Method for optimizing an organic, environmentally friendly water-based extinguishing agent

Abstract

This invention relates to an optimization method for an organic, environmentally friendly water-based fire extinguishing agent, belonging to the field of intelligent optimization technology for fire extinguishing agent formulations. The method parametrically models the fire extinguishing agent formulation and preparation process to construct a joint feasible domain; it establishes a comprehensive performance function including extinguishing time, cooling capacity, smoke suppression index, and reignition probability, and simultaneously constructs a function including TOC, COD, and EC. 50 An environmental risk function is derived based on VOC, organic fluorine content, and bio-based proportion, with environmental thresholds as constraints. Mechanism consistency constraints such as surface tension, viscosity window, and lower bound of cooling capacity are introduced. Through worst-case robust optimization and an active sampling closed-loop update mechanism, stable performance optimization under low-temperature and brine environments is achieved. This method can obtain optimal fire extinguishing agent formulations and process parameters that combine environmental compliance, smoke suppression and reignition resistance, and multi-condition adaptability, while meeting the requirements of fluorine-free and high bio-based proportions.

Description

Technical Field

[0001] This invention relates to the field of fire extinguishing agent formulation development and optimization, and more particularly to the interdisciplinary technical field of intelligent formulation optimization, green chemistry and fire protection materials engineering. Background Technology

[0002] Existing water-based fire extinguishing agents typically rely on empirical formulations and a small number of orthogonal experimental iterations, which have the following shortcomings: 1) Inconsistent objective functions: indicators such as extinguishing time, cooling capacity, foam stability, smoke suppression, and anti-reignition are evaluated separately, making it difficult to balance and optimize them within the same framework;

[0003] 2) Environmental and safety constraints lack calculable expression: They are often described using descriptive language such as "environmentally friendly, low toxicity, low VOC", which makes it difficult to directly couple with experimental data and regulatory thresholds to form verifiable constraints;

[0004] 3) Insufficient compatibility with low temperature and brine conditions: Under low temperature (even below 0℃) or brine conditions, the formula is prone to phase separation, viscosity drift, and decreased foaming, which leads to a reduction in fire extinguishing efficiency.

[0005] 4) High cost and poor reproducibility of small sample R&D: The formulation-process variables are highly dimensional, and blind experimentation leads to high costs and difficulty in reproducing the "optimal process window".

[0006] Therefore, there is an urgent need for an optimization method that can unify performance, environmental compliance, low temperature / salt water robustness, smoke suppression and reignition resistance into a calculable closed loop, in order to reduce the number of tests and improve the feasibility of engineering implementation. Summary of the Invention

[0007] The purpose of this invention is to provide an optimization method for organic and environmentally friendly water-based fire extinguishing agents. By embedding the comprehensive performance function, environmental risk function, and mechanism consistency constraint into the worst-case robust optimization and active sampling closed loop, the method achieves stable performance improvement under multiple operating conditions such as low temperature and salt water, while ensuring environmental compliance and engineering feasibility. It significantly improves optimization efficiency and result reliability within a limited experimental budget.

[0008] To achieve the above objectives, the present invention provides an optimization method for an organic environmentally friendly water-based fire extinguishing agent, which is executed by a processor and includes:

[0009] S1 Formulation and Process Modeling: Representing the extinguishing agent formulation as a mass fraction vector ,in Let i be the mass fraction of the i-th component, satisfying , , The process parameters are expressed as follows: and construct a joint feasible domain. ;

[0010] S2 performance and efficiency function modeling: The performance vector is obtained from the input. And calculate the overall performance function to be maximized:

[0011]

[0012] in , As the indicator weight, For the synergistic weighting of smoke suppression and reignition prevention, This is the normalization function; S3 compliance risk constraints; To calculate the environmental risk function.

[0013] and limit Each indicator was entered after standardization. Let i be the weights, i=1,…,6; TOC is total organic carbon, COD is chemical oxygen demand. These are acute toxicity indicators; TOC is total organic carbon, and COD is chemical oxygen demand. VOCs are acute toxicity indicators. This refers to the organic fluorine content. The proportion of bio-based components;

[0014] S4 Mechanism Feasibility Constraints: Establishing a set of mechanism constraint functions and the and Commonly constrained feasible solutions For the threshold;

[0015] S5 Robust Optimization Solution: With temperature and salinity perturbations... Below, the predictive performance of the surrogate model output. With prediction uncertainty Construct a robust objective and solve it.

[0016] in The penalty coefficient is... S6 is a function that increases with increasing uncertainty; closed-loop iterative update: selecting from the feasible region based on the acquired function. Conduct experiments to obtain real results After updating the proxy model parameters with the experimental data, the process returns to execution S5 until the termination condition is met, outputting the optimal result that satisfies robustness and compliance constraints. .

[0017] Compared with the prior art, the present invention has at least the following beneficial effects:

[0018] 1) Improve robust stability under multiple operating conditions: By constructing a worst-case robust optimization structure: This achieves the effect of ensuring the lower limit of performance under low temperature and salt water disturbance conditions, avoiding the problem of insufficient environmental adaptability caused by existing technologies that only optimize a single operating condition.

[0019] 2) Achieving pre-embedded environmental compliance: by using environmental risk functions:

[0020]

[0021] By directly embedding optimization constraints, regulatory requirements are ensured to be met during the optimization phase, thus avoiding the waste of repeated experiments in the existing technology of "optimization followed by screening".

[0022] 3) Enhancing mechanistic consistency and physical feasibility: By constructing mechanistic constraint functions: This ensures that physical mechanisms such as wetting, atomization, and cooling are effective, avoiding situations where numerical values ​​are optimal but engineering implementation is impossible.

[0023] 4) Reduce the number of experiments and R&D costs: By introducing an acquisition function: This achieves the effect of prioritizing experimental sites with high potential and high compliance probability, significantly reducing the number of experimental rounds compared to randomized trials or response surface methodology.

[0024] 5) Improve prediction stability and engineering reliability: by introducing an uncertainty penalty term:

[0025]

[0026] This achieves the effect of suppressing "spurious optimality" in areas of high predictability uncertainty, thereby improving the reproducibility and stability of the final formulation.

[0027] 6) Achieve triple coupling optimization of performance, environmental protection, and mechanism: through unified construction:

[0028] This forms a closed-loop structure of variables, performance, risk, constraints, robustness, and experimental updates, thereby achieving multi-dimensional performance synergistic optimization. Attached Figure Description

[0029] Figure 1 This is a flowchart of the optimized method for the organic environmentally friendly water-based fire extinguishing agent provided by the present invention. Detailed Implementation

[0030] To enable those skilled in the art to better understand the technical solutions of the present invention, the clock loop forming processing method and apparatus provided by the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0031] The invention will be described more fully below with reference to the accompanying drawings; however, the embodiments shown may be embodied in different forms, and the invention should not be construed as limited to the embodiments set forth below. Rather, these embodiments are provided so that the invention will be thorough and complete, and that those skilled in the art will fully understand the scope of the invention.

[0032] The accompanying drawings of the embodiments of the present invention are provided to further illustrate the embodiments of the present invention and form part of the specification. They are used together with the detailed embodiments to explain the present invention and do not constitute a limitation thereof. The above and other features and advantages will become more apparent to those skilled in the art from the description of the detailed embodiments with reference to the accompanying drawings.

[0033] The present invention is not limited to the embodiments shown in the accompanying drawings, which illustrate illustrative properties but are not intended to be limiting.

[0034] Figure 1 This is a flowchart of the optimized method for the organic environmentally friendly water-based fire extinguishing agent provided by the present invention, as shown below. Figure 1 As shown, the optimization method for organic environmentally friendly water-based fire extinguishing agents provided by the present invention is executed by a processor and is aimed at water-based fire extinguishing agent formulations composed of an aqueous phase and at least one organic component. The method includes:

[0035] S1 Formulation and Process Modeling: Representing the extinguishing agent formulation as a mass fraction vector The process parameters are expressed as and construct a joint feasible domain. Specifically including

[0036] S1.1 Define the raw material / component set and perform variable mapping, specifically including:

[0037] S1.1.1 Obtain a list of candidate components for the organic environmentally friendly water-based fire extinguishing agent to be optimized. The candidate component list includes at least an aqueous phase component and an organic component. The organic component can be functionally classified into at least two categories: solubilizing aids, surfactants, film-forming / barrier agents, smoke suppressants, and flame retardants / anti-reignition agents. The solubilizing aid is selected from at least one of polyols, organic alcohols, or organic salts; the surfactant is selected from at least one of anionic, nonionic, amphoteric, or silicone surfactants; the film-forming agent is selected from at least one of cellulose derivatives, polyvinyl alcohol, acrylic polymers, or chitosan; the smoke suppressant is selected from at least one of aluminum hydroxide, magnesium hydroxide, phosphates, or melamine; and the flame retardant is selected from at least one of phosphorus-containing flame retardants, nitrogen-containing flame retardants, or phosphorus-nitrogen synergistic flame retardant systems.

[0038] S1.1.2 represents each component in the candidate component list. Establish the formula decision variables and represent the formula as a quality score vector:

[0039]

[0040] in, The total number of components included in the optimization; For the first The mass fraction (or the normalized percentage of mass fractions) of each component in the final formulation.

[0041] S1.1.3 Establish a process decision variable vector for the preparation process. The process decision variables include at least one of the following: rotation speed, dispersion time, temperature, pH, shear rate, and feeding sequence coding:

[0042]

[0043] in, The number of variables related to art; For the first Process variables (such as stirring speed, dispersion time, temperature, pH, shear rate, etc.).

[0044] This invention maps "component selection / proportioning" to Mapping "preparation process" to This transforms the formulation development problem into a computable joint optimization variable space, facilitating subsequent optimization of performance and compliance within the same mathematical framework.

[0045] S1.2 Constructing the basic feasibility constraints of the formulation, specifically including:

[0046] Apply nonnegativity and normalization constraints to the formulation mass fraction; apply upper and lower bound constraints to each component (sources may be solubility, regulatory restrictions, stability boundaries, or process manufacturability boundaries); when using "functional grouping constraints," divide the component set into... Functional Classes And set the percentage variable within the group. ,

[0047] By establishing non-negative, normalized, upper and lower bounds, and functional grouping constraints, the system can automatically exclude incompatible, unmanufacturable, or obviously non-compliant formulation regions during the search process, thereby improving optimization efficiency and enhancing the feasibility of the solution.

[0048] S1.3 Construct the feasible region of process variables and encode the discrete process parameters:

[0049] S1.3.1 For continuous process variables (For parameters such as rotation speed, temperature, pH, and dispersion time) give the allowable range:

[0050]

[0051] in, A set of continuous process variable indices; These correspond to the upper and lower limits of process variables, respectively.

[0052] S1.3.2 When the feeding sequence is used as a discrete process variable, define the feeding sequence code. : ,

[0053]

[0054] And the process variables are expressed in a mixed form, where, This represents the number of optional feeding sequence schemes; This is a continuous process variable subvector.

[0055] This invention achieves the goal of endogenizing "process feasibility" into the optimization space by establishing codes for discrete processes such as the manufacturable range for continuous processes and the feeding sequence, thereby reducing performance fluctuations caused by the non-reproducibility of processes.

[0056] S1.4 Forming a joint feasible region Specifically, this includes: combining formulation variables and process variables to form decision pairs. and define the joint feasible region. :

[0057] .

[0058] This invention constructs a joint feasible domain. This achieves the goal of unifying the constraints of formulation feasibility and process feasibility, ensuring that subsequent optimizations are carried out under the boundary conditions of "preparable, feasible, and reproducible" throughout the entire process.

[0059] By parameterizing the component mass fraction and the process window respectively. and This allows the formulation development problem to be standardized into a computable joint optimization problem;

[0060] S2: Performance and efficiency function modeling, The performance vector is obtained from the input. And calculate the overall performance function to be maximized. Specifically, it includes:

[0061] S2.1.1 Receive the recipe variables output from step S1 With process variables And construct performance index vectors based on experimental measurement data or prediction models. :

[0062]

[0063] in, This refers to the time required for fire extinguishing; This refers to the cooling capacity per unit time or the integral of temperature drop. As an indicator of foam stability; Surface tension; Dynamic viscosity; The smoking suppression index; This represents the probability of reignition. Efficiency decreases under saline conditions; This is due to performance degradation at low temperatures.

[0064] Among them, the smoke suppression index for:

[0065]

[0066] in, The reference light transmittance; For a moment The light transmittance; This is the measurement time window.

[0067] The probability of reignition for:

[0068]

[0069] in, For the Sigmoid function; The residual surface temperature after fire extinguishing; The mass of residual combustible material; This refers to the time of water loss. , where i = 1, 2, 3.

[0070] The salt water robustness for:

[0071]

[0072] in, Salt concentration; Preset brine operating conditions; This is the overall performance function.

[0073] The low-temperature robustness term for:

[0074]

[0075] in, At room temperature; It is at a low temperature.

[0076] This invention achieves the effect of simultaneously optimizing the performance of multiple operating conditions within a single variable space by unifying fire extinguishing time, cooling capacity, smoke suppression, anti-reignition, salt water resistance, and low-temperature robustness into a performance vector.

[0077] S2.2 Constructing the comprehensive performance objective function, specifically including: normalizing and mapping each indicator to construct the comprehensive performance function:

[0078] ,

[0079] in, For a vector of formulation or structural parameters, For control parameter vectors; For the first One performance evaluation metric The total number of indicators; The corresponding indicator weights; For index normalization function; For smoke suppression performance indicators; This represents the probability of reignition. For the synergistic weighting of smoke suppression and anti-reignition

[0080] in, For a vector of formulation or structural parameters, For control parameter vectors; For the first One performance evaluation metric The total number of indicators; The corresponding indicator weights; For index normalization function; For smoke suppression performance indicators; This represents the probability of reignition. The weighting is for the synergistic effect of smoke suppression and re-ignition prevention.

[0081] This invention achieves the effect of avoiding extreme optimization of a single indicator and improving overall safety performance by constructing a comprehensive objective function and introducing synergistic terms for smoke suppression and anti-reignition.

[0082] S3: Compliance risk constraints, with To calculate the environmental risk function. :

[0083] ,

[0084] Among them, organic fluorine items for: Bio-based items for: OC stands for Total Organic Carbon, and COD stands for Chemical Oxygen Demand. For acute toxicity indicators, VOC refers to volatile organic compounds; risk constraints: , The threshold value is used.

[0085] The organic fluorine content term in the environmental risk function satisfies:

[0086]

[0087] in, For the first The mass fraction of organic fluorine in the components. The threshold value is used.

[0088] 9. The method according to claim 1, characterized in that: the bio-based proportion term in the environmental risk function satisfies:

[0089]

[0090] in The contribution coefficient of bio-based This is the lower bound threshold.

[0091] This invention incorporates TOC, COD, EC50, VOC, organic fluorine, and bio-based content into a risk function, thereby transforming environmental compliance from a descriptive concept into a calculable constraint and improving patent enforceability and examination resistance.

[0092] This invention achieves the effect of unified modeling of low temperature, brine, smoke suppression, and anti-reignition by constructing a unified performance index vector; avoids the optimization bias of a single performance index by constructing a comprehensive performance function; and automatically meets environmental compliance requirements by constructing an environmental risk function and applying threshold constraints.

[0093] S4: Mechanism Feasibility Constraints: Establishing a set of mechanism constraint functions and the and Commonly constrained feasible solutions Threshold; Mechanism constraint function It must include at least the surface tension deviation function: ,in, The surface tension of the extinguishing agent under a given formulation and process; : Wetting critical threshold. The mechanism constraint function. It also includes window functions. and the lower bound function of cooling capacity :

[0094] ,in, In a given formula With process Dynamic viscosity measured or predicted under certain conditions; Indicates the upper limit of viscosity; This indicates the lower limit of viscosity;

[0095] ,in, This refers to the temperature drop or heat absorption capacity per unit time. This represents the minimum cooling capacity.

[0096] S5: Robust optimization solution: in the case of temperature and salinity perturbations Below, the predictive performance of the surrogate model output. With prediction uncertainty Construct a robust objective and solve it. Specifically, it includes:

[0097] S5.1 Input data and variables are consistent, specifically including:

[0098] S5.1.1 Receive the joint feasible region output from step S1 and decision variables ,in Let the mass fraction of the formula be a vector. This is a vector of process variables.

[0099] S5.1.2 Receive the overall performance objective function output from step S2 With environmental risk function and the family of mechanism constraint functions output in step S3 The mechanism constraint function is composed of the performance vector. Calculated.

[0100] S5.2 defines the set of disturbances and the worst-case operating conditions, specifically including:

[0101] S5.2.1 Define the disturbance variable: And define the perturbation set:

[0102] in, For perturbation state variables; Ambient temperature; Salinity (mass fraction or concentration); This is the lowest design temperature; For reference temperature; Maximum salinity; This is a set of temperature-salinity uncertainties.

[0103] S5.2.2 For any candidate solution Define worst-case prediction performance:

[0104]

[0105] in, For the proxy model in working conditions The overall effectiveness of forecasting; This represents the lower bound of the prediction performance under worst-case conditions. The surrogate model can be one of the following: Gaussian process model, BP neural network, deep neural network, graph neural network, support vector regression (SVR), or multinomial response surface model (RSM).

[0106] This invention uses The worst-case scenario is defined to ensure that the optimization results have robust stability in low-temperature and saltwater environments.

[0107] S5.3 Output the surrogate model and uncertainties, specifically including:

[0108] S5.3.1 Based on existing experimental datasets

[0109]

[0110] Training agent model This enables it to output predicted values ​​and prediction uncertainties for candidate solutions:

[0111]

[0112] S5.3.2 By And based on perturbation variables Obtained through the index correction function Then substitute it into the comprehensive performance function get , and by Calculate the uncertainty penalty function:

[0113]

[0114] in, For the first Experimental dataset for round-by-round iteration; For the first The performance vector obtained from this experiment (including each index); For the proxy model, These are model parameters; For predicting performance vectors; For the prediction uncertainty vector; For its first One component; This is an uncertainty penalty function used to suppress "spurious optima" in regions of high uncertainty.

[0115] This invention achieves the effect of reducing overfitting and improving the reproducibility of solutions under small sample conditions by outputting predicted values ​​and uncertainties and introducing a penalty term.

[0116] S5.4 Robust optimization for solving joint constraints, specifically including:

[0117] S5.4.1 In the set of temperature and salinity perturbations Below, the predictive performance of the surrogate model output. With prediction uncertainty Construct a robust objective and solve it.

[0118]

[0119] in The penalty coefficient is... Let be a function that increases with increasing uncertainty, and satisfy:

[0120] , ,

[0121] in, Decision variables selected from the feasible region; This is a predicted value for environmental risks. Environmental threshold; For the first A mechanism consistency constraint function; Uncertainty penalty coefficient; : Returns the decision variable that maximizes the objective function.

[0122] According to one embodiment of the present invention, the process of training, for example, a deep ensemble model into a proxy model includes: constructing a training dataset: ,in: Let i be the formula variable for the i-th experiment; Let i be the process variable for the i-th experiment; Let i be the environmental disturbance variable in the i-th experiment; For true comprehensive effectiveness;

[0123] Constructing the network input vector: ,Right now:

[0124] Input z into the l-th layer of the m-th sub-model of the deep ensemble model:

[0125] , :

[0126] Output layer output: ,in, Let m be the weight matrix of the l-th layer of the m-th model; The bias of the l-th layer in the m-th model; The number of network layers; The activation function is (ReLU / Tanh). Represented as a hidden layer;

[0127] Each sub-model is trained independently, and its loss function is:

[0128]

[0129] in, True overall effectiveness; To predict overall effectiveness; The regularization coefficient is used. This represents the complete set of trainable parameters for the neural network proxy model, including all weights and biases.

[0130] Update via gradient descent:

[0131]

[0132] in: The learning rate is used until training reaches convergence:

[0133] , This is the set value.

[0134] The predicted overall performance is: ,

[0135] The prediction uncertainty is: .

[0136] This invention achieves simultaneous optimization of both "robust performance lower bound" and "predictive reliability" by placing the worst-case objective and uncertainty penalty within the same set of parentheses; through... and As a joint constraint, the output solution naturally satisfies both environmental compliance and physical mechanism feasibility; through The internal solution ensures that the solution is manufacturable and the process is executable.

[0137] By performance vector The calculated mechanism constraint function Environmental risk constraints Simultaneously, a worst-case robust optimization objective is embedded, and an uncertainty penalty term is introduced. Thus, the solution is obtained. It has a higher performance lower bound and higher prediction reliability under multiple operating conditions, and forms a reproducible closed-loop optimization process under limited experimental budget.

[0138] Top of form

[0139] bottom of form

[0140] S6 Closed-loop iterative update: Selecting from the feasible region based on the acquisition function. Conduct experiments to obtain real results After updating the proxy model parameters with the experimental data, the process returns to execution S5 until the termination condition is met, outputting the optimal result that satisfies robustness and compliance constraints. Specifically including

[0141] S6.1 Predicted target value based on the output of step S4 Predicted risk value Prediction uncertainty Based on the mechanistic constraint prediction results, a multi-factor joint acquisition function is constructed:

[0142] ,

[0143] in, This indicates a desire for improvement; This represents the overall performance value predicted by the proxy model. This indicates that the mechanism constraint satisfies the probability product; Indicates the first A mechanism constraint function; This indicates the probability that the constraint is satisfied, as given by the prediction model. : Predicted environmental risk function value; This indicates the permissible threshold for environmental risks; Indicates environmental margin; For the lower bound of robust performance, This indicates a robust performance metric.

[0144] The expected improvement function is defined as:

[0145] in, This is the best true value observed so far.

[0146] This invention achieves the effect of prioritizing experimental points that are "high-performance, highly reliable, and highly compliant" by incorporating improvement potential, mechanism feasibility probability, environmental margin, and robustness conditions into the acquisition function.

[0147] S6.2 Selecting active sampling points specifically includes:

[0148] S6.2.1 In the joint feasible region Internal solution to obtain candidate points:

[0149]

[0150] S6.2.2 The candidate points shall be used as the experimental formula and process scheme for the next round.

[0151] By maximizing the acquisition function within the feasible region, we can achieve the effect of maximizing information gain and performance improvement efficiency with a limited number of experiments.

[0152] S6.3 Experimental execution and real data acquisition, specifically including:

[0153] S6.3.1 According to the selected Preparation of fire extinguishing agent samples;

[0154] S6.3.2 Measure performance indicators under standardized experimental conditions:

[0155]

[0156] S6.3.3 Calculate the actual overall performance With risk :

[0157] , By replacing predicted values ​​with real experimental data, the effect of correcting model bias and improving prediction accuracy can be achieved.

[0158] S6.4 Model parameter updates, specifically including:

[0159] S5.4.1 Update the dataset:

[0160] S6.4.2 Retrain the agent model:

[0161] in, Regularization coefficient; Model parameters.

[0162] This invention achieves the effect of gradually reducing prediction error and uncertainty by updating model parameters based on real experimental data.

[0163] S6.5 Convergence determination mechanism, specifically including:

[0164] S5.5.1 Stop iteration when any of the following conditions are met: or ,in, To improve the performance threshold; To maximize the number of experiments and output the optimal solution that satisfies robustness and compliance constraints. .

[0165] This invention achieves the effect of prioritizing the exploration of high-potential and compliant regions by constructing a multi-factor joint acquisition function; it achieves the effect of gradually approaching the true optimal solution by updating the model through experimental data in a closed loop; and it achieves the effect of engineering optimization with limited experimental cost through convergence control.

[0166] experiment

[0167] I. Experimental Design:

[0168] 1. Repetition and Randomization: Each method is independently repeated M = 8 times; each time, a different random seed is used to initialize the initial samples; the initial sample set size... Total experimental budget .

[0169] 2. Set of disturbance conditions:

[0170]

[0171] Calculate worst-case performance: .

[0172] II. Definition of Statistical Indicators

[0173] For each method, record the following statistics: final worst-case performance. Reignition rate (%); Compliance pass rate (meets requirements) and );achieve Number of experimental rounds required

[0174] Calculate for each indicator:

[0175]

[0176] 95% confidence interval:

[0177] in: ;

[0178] III. Statistical Test Methods

[0179] 1. Normality test

[0180] The Shapiro-Wilk test is used to determine whether the distribution follows a normal distribution.

[0181] 2. Pairwise comparison

[0182] If normality is satisfied, a two-tailed independent samples t-test is used:

[0183] If normality is not satisfied, the Mann–Whitney U test is used.

[0184] Significance level setting:

[0185]

[0186] When the probability value in statistical detection The difference was determined to be significant at that time.

[0187] IV. Statistical Table of Experimental Results:

[0188] Table 1: Worst-case performance

[0189]

[0190] RSM stands for Response Surface Methodology.

[0191] Conclusion: A was significantly different from all comparison methods (p < 0.01).

[0192] Table 2: Reignition Rate (%)

[0193] Table 3: Number of rounds required to achieve target performance

[0194] Based on statistical analysis of eight independent and repeated experiments, the overall effectiveness of the method of this invention under worst-case conditions is demonstrated. The mean value was 0.812±0.021, significantly higher than that of the non-robust method (0.645±0.037), the random method (0.533±0.042), and the response surface methodology (0.612±0.029). All two-tailed t-tests were satisfied. The results indicate that the differences are statistically significant. Regarding the reignition rate, the method of this invention significantly reduces it to 2.5%, a significant difference compared to the comparative method (p<0.01). Furthermore, the method of this invention significantly reduces the number of experimental rounds required to reach the target performance threshold, demonstrating higher sample efficiency and optimization stability. These statistical results demonstrate that the method of this invention has significant robustness and engineering feasibility under low temperature and saltwater disturbance conditions.

[0195] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. An optimization method for an organic, environmentally friendly water-based fire extinguishing agent, characterized in that, Executed by the processor, including: S1: Formulation and Process Modeling: Representing the extinguishing agent formulation as a mass fraction vector ,in Let i be the mass fraction of the i-th component, satisfying , , The process parameters are expressed as follows: and construct a joint feasible domain S2 performance and efficiency function modeling: based on The performance vector is obtained from the input. And calculate the overall performance function to be maximized: in , As the indicator weight, For the synergistic weighting of smoke suppression and anti-reignition, This is the normalization function; S3: Compliance Risk Constraints: Based on To calculate the environmental risk function: and limit Each indicator was entered after standardization. Let i be the weights, i=1,…,6; TOC is total organic carbon, COD is chemical oxygen demand. COD is an acute toxicity indicator, representing chemical oxygen demand. VOCs are acute toxicity indicators. This refers to the organic fluorine content. The proportion of bio-based components; S4: Mechanism Feasibility Constraints: Establishing a set of mechanism constraint functions and the and Commonly constrained feasible solutions For the threshold; S5: Robust optimization solution: in the case of temperature and salinity perturbations Below, the predictive performance of the surrogate model output. With prediction uncertainty Construct a robust objective and solve for it: in, The penalty coefficient is... For uncertainty A function that increases with increasing size; S6: Closed-loop iterative update: Selecting from the feasible region based on the acquisition function. Conduct experiments to obtain real results After updating the proxy model parameters with the experimental data, the process returns to execution S5 until the termination condition is met, outputting the optimal result that satisfies robustness and compliance constraints. .

2. The method according to claim 1, characterized in that: The performance vector At least include the smoking suppression index ,and in As a reference light transmittance, For a moment Light transmittance, For the purpose of assessment duration.

3. The method according to claim 2, characterized in that: The performance vector It also includes the probability of reignition. : ， in, For the Sigmoid function, The residual surface temperature after fire extinguishing. For the quality or equivalent indicators of residual combustibles, This is an indicator of drying time.

4. The method according to claim 3, characterized in that: The comprehensive performance function is: ， in, For a vector of formulation or structural parameters, For the control parameter vector; For the first One performance evaluation metric The total number of indicators; The corresponding indicator weights; For index normalization function; For smoke suppression performance indicators; This represents the probability of reignition. The weighting is for the synergistic effect of smoke suppression and re-ignition prevention.

5. The method according to claim 4, characterized in that: The temperature and salinity perturbation variables And the worst-case set in the robust objective satisfies ， in, Indicates ambient temperature. To design the lowest temperature, For reference temperature; Indicates salt and, This is the preset maximum salinity.

6. The method according to claim 1, characterized in that: The mechanism constraint function It must include at least the surface tension deviation function: ,in, The surface tension of the extinguishing agent under a given formulation and process; : Wetting critical threshold.

7. The method according to claim 6, characterized in that: The mechanism constraint function It also includes window functions. and the lower bound function of cooling capacity : ,in, In a given formula With process Dynamic viscosity measured or predicted under certain conditions; This indicates the upper limit of viscosity; This indicates the lower limit of viscosity; ,in, This refers to the temperature drop or heat absorption capacity per unit time. This represents the minimum cooling capacity.

8. The method according to claim 1, characterized in that: The organic fluorine content term in the environmental risk function satisfies: in For the first The mass fraction of organic fluorine in the components. The threshold value is used.

9. The method according to claim 1, characterized in that: The bio-based proportion term in the environmental risk function satisfies: in The contribution coefficient of bio-based This is the lower bound threshold.

10. The method according to claim 1, characterized in that: The acquisition function is: ， in, This indicates a desire for improvement; This represents the overall performance value predicted by the proxy model. This indicates that the mechanism constraint satisfies the probability product; Indicates the first A mechanism constraint function; This indicates the probability that the constraint is satisfied, as given by the prediction model. : Predicted environmental risk function value; This indicates the permissible threshold for environmental risks; Indicates environmental margin; For the lower bound of robust performance, This indicates a robust performance metric.

Images