Genetic algorithm-based polycaprolactone polyol synthesis path optimization method

Abstract

The invention relates to the technical field of data processing, in particular to a polycaprolactone polyol synthesis path optimization method based on a genetic algorithm, which comprises the following steps: constructing an antagonism framework taking the genetic algorithm as a generator and a process stability discriminator; directionally generating virtual failure data by using a failure path deduction model so as to solve the problem of data sparsity and train a discriminator; quantifying the robustness risk of the candidate process parameters through a virtual disturbance unit, and taking the risk as a key part of a fitness function of the genetic algorithm; meanwhile, the interpretability analysis result of the discriminator is used for guiding the variation direction of the genetic algorithm. According to the method, the process path with high performance and high robustness can be found, and the industrial practicability and decision-making efficiency of an optimization result are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a polycaprolactone polyol synthesis path optimization method based on a genetic algorithm. BACKGROUND

[0002] Heuristic search models such as genetic algorithms are key computing tools for solving complex engineering optimization problems. For example, in the synthesis path optimization of high-performance materials such as polycaprolactone polyols, how to apply these computing models to deal with real-world systems with uncertainty and sparse failure data, and ensure the robustness and practicality of the optimization results, is a core challenge in the field of artificial intelligence.

[0003] Existing computing methods usually couple genetic algorithms with physical models or data-driven surrogate models. The former is inefficient due to high evaluation cost. The latter generally uses models such as artificial neural networks as surrogates, but as a black box model, the prediction logic is difficult to explain, and its training effect is highly dependent on data quality; in industrial applications such as polycaprolactone synthesis, failure condition data representing unqualified product quality is extremely sparse, resulting in poor model generalization ability and inability to accurately assess risks. Although existing research has attempted to use generative adversarial networks to enhance data, conventional generative adversarial networks have problems such as unstable training and pattern collapse when dealing with high-dimensional and complex process data, making it difficult to generate high-fidelity virtual samples that meet physical laws. More importantly, regardless of the coupling method, existing technologies mostly find isolated optimal points, lack mechanisms to directly quantify and optimize the robustness of solutions to parameter perturbations at the algorithm level, and lack feedback loops that use internal insights from the model to intelligently guide evolutionary search.

[0004] Therefore, there is an urgent need in the art for a new computing optimization method to solve the core algorithmic challenges encountered in the optimization of complex processes such as polycaprolactone synthesis: how to create high-quality virtual data through a failure path deduction model to train an accurate process stability discriminator when negative sample data is sparse; and how to directly quantify and optimize the robustness of solutions to parameter perturbations in evolutionary search.

[0005] To this end, a polycaprolactone polyol synthesis path optimization method based on a genetic algorithm is proposed. SUMMARY

[0006] The present application aims to provide a polycaprolactone polyol synthesis path optimization method based on a genetic algorithm, to solve the technical problem that the production decision-making has great uncertainty due to the inability to construct an accurate risk prediction model because of the sparse production failure case data in the prior art. The present application constructs an adversarial optimization framework containing a generator and a discriminator, generates high-quality virtual failure data using a failure path deduction model to enhance the training of the process stability discriminator, and combines the robustness risk of the process parameters under real fluctuations using a virtual disturbance unit to quantify the robustness risk, finally incorporates the risk indicator into the fitness function of the genetic algorithm for iterative optimization, thereby balancing product performance and production stability to provide managers with production strategies with the highest business return on investment.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solutions: The polycaprolactone polyol synthesis path optimization method based on a genetic algorithm comprises a genetic algorithm as a generator and a process stability discriminator. The failure path deduction model is trained using historical synthesis data to learn the conditional probability distribution of synthesis parameters consisting of reaction temperature, reaction time, monomer to initiator molar ratio, and catalyst to initiator molar ratio under the condition of a given failure mode label; and virtual failure data is generated by inputting the failure mode label into the model. For the candidate synthesis parameters generated by the genetic algorithm, parameter clouds are generated in the neighborhood of the candidate synthesis parameters by a virtual disturbance unit; the failure risk values of the disturbed parameter points in the parameter clouds are evaluated by the process stability discriminator to construct a strategy volatility risk matrix; and the strategy volatility risk matrix is aggregated to generate a robustness risk indicator representing the overall failure risk of the candidate synthesis parameters. The robustness risk indicator is input into the fitness function of the genetic algorithm, and the results of the explainability analysis of the process stability discriminator are used to guide the mutation operator of the genetic algorithm to accelerate the convergence of the optimization process to the robustness optimal solution.

[0008] Preferably, the training step of the process stability discriminator comprises: The historical synthesis data and the virtual failure data are merged to form an enhanced training set; The process stability discriminator is supervised learning trained using the enhanced training set, so that the discriminator can input a set of synthesis parameters consisting of reaction temperature, reaction time, monomer to initiator molar ratio, and catalyst to initiator molar ratio, and output the corresponding failure risk probability.

[0009] Preferably, the explainability analysis result is quantified by applying a SHAP analysis method to the process stability discriminator, to quantify the positive and / or negative contribution of each candidate synthesis parameter to the overall failure risk; and the mutation operator is specifically configured to dynamically adjust the mutation probability and mutation direction of the corresponding parameter gene according to the contribution, to apply reverse mutation to the parameter that contributes to the positive risk, and to increase the mutation probability of the parameter.

[0010] Preferably, the failure path inference model comprises: a forward diffusion unit configured to gradually add Gaussian noise to the real synthesis parameter in the historical synthesis data through a Markov chain until the distribution of the real synthesis parameter is a standard normal distribution; a conditional denoising unit configured as a trained neural network, which receives the noisy synthesis parameter at a time step and the failure mode label as common inputs at any time step of the forward diffusion process, and predicts the noise added to the noisy synthesis parameter; and a sampling generation unit configured to sample an initial noise from a standard normal distribution, and to gradually denoise by iteratively calling the conditional denoising unit under the guidance of the failure mode label, to reversely reconstruct the virtual failure data.

[0011] Preferably, the virtual disturbance unit is specifically a parameterized noise generation module: For each parameter in the candidate synthesis parameter, multiple random samplings are independently performed from a probability distribution based on a process tolerance range matched with the physical characteristics of the parameter, to generate the parameter cloud, wherein the process tolerance range is set based on statistical fluctuations of historical production data and / or expert experience.

[0012] Preferably, the generation of the robust risk indicator representing the overall failure risk of the candidate synthesis parameter comprises: receiving a single candidate synthesis parameter generated by the genetic algorithm; the virtual disturbance unit generates a parameter cloud containing multiple disturbed parameter points around the single candidate synthesis parameter by multiple random samplings; the failure risk values corresponding to the disturbed parameter points in the parameter cloud are constructed into a strategy volatility risk matrix; and a single robust risk indicator representing the overall failure risk of the candidate synthesis parameter is generated by aggregating the strategy volatility risk matrix.

[0013] Preferably, the robustness risk indicator is input into the fitness function of the genetic algorithm, specifically, a composite fitness function is constructed, which combines the robustness risk indicator as a negative weighted item with a positive weighted item based on the expected product performance indicator of the candidate synthesis parameters; the mutation operator of the genetic algorithm is guided, specifically, the key risk parameters that have positive contribution to the overall failure risk are determined by using the explainability analysis results, and when performing the mutation operation, the probability of the key risk parameters being selected for mutation is increased, and a directional mutation bias that reduces the key risk parameters is applied.

[0014] Preferably, the aggregation processing, specifically, the maximum value and / or a preset high percentile value are selected from the failure risk values of all perturbed parameter points in the parameter cloud as the robustness risk indicator, to evaluate the worst possible performance under process fluctuations.

[0015] Preferably, the expected product performance indicator is calculated by a proxy model trained on the historical synthesis data for predicting product performance, which receives the candidate synthesis parameters as input and outputs performance prediction values including expected molecular weight and expected polydispersity index.

[0016] Compared with the prior art, the present application has the following beneficial effects: 1. The technical problem of being difficult to construct an accurate risk model due to sparse failure case data in industrial production is solved. The present application generates high-quality and diversified virtual failure data by using the failure path deduction model, significantly enhances the training set of the process stability discriminator, and enables it to accurately predict the potential failure risk of different process paths, thereby providing reliable data support for production decisions.

[0017] 2. The industrial practicability and robustness of the process optimization result are significantly improved. The present application simulates process fluctuations in real production through the virtual perturbation unit, and incorporates the quantified robustness risk into the optimization objective, so that the finally found process parameters not only have the best performance in theory, but also have a wider process window, which can effectively resist disturbances in actual production, thereby reducing the batch failure rate and improving the stability of product quality.

[0018] 3. A more efficient and intelligent optimization process is realized. By introducing explainability analysis to guide the mutation direction of the genetic algorithm, the present application changes the optimization process from a "black box" search to a "white box" evolution with clear guidance and the ability to learn from mistakes, avoiding a large amount of invalid exploration, accelerating the convergence speed to the robustness optimal solution, and improving the decision-making efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0019] Fig. 1The construction flow chart of the polycaprolactone polyol synthesis path optimization method based on a genetic algorithm for the embodiment of the present application is shown in the figure. Fig. 2 The method flow chart of the polycaprolactone polyol synthesis path optimization method based on a genetic algorithm for the embodiment of the present application is shown in the figure. Fig. 3 The model structure diagram of the failure path deduction model for the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0021] Embodiment one Please refer to Figs. 1 to 3 The present application provides a polycaprolactone polyol synthesis path optimization method based on a genetic algorithm, and the technical solutions are as follows: The polycaprolactone polyol synthesis path optimization method based on a genetic algorithm, as shown in the figures Fig. 1 and Fig. 2 includes a genetic algorithm as a generator and a process stability discriminator; The failure path deduction model is trained using historical synthesis data to learn the conditional probability distribution of synthesis parameters consisting of reaction temperature, reaction time, monomer to initiator molar ratio, and catalyst to initiator molar ratio under the condition of a given failure mode label; and by inputting the failure mode label to the model, virtual failure data is generated in a targeted manner; For the candidate synthesis parameters generated by the genetic algorithm, the parameter cloud is generated in the neighborhood of the candidate synthesis parameters by the virtual disturbance unit; the failure risk values of the disturbed parameter points in the parameter cloud are evaluated by the process stability discriminator to construct a strategy volatility risk matrix; and the strategy volatility risk matrix is aggregated to generate a robustness risk indicator representing the overall failure risk of the candidate synthesis parameters; The robustness risk indicator is input into the fitness function of the genetic algorithm, and the explainability analysis result of the process stability discriminator is used to guide the mutation operator of the genetic algorithm to accelerate the convergence of the optimization process to the robustness optimal solution.

[0022] Further, the optimization process adopts a two-stage hierarchical optimization strategy: the first stage is a coarse-grained performance optimization stage, the genetic algorithm uses a simplified fitness function containing only the expected product performance indicators for iteration to quickly locate several high-performance regions in the entire parameter space, and obtain an elite population composed of candidate synthesis parameters with excellent performance; the second stage is a fine-grained robustness tuning stage, taking the elite population generated in the first stage as the initial population, and switching to a composite fitness function containing the robustness risk indicators for iterative optimization, thereby concentrating computing resources for fine robustness optimization within the determined high-performance region.

[0023] This hierarchical strategy significantly improves the overall optimization efficiency. It avoids time-consuming robustness risk assessment of a large number of low-performance solutions in the early optimization stage, and instead concentrates valuable computing resources in the most promising candidate regions, thereby significantly shortening the total time required to find the optimal solution and reducing the computational cost while ensuring the quality of the final solution.

[0024] Further, the training step of the process stability discriminator includes: Merge the historical synthesis data and the virtual failure data to form an enhanced training set; Use the enhanced training set to perform supervised learning training on the process stability discriminator, so that the discriminator can input a set of synthesis parameters composed of reaction temperature, reaction time, monomer to initiator molar ratio, and catalyst to initiator molar ratio, and output the corresponding failure risk probability.

[0025] By merging a small amount of real historical data with a large amount of generated virtual failure data, the key technical problem of being unable to train an effective risk model due to the sparsity of failure cases and the lack of negative samples in industrial applications is effectively solved. The data-balanced enhanced training set significantly improves the prediction accuracy and generalization ability of the process stability discriminator, enabling it to accurately identify potential process risks and providing a solid and reliable decision basis for the subsequent robustness optimization process.

[0026] Further, the explainability analysis result is obtained by applying the SHAP analysis method to the process stability discriminator, quantifying the positive and / or negative contribution of each parameter in the candidate synthesis parameter to the overall failure risk; the mutation operator specifically dynamically adjusts the mutation probability and mutation direction of the corresponding parameter gene according to the contribution, applies reverse mutation to parameters that contribute positively to risk, and increases the mutation probability of the parameters.

[0027] For each of the candidate synthesis parameters, the probability of being selected for mutation is dynamically calculated. First, the absolute value of the risk contribution of each parameter is calculated, and the absolute values of the risk contributions of all parameters are added to obtain a total. Then, the relative risk importance of the individual parameter is obtained by dividing the absolute value of the risk contribution of the parameter by the total. Finally, the probability of the parameter being selected for mutation is equal to a preset base probability plus the product of the relative risk importance of the parameter and an influence factor. The greater the risk impact of the parameter, the greater the likelihood of being selected for modification, while all parameters have a basic mutation opportunity.

[0028] When a parameter is selected for mutation according to its probability, the new value is calculated as follows: First, determine whether the risk contribution of the parameter is positive or negative. Then, calculate the mutation amplitude. The amplitude is equal to a preset "base step size" multiplied by "the absolute value of the risk contribution of the parameter plus one". The greater the risk contribution of the parameter, the greater the adjustment amplitude. Finally, update the parameter value according to the risk direction determined in the first step. If the risk contribution is positive, subtract the calculated mutation amplitude from the original parameter value; if the risk contribution is negative, add the calculated mutation amplitude to the original parameter value. At the same time, boundary check is needed for the calculated new parameter value to ensure that its value is within the preset reasonable process range.

[0029] The genetic algorithm is transformed from "blind trial and error" to "intelligent optimization". Through SHAP analysis, the contribution of each parameter to the failure risk is accurately located, and the algorithm can preferentially and directionally correct the parameter with the highest risk. This strategy not only significantly speeds up the search for the optimal solution, improving decision-making efficiency, but more importantly, it can find a process path that combines high performance and high robustness. The final solution can better resist process fluctuations in actual production, thereby reducing the failure rate and improving industrial applicability.

[0030] Further, as shown in Fig. 3 The failure path derivation model comprises: a forward diffusion unit that gradually adds Gaussian noise to the real synthesis parameters in the historical synthesis data through a Markov chain until their distribution is a standard normal distribution; a conditional denoising unit, which is a trained neural network, receives the noisy synthesis parameters at any time step of the forward diffusion process and the failure mode label as common input, and predicts the noise added to the noisy synthesis parameters; and a sampling generation unit for sampling an initial noise from a standard normal distribution and iteratively calling the conditional denoising unit to gradually denoise under the guidance of the failure mode label to reversely reconstruct the virtual failure data.

[0031] The conditional denoising unit is specifically a neural network based on a U-Net architecture. The network includes an encoder path for down-sampling, a bottleneck layer, and a decoder path for up-sampling that receives information from the encoder using a skip connection. In the residual blocks of each path, a self-attention module can also be selectively added to capture the internal dependencies between parameters.

[0032] The specific structure of the neural network based on the U-Net architecture is as follows: both the encoder and decoder paths of the network include 4 levels. The number of channels of the convolutional blocks of the encoder path is [64, 128, 256, 512] in turn, each convolutional block includes two convolutional layers (with a convolution kernel size of 3x1 and a padding of 1), followed by a SiLU activation function and a batch normalization layer. A convolution with a step size of 2 is used for down-sampling at the end of each level. The decoder path is symmetrical to the encoder and uses transposed convolution for up-sampling. The number of channels of the bottleneck layer is 1024. A self-attention module is added in the 3rd level, using the standard scaled dot-product attention mechanism.

[0033] The forward diffusion unit adopts a linear noise scheduling scheme. In a Markov chain with a total of T (for example, T = 1000) time steps, the variance of the Gaussian noise added at the t-th step is increased linearly from a small initial value (for example, 0.0001) to a larger end value (for example, 0.02).

[0034] The failure mode label is converted into a conditional embedding vector through an embedding layer. The vector is added to the time embedding of the current time step t, and then processed through a small feedforward network, and then applied as a bias term or scaling term after the normalization layer of each residual block in the U-Net architecture, thereby guiding the denoising process at each step.

[0035] The model can generate virtual failure data for supplementing the training set with high quality and controllability, to solve the problem of the sparsity of real failure cases in industrial applications. The use of the U-Net architecture ensures that the generated virtual data is highly similar in distribution to the real process parameters, ensuring the authenticity of the data. The ingenious label embedding guidance mechanism enables the model to generate failure data of specific types according to specific requirements, achieving controllability. This high-quality and controllable virtual data can significantly enhance the training effect of the downstream process stability discriminator, making its prediction more accurate and reliable.

[0036] Further, the population selection strategy of the genetic algorithm adopts an elite preservation strategy combined with tournament selection, wherein the elite proportion is 0.1, and the tournament size is 3. The crossover operation adopts simulated binary crossover, and the crossover probability is set to 0.9; the mutation operation adopts the directed mutation operator guided by the explainability analysis result.

[0037] The population number of the genetic algorithm is set to 120, and the iteration number is 250 generations. Each individual represents a candidate polycaprolactone polyol synthesis path, and the parameters are optimized in the following preset ranges: the reaction temperature is [120.0, 160.0] ℃, the residence time is [30, 120] minutes, the monomer to initiator molar ratio is [50, 200], and the catalyst to initiator molar ratio is [0.001, 0.01].

[0038] Further, the virtual disturbance unit is specifically a parameterized noise generation module. For each parameter in the candidate synthesis parameters, according to a process tolerance range matched with the physical characteristics of the parameter, multiple random samplings are independently performed from a probability distribution to generate the parameter cloud, wherein the process tolerance range is set based on statistical fluctuations of historical production data and / or expert experience.

[0039] The probability distribution is specifically a truncated normal distribution. The mean of the distribution is the numerical value of the current candidate parameter, the standard deviation is set according to the process tolerance range, and the truncation boundary is the upper and lower limits of the process tolerance range. This can not only simulate the characteristics that fluctuations in the real process are more frequent around the center value, but also ensure that the sampling points will not exceed the physically or experientially allowed range.

[0040] The process tolerance range is quantitatively set as follows: for parameters with sufficient historical data, the range is set to plus or minus three times the standard deviation of the historical data mean; for parameters lacking data or specified by experts, the operating upper and lower limits given by the experts are directly used as their tolerance range.

[0041] The multiple random samplings are specifically to generate a preset number N of parameter points, which together constitute the parameter cloud. The number N is determined after balancing the stability of the evaluation results and the calculation efficiency. In this embodiment, N = 100.

[0042] By using the tolerance range set based on historical data statistics and using the truncated normal distribution for sampling, the generated parameter cloud can accurately reflect the fluctuation characteristics of various parameters in the real world. This makes the robustness evaluation results of the candidate process scheme more accurate and reliable.

[0043] Further, the disturbance amplitude applied by the virtual disturbance unit is dynamically self-adaptive: in the early stage of the optimization process, a larger disturbance amplitude is adopted to quickly eliminate the candidate synthesis parameters that are extremely sensitive to process fluctuations; in the middle and later stages of the optimization process, the average fitness change rate of the genetic algorithm population is monitored, and when the change rate is lower than a preset convergence threshold, the disturbance amplitude is automatically reduced to perform a more refined local robustness search in the neighborhood of the current optimal solution; when the change rate is lower than the convergence threshold for a plurality of generations, the disturbance amplitude is temporarily increased to stress test the current optimal solution and assist the algorithm in jumping out of the local optimum.

[0044] The adaptive adjustment mechanism realizes the intelligent balance between global exploration and local optimization of the algorithm; the large disturbance in the early stage ensures the breadth of the search, which can effectively avoid falling into a local optimum; the fine adjustment in the later stage ensures the depth of the search, which can more accurately depict the robustness boundary of the optimal solution; this dynamic strategy makes the optimization process more intelligent and efficient.

[0045] Further, the generation process of the robustness risk indicator includes: receiving a single candidate synthesis parameter generated by the genetic algorithm; the virtual disturbance unit generates a parameter cloud containing a plurality of disturbed parameter points around the single candidate synthesis parameter through multiple random samplings; the failure risk values of the disturbed parameter points in the parameter cloud are constructed into a strategy volatility risk matrix; and the strategy volatility risk matrix is aggregated to generate a single robustness risk indicator representing the overall failure risk of the candidate synthesis parameter.

[0046] The aggregation processing is specifically selecting the maximum value and / or a preset high percentile value (e.g., the 95th percentile value) of the failure risk values of all disturbed parameter points in the parameter cloud as the single robustness risk indicator, which is used to evaluate the worst possible performance under process fluctuations.

[0047] By selecting the maximum risk value or the high percentile risk value in the parameter cloud as the final indicator, it no longer focuses on the average performance, but focuses on the "worst possible situation" under real production fluctuations. This evaluation method is more in line with the core needs of "safety first, stability first" in industrial production. Therefore, the process parameters optimized for this purpose will naturally have stronger anti-interference ability and higher production reliability, which can effectively reduce the risk of batch failure.

[0048] Further, the robustness risk indicator is inputted into the fitness function of the genetic algorithm, specifically, a composite fitness function is constructed, which takes the robustness risk indicator as a negative weighted item and combines it with a positive weighted item based on the expected product performance indicator of the candidate synthesis parameters; the mutation operator of the genetic algorithm is guided, specifically, the key risk parameters that have positive contribution to the overall failure risk are determined by using the explainability analysis results, and when performing mutation operation, the probability of the key risk parameters being selected for mutation is increased, and a directional mutation bias that reduces the key risk parameters is applied.

[0049] The specific form of the composite fitness function F is: . Wherein, and are the values after the expected product performance indicator and the robustness risk indicator are mapped to the [0, 1] interval by the maximum-minimum normalization method; and are preset weight coefficients representing the importance of performance and robustness respectively, and the sum of the two is 1 (for example, , ).

[0050] The specific operation of guiding the mutation operator of the genetic algorithm is: Using the explainability analysis results (such as SHAP value ), the mutation probability of each parameter is calculated , which can be set as a basic mutation probability plus an increment proportional to the absolute value of .

[0051] When a parameter determined as a key risk parameter (i.e. ) is selected for mutation, a directional mutation bias is applied to it, specifically, an adjustment amount proportional to the size of is subtracted from its current value. New value = original value , wherein is a preset learning rate or step factor.

[0052] The composite fitness function sets a clear and controllable optimization goal for the algorithm, allowing users to accurately balance the two sometimes conflicting indicators of high performance and high robustness according to actual needs. Secondly, the directional mutation operator driven by explainability analysis provides the most efficient path to achieve this goal. It can accurately identify and correct specific process parameters that cause risks, avoiding blind search. The synergistic effect of this goal + path ensures that the optimization process can quickly and stably converge to find the best process solution that truly balances performance and reliability.

[0053] Further, the process stability discriminator is an ensemble learning model composed of multiple base models: K independent risk assessment base models are trained through K-fold cross-validation on the augmented training set; when evaluating the failure risk of each perturbed parameter point in the parameter cloud, the parameter point is input into the K base models respectively to obtain K independent risk prediction values; the K independent risk prediction values are processed through an aggregation function to obtain a final, more robust failure risk value; the aggregation function is specifically the average and / or maximum of the K risk prediction values.

[0054] The use of an ensemble learning model significantly enhances the accuracy and stability of risk assessment. By aggregating the prediction results of multiple independently trained base models, the risk of misjudgment due to accidental bias of a single model or randomness of training data division can be effectively reduced, making the final robust evaluation result more reliable, thereby providing a more reliable decision basis for optimization.

[0055] Further, the aggregation processing is specifically selecting the maximum value and / or a pre-set high percentile value from the failure risk values of all perturbed parameter points in the parameter cloud as the robustness risk indicator to evaluate the worst possible performance under process fluctuations.

[0056] The aggregation processing is specifically selecting the maximum value from the failure risk values of all perturbed parameter points in the parameter cloud as the single robustness risk indicator; or selecting a pre-set high percentile value (e.g., the 95th percentile value) as the single robustness risk indicator; or selecting the larger value between the aforementioned calculated maximum value and high percentile value as the final single robustness risk indicator to evaluate the worst possible performance under process fluctuations.

[0057] By selecting the potential maximum risk value or high percentile risk value in the parameter cloud as the final indicator, this method no longer focuses on the average performance, but focuses on the "worst possible situation" under real production fluctuations. This evaluation method is more in line with the core needs of "safety first, stability first" in industrial production. Therefore, the process parameters optimized for this purpose will naturally have stronger anti-interference ability and higher production reliability, effectively reducing the risk of batch failure.

[0058] Further, the expected product performance indicator is calculated by a proxy model trained on the historical synthesis data for predicting product performance, which receives the candidate synthesis parameters as input and outputs performance prediction values including expected molecular weight and expected polydispersity index.

[0059] The proxy model is specifically a gradient boosting decision tree model.

[0060] The output multiple performance prediction values are combined into a single performance index by a preset utility function. × (normalized molecular weight) - × (normalized polydispersity index).

[0061] wherein, and are preset positive weights representing the importance of molecular weight and polydispersity, respectively; the polydispersity index term is negative because a lower index represents better performance. Before combination, each performance prediction value needs to be processed by maximum-minimum normalization or the like to eliminate dimensional differences.

[0062] The gradient boosting decision tree model used in this embodiment can accurately capture the complex nonlinear relationship between process parameters and product performance, ensuring the accuracy of performance prediction. Secondly, through an explicit weighted utility function, it combines multiple sometimes conflicting indicators such as molecular weight and polydispersity into a single quantifiable performance score.

[0063] The present application generates high-quality virtual failure data through a failure path deduction model, solves the problem of difficulty in constructing an accurate risk model due to sparse failure case data in industrial production, and provides reliable data support for production decision-making. By simulating process fluctuations in real production through a virtual disturbance unit and incorporating the quantified robustness risk into the optimization objective, the found process path not only has the best performance, but also has a wider process window to resist interference, thereby reducing the batch failure rate. By introducing explainability analysis to guide the mutation direction of the genetic algorithm, the optimization process is transformed from a "black box" search to a "white box" evolution with clear guidance, significantly accelerating the convergence speed to the robustness optimal solution and improving the decision-making efficiency.

[0064] Embodiment Two The application scenario of this embodiment is that an A enterprise develops a core raw material, polycaprolactone (PCL) polyol, for a special polyurethane (PU) adhesive required for electric vehicle (EV) battery package packaging for downstream new energy automobile customers.

[0065] The core requirement of this application scenario is: the adhesive must provide excellent bonding strength while having extremely high resistance to heat aging and electrolyte corrosion, to ensure the structural integrity and safety of the battery pack under long-term harsh working conditions. Therefore, the optimization goal is to maximize the robustness of the process path while ensuring bonding performance, minimizing the risk of batch material aging performance not meeting standards due to process fluctuations.

[0066] A enterprise uses its hundred-ton pilot production line to collect 180 batches of PCL polyol production data. These data are generated using the leading micro-reaction continuous flow process, including reaction temperature, residence time, monomer to initiator molar ratio, catalyst dosage, and other key synthesis parameters. Among them, 172 batches of products have been verified by downstream customers and have excellent performance (labeled as "success"). But 8 batches have a bonding strength decay exceeding the acceptable threshold after accelerated aging test, marked as "aging failure".

[0067] Due to the severe shortage of real data samples (8 batches) of "aging failure", first use the failure path reasoning model for data enhancement. Use all 180 batches of data to train a failure path reasoning model based on the U-Net architecture. After training, input the "aging failure" label to the model to generate 500 sets of virtual synthesis parameter data that may cause this failure mode.

[0068] Merge the 172 batches of successful data (label 0), 8 batches of real failure data (label 1), and 500 batches of virtual failure data (label 1) to train a LightGBM classifier. This model is used to predict the probability of "aging failure" of the final product for any set of synthesis parameters.

[0069] Only use 172 batches of successful data to train a gradient boosting regression tree model. This model is used to input a set of synthesis parameters to quickly predict the corresponding core product performance indicator - initial peel strength (N / mm).

[0070] Set the population size of the genetic algorithm to 120 and the number of iterations to 250 generations. Each individual represents a candidate PCL polyol synthesis path. The evaluation (calculate fitness) process for each individual is as follows: Input the synthesis parameters of the individual into the trained performance proxy model to get the predicted "initial peel strength" value, which is directly used as the single performance indicator P.

[0071] For each parameter of this individual, perturb around its central value to generate a parameter cloud containing N=100 points. The perturbation range is set taking into full consideration the "precise control" feature of A enterprise's micro-reaction continuous flow process, setting a narrower process tolerance than traditional tank reactors (for example, temperature fluctuation range is ±0.5°C).

[0072] The 100 points in the parameter cloud are input into the process stability discriminator one by one, and 100 corresponding "aging failure resistance" risk probability values are obtained.

[0073] Considering the extremely important battery safety, the maximum value in the 100 risk values is selected as the robustness risk indicator R of the individual, representing the risk performance of the process path under the worst fluctuation.

[0074] A composite fitness function is constructed. For this safety-critical application, the robustness risk term is given a very high weight to ensure that the optimization direction converges preferentially to "safest": .

[0075] wherein, and are the normalized values of the performance indicator and the risk indicator, respectively.

[0076] In the mutation stage of the genetic algorithm, for a selected individual to be mutated, first apply the SHAP analysis method to the process stability discriminator to quantify the contribution of each parameter to the "aging failure resistance" risk .

[0077] For example, the analysis found that the "catalyst dosage" value was significantly positive, which was the main factor leading to potential aging risk. Therefore, the probability of being selected for mutation of this parameter is significantly increased, and the mutation operation performed on it will be directed and small. The dosage is reduced.

[0078] After 250 generations of iteration, the algorithm converges to an optimal synthesis path. This result is compared with a path found by a traditional optimization method that only maximizes the "initial peel strength". As shown in Table 1: Table 1 Comparison of the present invention and the traditional optimization method

[0079] The traditional optimization method found a path that achieved the highest initial adhesive strength, but its robustness risk was as high as 25%, meaning that in the actual continuous flow production of A Company, there is a high risk of batch quality failure even with slight process fluctuations.

[0080] The path found by the present invention has an initial peel strength that is only slightly reduced by about 4.6%, which is completely within the customer's acceptable range, but its robustness risk is reduced by 92%, almost zero. This indicates that the process path has a very wide process window and very high production stability.

[0081] Example Three ​The application scenario of this embodiment is that enterprise A develops a long-acting protective coating for wind turbine blades in marine environment for its downstream high-end coating customers. The coating is prepared from polyurethane dispersion based on the special polycaprolactone polyol produced by enterprise A.

[0082] The blade coating needs to meet two core requirements: one is extreme weather resistance, which can maintain gloss and mechanical properties for a long time under strong ultraviolet radiation; the other is excellent salt spray corrosion resistance to resist the erosion of marine environment. The key failure mode is “coating salt spray blistering / peeling”, which is difficult to detect in the laboratory research and development stage, but is disastrous in actual application. Therefore, the optimization goal is to maximize the robustness of the process path while ensuring excellent weather resistance, and to minimize the risk of coating corrosion caused by process fluctuations.

[0083] The research and development team has accumulated historical data by synthesizing and evaluating 250 different formulations of PCL polyols in their advanced detection and analysis room. These data include synthesis parameters (reaction temperature, residence time, monomer / initiator molar ratio, catalyst dosage) and performance test results of PUD coatings prepared therefrom. Among them, 238 formulations of coatings show excellent performance. However, 12 formulations show slight “coating blistering” phenomenon in more than 1000 hours of accelerated salt spray test.

[0084] Since the real failure data sample (12) of “coating blistering” is too small to effectively train the model, a failure path deduction model is first used for data augmentation. All 250 sets of data are used to train a failure path deduction model based on U-Net architecture. After training, the model is input with the label of “coating blistering” to generate 600 sets of virtual synthesis parameter data that may cause this failure mode.

[0085] Merge the 238 sets of successful data (label 0), 12 sets of real failure data (label 1) and 600 sets of virtual failure data (label 1) to train a random forest (Random Forest) classifier. This model is used to predict the risk probability of “coating blistering” of the final coating for any set of synthesis parameters.

[0086] A gradient boosting regression tree model is trained using 238 sets of successful data. This model is used to input a set of synthesis parameters to quickly predict its core weather resistance indicator, gloss retention rate (%) after 2000 hours of QUV accelerated aging.

[0087] The population size of the genetic algorithm is set to 150, and the number of iterations is set to 300 generations. Each individual represents a candidate PCL polyol synthesis path. The evaluation (calculate fitness) process of each individual is as follows: The synthetic parameters of the individual are input into the trained performance proxy model to obtain a predicted "gloss retention" value, which is directly used as a single performance indicator P.

[0088] For each parameter of the individual, a perturbation is performed around its central value to generate a parameter cloud containing N = 100 points. The perturbation range is set according to the accuracy of the A-lab system. The 100 points in the parameter cloud are input one by one into the process stability discriminator to obtain 100 corresponding "coating blistering" risk probability values. For wind turbine blades, which require ultra-long service life and high reliability, the maximum value of the 100 risk values is selected as the robustness risk indicator R of the individual, which is used to evaluate the corrosion resistance stability in the worst case.

[0089] A composite fitness function is constructed. For this application, weather resistance (performance) and corrosion resistance (risk) are equally important, so a balanced weight is adopted: . wherein, and are the normalized values of the performance indicator and the risk indicator, respectively.

[0090] In the mutation stage of the genetic algorithm, for a selected individual to be mutated, the SHAP analysis method is first applied to the process stability discriminator to quantify the contribution of each parameter to the "coating blistering" risk .

[0091] For example, analysis finds that "monomer to initiator molar ratio" is too high, which is the main factor leading to potential blistering risk. Then in mutation, the probability of this parameter being selected is significantly increased, and the mutation operation performed on it will be directed to reduce its value.

[0092] After 300 generations of iteration, the algorithm converges to an optimal synthesis path. This result is compared with the path found by a traditional optimization method that only maximizes "gloss retention", as shown in Table 2: Table 2 Comparison of the present invention and the traditional optimization method

[0093] The traditional optimization method finds a formula with extremely high weather resistance indicator, but its potential salt spray blistering risk is also relatively high (22%), which is unacceptable for marine wind power coatings that require more than 20 years of life.

[0094] The path found by the present invention has a slightly lower gloss retention (about 2.6%), but its robustness risk is reduced by more than 95%, reaching an extremely high reliability level. This indicates that the process path has a very wide process window, which can effectively resist production fluctuations and ensure that each batch of products has excellent and stable corrosion resistance.

[0095] While embodiments of the application have been shown and described, it is to be understood that the application is not limited to the details of the embodiments described, since numerous changes, modifications, substitutions and variations can be made thereto without departing from the spirit and scope of the application as defined by the appended claims and their equivalents.

Claims

1. A method for optimizing the synthesis path of polycaprolactone polyols based on genetic algorithm, characterized in that: include: a genetic algorithm as a generator and a process stability discriminator; A failure path deduction model is trained using historical synthesis data to learn the conditional probability distribution of synthesis parameters consisting of reaction temperature, reaction time, monomer-to-initiator molar ratio, and catalyst-to-initiator molar ratio under given failure mode labels; and virtual failure data is generated in a targeted manner by inputting failure mode labels into the failure path deduction model. For candidate synthetic parameters generated by the genetic algorithm, a parameter cloud is generated within the neighborhood of the candidate synthetic parameters using a virtual perturbation unit; the failure risk value of the perturbed parameter point in the parameter cloud is evaluated using the process stability discriminator to construct a strategy volatility risk matrix; and performing aggregation processing on the volatility risk matrix of the strategy to generate a robustness risk indicator representing the overall failure risk of the candidate synthetic parameters; The robustness risk index is input into the fitness function of the genetic algorithm, and the interpretability analysis result of the process stability discriminator is used to guide the mutation operator of the genetic algorithm to accelerate the optimization process to converge to the robust optimal solution.

2. The method for optimizing the synthesis path of polycaprolactone polyols based on genetic algorithm according to claim 1, wherein The training steps of the process stability discriminator include: Merging the historical synthetic data with the virtual failure data to form an enhanced training set; The process stability discriminator is trained in a supervised learning manner using the enhanced training set, so that the discriminator can input a set of synthesis parameters consisting of reaction temperature, reaction time, monomer-to-initiator molar ratio, and catalyst-to-initiator molar ratio, and output the corresponding failure risk probability.

3. The method for optimizing the synthesis path of polycaprolactone polyols based on genetic algorithm according to claim 1, wherein: The interpretability analysis result is obtained by applying the SHAP analysis method to the process stability discriminator to quantify the positive and / or negative contribution of each parameter in each candidate synthesis parameter to the overall failure risk; the mutation operator, specifically, dynamically adjusts the mutation probability and mutation direction of the corresponding parameter gene according to the contribution, applies reverse mutation to the parameters that produce positive risk contribution, and increases the mutation probability of the parameters.

4. The method for optimizing the synthesis path of polycaprolactone polyols based on genetic algorithm according to claim 1, wherein: The failure path deduction model includes: a forward diffusion unit that gradually adds Gaussian noise to the real synthetic parameters in the historical synthetic data through a Markov chain until the distribution becomes a standard normal distribution; a conditional denoising unit, which is a trained neural network, receiving the noisy composite parameter and the failure mode label at any time step of the forward diffusion process as common inputs, and predicting the noise added to the noisy composite parameter; and a sampling generation unit for sampling an initial noise from a standard normal distribution, and under the guidance of the failure mode label, iteratively calling the conditional denoising unit to perform step-by-step denoising and reversely reconstruct the virtual failure data.

5. The method for optimizing the synthesis path of polycaprolactone polyols based on genetic algorithm according to claim 1, wherein The virtual disturbance unit is specifically a parameterized noise generation module: For each of the candidate synthesis parameters, multiple random samplings are independently performed from a probability distribution according to a process tolerance range that matches the physical characteristics of the parameter to generate the parameter cloud, wherein the process tolerance range is set based on statistical fluctuations of historical production data and / or expert experience.

6. The method for optimizing the synthesis path of polycaprolactone polyols based on genetic algorithm according to claim 1, wherein Generating a robustness risk indicator representing the overall failure risk of the candidate synthesis parameters includes: A single candidate synthetic parameter generated by the genetic algorithm is received; the virtual perturbation unit generates a parameter cloud containing multiple perturbed parameter points around the single candidate synthetic parameter through multiple random samplings; the failure risk values ​​corresponding to the perturbed parameter points in the parameter cloud are constructed into a strategy volatility risk matrix; and the strategy volatility risk matrix is ​​aggregated to generate a single robustness risk indicator representing the overall failure risk of the candidate synthetic parameter.

7. The method for optimizing the synthesis path of polycaprolactone polyols based on genetic algorithm according to claim 6, wherein: The robustness risk indicator is input into the fitness function of the genetic algorithm, specifically by constructing a composite fitness function, which uses the robustness risk indicator as a negatively weighted term and combines it with a positively weighted term of the expected product performance indicator based on the candidate synthetic parameter; the mutation operator of the genetic algorithm is guided, specifically by using the interpretability analysis results to determine the key risk parameters that have a positive contribution to the overall failure risk, and when performing the mutation operation, the probability of the key risk parameter being selected for mutation is increased, and a directional mutation bias is applied to reduce the key risk parameter.

8. The method for optimizing the synthesis path of polycaprolactone polyols based on genetic algorithm according to claim 6, wherein: The aggregation processing is specifically to select the maximum value and / or a preset high percentile value from the failure risk values ​​of all the perturbed parameter points in the parameter cloud as the robustness risk indicator to evaluate the worst possible performance under process fluctuations.

9. The method for optimizing the synthesis path of polycaprolactone polyols based on genetic algorithm according to claim 7, wherein: The expected product performance indicators are calculated by a proxy model trained on the historical synthesis data for predicting product performance. The proxy model receives the candidate synthesis parameters as input and outputs performance prediction values ​​including expected molecular weight and expected polydispersity index.

Images