the present invention;

[0035] Figure 3 Schematic diagram of the defect assessment algorithm flow of the present invention. DETAILED DESCRIPTION

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the specification of the present invention. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0037] Example:

[0038] Please see the attached Figure 1 -Attached Figure 3 The embodiment of the present invention provides a method for producing an environmentally friendly aluminum alloy, comprising the following steps:

[0039] S1: Raw material screening and classification steps

[0040] Screening usable materials from various waste aluminum alloy sources, performing detailed classification, and initially removing obvious impurities;

[0041] S2: Raw material pretreatment step

[0042] The screened and classified recycled aluminum is used as the main raw material, and is crushed, sorted and pre-treated to remove impurities. Metal silicon, magnesium ingots and a mixture of rare earth elements are added in precise proportions.

[0043] S3: Efficient Melting Step

[0044] Using electromagnetic induction heating furnace, in the temperature range of 720℃-760℃, with argon gas introduced at a steady flow rate for protection, the raw materials are melted for 1.5-2.5 hours, stirring at appropriate times during the period;

[0045] S4: Green Refining Step

[0046] Apply green self-cleaning physical refining technology for aluminum alloy melt, add environmentally friendly refining agent composed of fluorine-free and chlorine-free inorganic salts and surfactants, and stir thoroughly to make the refining agent evenly dispersed;

[0047] S5: Ingredient fine-tuning step

[0048] Conduct composition testing on the refined aluminum alloy liquid and adjust the alloy composition slightly to the preset standard based on the test results;

[0049] S6: Semi-solid casting step

[0050] The aluminum alloy liquid with fine-tuned composition is cooled to a semi-solid state and injected into the mold for casting in a specific manner while precisely controlling the mold cooling rate;

[0051] S7: Chrome-free surface treatment step

[0052] The aluminum alloy surface is treated with a chromium-free chemical conversion solution using cobalt salt as the main film-forming substance, and then coated with an environmentally friendly water-based paint and cured under specific low-temperature conditions.

[0053] In the raw material screening and classification step, the various sources of waste aluminum alloys include discarded aluminum alloy doors and windows, auto parts, electronic equipment casings, and aerospace scrap aluminum materials, and are classified according to the alloy composition, impurity content, and physical form of the aluminum alloys, and obvious impurities including plastics, rubber, and other metal impurities are preliminarily removed.

[0054] In the raw material pretreatment step, the crushing adopts a multi-stage crushing method, first coarse crushing is performed to crush the raw materials into larger particles, and then fine crushing is performed to make them reach the appropriate particle size range; the sorting includes magnetic separation and flotation. The deep impurity removal further removes tiny impurities through chemical cleaning and high-temperature roasting.

[0055] In the efficient smelting step, the argon flow rate is controlled at 5-10 L / min to maintain a slightly positive pressure in the furnace, and the argon is dried and filtered before being introduced to ensure purity; the timely stirring adopts a combination of electromagnetic stirring and mechanical stirring, and the stirring frequency and intensity are dynamically adjusted according to the smelting process.

[0056] In the green refining step, the physical refining technology includes introducing rotating argon bubbles into the alloy liquid, the rotation speed and bubble size of the argon bubbles are adjustable, and an electromagnetic stirring device is used. The intensity and direction of the electromagnetic stirring are changed in real time according to the refining stage; the environmentally friendly refining agent is preheated before addition, and is added gradually and evenly. After addition, stirring is continued for a certain period of time to allow the refining agent to fully exert its effect.

[0057] In the green refining step, the refining process includes the gas flow rate, stirring speed, and refining agent addition operating parameters as the action space, and the inclusion removal rate and gas content reduction degree indicators as the reward function. A reinforcement learning model is constructed. The algorithm continuously learns by trial and error in the actual production environment, dynamically adjusts the operating parameters according to the real-time monitoring of the refining effect, and automatically adapts to the characteristic differences of different batches of aluminum alloy raw materials. The following algorithm is established here:

[0058] Step 1: Define state, action, and reward functions

[0059] State

[0060] The state s is composed of relevant parameters of the refining process, such as the current inclusion content I, gas content G, refining time t, etc. These parameters can be discretized and combined into a multidimensional vector to represent the state, that is, s = (I, G, t).

[0061] Action

[0062] Action a is a combination of operating parameters, including gas flow rate F, stirring speed S, and refining agent addition amount A. The action space is a combination set of different values ​​of these parameters.

[0063] Reward Function

[0064] The reward function r is used to evaluate the quality of each action, according to the inclusion removal rate R inclusion and the gas content reduction degree R gasto define;

[0065] Assuming the inclusion removal rate is The gas content is reduced to Then the reward function r can be defined as:

[0066] r=w1R inclusion +w2R gas

[0067] Where w1 and w2 are weight coefficients, and w1+w2=1,w1,w2≥0;

[0068] Step 2: Initialize the Q table

[0069] The Q-table Q(s,a) is used to store the expected value of taking each action a in each state s. Initially, all values ​​in the Q-table are set to 0. For each state s in the state space and each action a in the action space, Q(s,a) = 0;

[0070] Step 3: Select an action

[0071] At each time step, the agent needs to choose an action a based on the current state s. Here, we use the ∈-greedy strategy to balance exploration and exploitation.

[0072]

[0073] Where ∈ is the exploration rate, which can be initially set to 1 and gradually decreases as the training progresses, for example, according to ∈=∈0×0.99 n where ∈ 0 is the initial exploration rate and n is the number of training rounds.

[0074] Step 4: Perform actions and observe environmental feedback

[0075] The agent performs the selected action a, which is to adjust the gas flow, stirring speed and refining agent addition. After a period of time, the new state s of the environment feedback is observed. ′ and reward r. New state s ′ =(I ′ ,G ′ ,t ′ ), where I ′ , G ′ is the new inclusion content and gas content, t ′ It's time for new refinement

[0076] Step 5: Update the Q table

[0077] Based on the observed reward r and the new state s′, the Q-learning update formula is used to update the values ​​in the Q table.

[0078]

[0079] in:

[0080] α is the learning rate, which ranges from (0,1] and controls the extent to which new information covers old information. For example, you can set α = 0.1.

[0081] γ is a discount factor in the range [0,1] that balances the importance of immediate rewards and future rewards. For example, you can set γ to 0.9.

[0082] max a′ Q(s′,a′) represents the maximum Q value among all possible actions a′ taken in the next state s′.

[0083] Step 6: Determine whether it is finished

[0084] Determine whether the refining process has reached the termination condition, such as the refining time reaching the preset maximum value, or the inclusion content and gas content both reaching the expected standards. If the termination condition is met, the current training round ends; otherwise, the new state s′ is used as the current state and the training is continued in step 3.

[0085] Step 7: Repeat the training

[0086] Repeat steps 3-6 for multiple rounds of training. As the number of training rounds increases, the values ​​in the Q table will gradually converge, and the agent will be able to learn the optimal action strategy under different states.

[0087] In the semi-solid casting step, the volume fraction of solid phase particles in the alloy liquid in the semi-solid state is 20% to 40%, and the volume fraction of the solid phase particles is accurately adjusted by controlling the cooling rate and stirring intensity. The specific injection method is low-pressure injection or extrusion injection, and the appropriate injection method is selected according to the mold shape and size.

[0088] In the semi-solid casting step, a digital twin model of the aluminum alloy semi-solid casting process is established. Actual production data is collected in real time through sensors to update model parameters. The model predictive control algorithm is used to predict the risk of defects in the future casting process based on the current production status and preset product quality goals. The mold temperature distribution and slurry injection speed are optimized in advance. At the same time, the simulation results of the digital twin model are compared and analyzed with the actual production data. The algorithm established here is as follows:

[0089] Step 1: Establish a digital twin model of aluminum alloy semi-solid casting

[0090] Algorithm and formula: Based on the heat conduction equation and fluid mechanics equation to describe the casting process, the three-dimensional unsteady heat conduction equation is:

[0091]

[0092] Where ρ is the material density, c p is the specific heat capacity, T is the temperature, t is the time, k is the thermal conductivity, and Q is the internal heat source term;

[0093] Machine learning modeling: A neural network is used to learn the nonlinear relationship between process parameters and defects. Taking the LSTM unit as an example, its core formula is:

[0094] Forget gate:f t =σ(W f ·[h t-1 ,x t ]+b f )

[0095] Input gate:i t =σ(W i ·[h t-1 ,x t ]+b i )

[0096] Cell status update:

[0097] Output gate: o t =σ(W o ·[h t-1 ,x t ]+b o )

[0098] Cell status:

[0099] Hidden state:h t =o t tanh(C t )

[0100] Among them, σ is the Sigmoid function, tanh is the hyperbolic tangent function, W and b are weights and biases, and h t is the hidden state, x t For input data.

[0101] Step 2: Real-time data collection and model update

[0102] Algorithms and formulas:

[0103] Data (such as temperature, pressure, and slurry flow rate) is collected in real time through sensors, and the Kalman filter algorithm is used to reduce noise and fuse the data. The formula is as follows:

[0104] Prediction steps:

[0105]

[0106] Update steps:

[0107]

[0108] P t|t =(IK t H t )P t|t-1

[0109] in, is the state estimation, P is the covariance matrix, F is the state transfer matrix, B is the control matrix, u is the control input, H is the observation matrix, z is the observation value, Q and R are the noise covariance, and K is the Kalman gain;

[0110] Input the processed data into the digital twin model and update the model parameters (such as thermal conductivity and fluid viscosity);

[0111] Step 3: Defect risk prediction and parameter optimization based on MPC

[0112] Algorithm and formula: Use discrete time state space model to describe system dynamics:

[0113]

[0114] Among them, x(k) is the state vector (such as temperature distribution, slurry flow rate), u(k) is the control input (mold temperature, injection speed), and y(k) is the output (casting quality index);

[0115] Rolling optimization: At each sampling time k, solve the optimization problem:

[0116]

[0117] Constraints:

[0118] x m in≤x(k+i|k)≤x max

[0119] u m in≤u(k+i|k)≤u max

[0120] Among them, N y is the prediction time domain, N u To control the time domain, Q and R are weight matrices, and r(k+i) is the reference trajectory;

[0121] Solve the optimization problem: Use a quadratic programming (QP) algorithm (such as the interior point method) to calculate the optimal control sequence u * , and only execute the first control quantity u(k);

[0122] Step 4: Comparative analysis of simulation results and actual data

[0123] Algorithm and formula: Calculate the mean square error (MSE) to evaluate the model accuracy:

[0124]

[0125] If the MSE exceeds the threshold, the model correction mechanism is triggered:

[0126] Use the back propagation (BP) algorithm to update the neural network parameters or adjust the coefficients of the physical model (such as thermal conductivity k);

[0127] In the correction formula, the calculation error is:

[0128] δ=(y sim -y actual )·f′(z)

[0129] Update weights:

[0130] w ji =w ji -α·δ·x i

[0131] Among them, α is the learning rate, f'(z) is the activation function derivative

[0132] Step 5: Loop Iteration Optimization

[0133] Repeat steps 2-4, continuously update the model based on real-time data, optimize control parameters, and achieve continuous closed-loop control. In each iteration, the algorithm predicts defect risks based on the latest digital twin model and adjusts process parameters, ultimately achieving dynamic optimization of the casting process.

[0134] The temperature of the chromium-free chemical conversion solution during treatment is 30° C.-40° C., the pH value is 5-6, and the treatment time is determined according to the surface condition of the aluminum alloy and the requirements of the conversion film. Before treatment, the aluminum alloy surface is subjected to degreasing and activation pretreatment operations.

[0135] After the chromium-free surface treatment step, the trained GAN model inputs the image to be detected into the discriminator in the detection phase. If the discriminator determines that the image is significantly different from the normal image, it is marked as having a defect. At the same time, the image of the defective area is input into the generator, which attempts to repair the defect and generates a simulated repaired image. By comparing the original image and the simulated repaired image, the severity and impact range of the defect are evaluated. The following algorithm is established here:

[0136] Step 1: Data collection and preprocessing

[0137] A large number of images of normal aluminum alloy surfaces after treatment are collected as real data. The images are uniformly adjusted to the same size (such as H×W×C, where H is the height, W is the width, and C is the number of channels), and normalized to scale the pixel values ​​to the range of [-1, 1].

[0138] Normalization formula: Where x is the original pixel value, x n orm is the normalized pixel value;

[0139] Step 2: Define the generator and discriminator network structures

[0140] Generator G: usually adopts a deconvolutional network structure, with the input being a random noise vector z and the output being a generated image G(z) of the same size as the real image;

[0141] Discriminator D: uses a convolutional network structure, with an input image (real image x or generated image G(z)) and an output scalar indicating the probability that the image is a real image;

[0142] The specific calculation process of the generator and discriminator consists of the convolution layer, deconvolution layer, activation function, etc. in the network. Taking the convolution layer as an example, the calculation formula of the output feature map y is:

[0143]

[0144] Where x is the input feature map, w is the convolution kernel weight, b is the bias, M and N are the height and width of the convolution kernel, and C in is the number of input channels;

[0145] Step 3: GAN training

[0146] GAN training is based on the minimax game, which alternately updates the parameters of the generator and discriminator.

[0147] Discriminator training:

[0148] Discriminator loss function L D :

[0149]

[0150] Update the discriminator parameters θ by gradient ascent D :

[0151]

[0152] Where α is the learning rate, is the gradient of the loss function with respect to the discriminator parameters;

[0153] Generator training:

[0154] Generator loss function L G :

[0155]

[0156] Update the generator parameters θ by gradient descent G :

[0157]

[0158] Step 4: Detection Phase - Defect Marking

[0159] The image to be detected x test Input it into the trained discriminator D, and judge whether there is a defect based on the probability value output by the discriminator;

[0160] Set the threshold τ, if D(x test )<τ, then the image is marked as defective;

[0161] Step 5: Detection Phase - Defect Fixing

[0162] Mark the image region x as a defect defect Input into the trained generator G to generate the simulated restored image G(x defect ).

[0163] This step is mainly the forward propagation calculation of the generator, that is, G(x defect ) is the generator network for the input x defect Output after convolution, deconvolution, activation, etc.

[0164] Step 6: Detection Phase - Defect Assessment

[0165] By comparing the original defect image x defect and simulated restored image G(x defect ), the mean square error (MSE) is used to evaluate the severity and impact range of the defect;

[0166]

[0167] Where m and n are the height and width of the image respectively, C is the number of channels, and x defect (i,j,k) and G(x defect )(i,j,k) are the pixel values ​​of the original defect image and the simulated repaired image at position (i,j,k), respectively.

[0168] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for producing an environmentally friendly aluminum alloy, characterized in that: The following steps are involved: S1: Raw material screening and classification steps Screening usable materials from various waste aluminum alloy sources, performing detailed classification, and initially removing obvious impurities; S2: Raw material pretreatment step The screened and classified recycled aluminum is used as the main raw material, and is crushed, sorted and pre-treated to remove impurities. Metal silicon, magnesium ingots and a mixture of rare earth elements are added in precise proportions. S3: Efficient Melting Step Using electromagnetic induction heating furnace, in the temperature range of 720℃-760℃, with argon gas introduced at a steady flow rate for protection, the raw materials are melted for 1.5-2.5 hours, stirring at appropriate times during the period; S4: Green Refining Step Apply green self-cleaning physical refining technology for aluminum alloy melt, add environmentally friendly refining agent composed of fluorine-free and chlorine-free inorganic salts and surfactants, and stir thoroughly to make the refining agent evenly dispersed; S5: Ingredient fine-tuning step Conduct composition testing on the refined aluminum alloy liquid and adjust the alloy composition slightly to the preset standard based on the test results; S6: Semi-solid casting step The aluminum alloy liquid with fine-tuned composition is cooled to a semi-solid state and injected into the mold for casting in a specific manner while precisely controlling the mold cooling rate; S7: Chrome-free surface treatment step The aluminum alloy surface is treated with a chromium-free chemical conversion solution using cobalt salt as the main film-forming substance, and then coated with an environmentally friendly water-based paint and cured under specific low-temperature conditions.

2. The environmentally friendly aluminum alloy production method according to claim 1, characterized in that: In the raw material screening and classification step, the various sources of waste aluminum alloys include discarded aluminum alloy doors and windows, auto parts, electronic equipment casings, and aerospace scrap aluminum materials, and are classified according to the alloy composition, impurity content, and physical form of the aluminum alloys, and obvious impurities including plastics, rubber, and other metal impurities are preliminarily removed.

3. The environmentally friendly aluminum alloy production method according to claim 1, characterized in that: In the raw material pretreatment step, the crushing adopts a multi-stage crushing method, first coarse crushing is performed to crush the raw materials into larger particles, and then fine crushing is performed to make them reach the appropriate particle size range; the sorting includes magnetic separation and flotation. The deep impurity removal further removes tiny impurities through chemical cleaning and high-temperature roasting.

4. The environmentally friendly aluminum alloy production method according to claim 1, characterized in that: In the efficient smelting step, the argon flow rate is controlled at 5-10 L / min to maintain a slightly positive pressure in the furnace, and the argon is dried and filtered before being introduced to ensure purity; the timely stirring adopts a combination of electromagnetic stirring and mechanical stirring, and the stirring frequency and intensity are dynamically adjusted according to the smelting process.

5. The environmentally friendly aluminum alloy production method according to claim 1, characterized in that: In the green refining step, the physical refining technology includes introducing rotating argon bubbles into the alloy liquid, the rotation speed and bubble size of the argon bubbles are adjustable, and an electromagnetic stirring device is used. The intensity and direction of the electromagnetic stirring are changed in real time according to the refining stage; the environmentally friendly refining agent is preheated before addition, and is added gradually and evenly. After addition, stirring is continued for a certain period of time to allow the refining agent to fully exert its effect.

6. The environmentally friendly aluminum alloy production method according to claim 5, characterized in that: The green refining step uses the gas flow rate, stirring speed, and refining agent addition operating parameters in the refining process as the action space, and the inclusion removal rate and gas content reduction degree indicators as the reward function to construct a reinforcement learning model. The algorithm continuously learns by trial and error in the actual production environment, dynamically adjusts the operating parameters according to the real-time monitoring of the refining effect, and automatically adapts to the differences in characteristics of different batches of aluminum alloy raw materials.

7. The environmentally friendly aluminum alloy production method according to claim 1, characterized in that: In the semi-solid casting step, the volume fraction of solid phase particles in the alloy liquid in the semi-solid state is 20% to 40%, and the volume fraction of the solid phase particles is accurately adjusted by controlling the cooling rate and stirring intensity. The specific injection method is low-pressure injection or extrusion injection, and the appropriate injection method is selected according to the mold shape and size.

8. The environmentally friendly aluminum alloy production method according to claim 7, characterized in that: In the semi-solid casting step, a digital twin model of the aluminum alloy semi-solid casting process is established, and actual production data is collected in real time by sensors to update the model parameters. The model predictive control algorithm is used to predict the defect risks in the future casting process based on the current production status and preset product quality goals, and optimize the mold temperature distribution and slurry injection speed in advance. At the same time, the simulation results of the digital twin model are compared and analyzed with the actual production data.

9. The environmentally friendly aluminum alloy production method according to claim 1, characterized in that: The temperature of the chromium-free chemical conversion solution during treatment is 30° C.-40° C., the pH value is 5-6, and the treatment time is determined according to the surface condition of the aluminum alloy and the requirements of the conversion film. Before treatment, the aluminum alloy surface is subjected to degreasing and activation pretreatment operations.

10. The environmentally friendly aluminum alloy production method according to claim 9, characterized in that: After the chromium-free surface treatment step, the trained GAN model inputs the image to be detected into the discriminator in the detection stage. If the discriminator determines that the image is significantly different from the normal image, it is marked as having a defect. At the same time, the image of the defective area is input into the generator, which attempts to repair the defect and generates a simulated repaired image. By comparing the original image and the simulated repaired image, the severity and impact range of the defect are evaluated.

Images