Method for extracting starch from wild acorn

Abstract

The invention relates to a method for extracting starch from wild acorn. A response surface method and a genetic algorithm are adopted to collaboratively optimize an acorn starch extraction process. The method comprises the following steps: firstly, carrying out an experiment through central composite design, establishing a secondary prediction model of soaking temperature, pH value, time, H2O2 concentration and solid-to-liquid ratio with starch whiteness, tannin residual quantity and purity, then taking an RSM model as a fitness function of GA, and carrying out a sensitivity-guided variation strategy and a dynamic weight adjustment mechanism so as to obtain a sensitivity-guided GA (Genetic Algorithm) model. And efficient optimization is carried out in the RSM-guided contraction search space. According to the scheme, double termination conditions are adopted, the optimal process parameters are finally obtained through experimental verification, other optimization algorithms do not need to be introduced, deep collaboration of RSM and GA is achieved, a suboptimal solution caused by premature termination is avoided, resource waste caused by excessive iteration is prevented, and the optimal balance of optimization efficiency and precision is achieved.

Description

Technical Field

[0001] This invention relates to the field of acorn starch extraction technology, specifically a method for extracting starch from wild acorns. Background Technology

[0002] Acorn starch extraction is a process that uses acorns as raw material and involves a series of physicochemical treatments to obtain edible starch. Acorns are rich in starch, but they also contain a large amount of tannins, pigments, and other anti-nutritional factors. Traditional extraction processes mainly include shell breaking and soaking, bleaching and purification, solid-liquid separation, and final fine drying to obtain acorn starch. Existing production processes largely rely on operator experience to set parameters, lacking systematic optimization methods. Complex interactions exist among five key parameters: soaking temperature, pH, time, H2O2 concentration, and solid-liquid ratio. Traditional single-factor experimental methods struggle to accurately capture these nonlinear relationships, leading to low process efficiency. Furthermore, starch quality evaluation involves three interdependent indicators: whiteness, tannin residue, and purity. Increasing whiteness often requires increasing H2O2 concentration, but this may lead to starch degradation and reduced purity. Extending soaking time is beneficial for tannin removal but increases energy consumption and may result in starch loss. Effective multi-objective collaborative optimization methods are lacking. While existing optimization algorithms such as genetic algorithms possess powerful search capabilities, traditional genetic algorithms suffer from blind search problems in process optimization. The lack of prior knowledge leads to slow convergence, requiring extensive experimental verification, resulting in high costs and low efficiency. Especially under multiple constraints, GA struggles to quickly find feasible solutions.

[0003] Technical issues have led to problems such as low optimization efficiency, high cost, and unstable product quality in acorn starch extraction processes, necessitating the development of a new intelligent optimization method to achieve systematic optimization of process parameters. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for extracting starch from a wild acorn, which has the advantage of optimizing the acorn starch extraction process. It solves the problems of blind search in process optimization by traditional genetic algorithms, slow convergence due to lack of prior knowledge guidance, the need for extensive experimental verification, high cost and low efficiency, especially the difficulty of finding feasible solutions quickly under multiple constraints.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for extracting starch from wild acorns, comprising the following steps: S1. Conduct preliminary experiments with wild acorns to obtain a set of process parameters. To set process parameters Chromosome encoding yields process parameter set Based on process parameter set Used as input to build the GA model; S2, and according to the process parameter set To construct the parameter range matrix As the initial search space for the GA model ; S3. Define the multi-objective optimization problem. And construct the initial fitness function; S4, using the parameter range matrix As an experimental factor Experiments are used to construct the experiment matrix D. The response value matrix was obtained by measurement. A quadratic regression model is then established to obtain the predicted response value matrix. and to The model was validated. S5. Validated models are tested using a genetic algorithm. The model seeks the optimal solution; S6, according to The optimal solution is used to construct the mapping function on the initial search space. To obtain the search space, we need to shrink the space. ; S7, in the updated search space A new population is generated, a fitness function judgment formula is constructed, the fitness value of the new population is calculated based on the judgment result, and the optimal solution and fitness value of each generation are output through genetic variation. S8, Settings based on The optimal solution output is used to construct the termination condition of the GA model. Once the termination condition is met, the algorithm stops running and outputs the optimal combination of process parameters.

[0006] Furthermore, step S4 also includes the following sub-steps: S4.1. A five-factor central composite design is used to generate an experimental matrix to obtain N sets of parameter combinations Z, and the parameter combinations Z are encoded and transformed into actual physical values. S4.2. Conduct starch extraction experiments on N sets of parameter combinations Z, and measure the experimental products of each starch extraction experiment. Given several response values, where k = (1, 2, 3), the response value matrix is ​​obtained. ; S4.3 For each response variable Three prediction models were obtained by fitting quadratic polynomial models respectively. The prediction model formula is as follows: S4.4, The response value matrix is ​​analyzed using three prediction models respectively. Three predicted response value matrices were obtained through prediction. ; S4.5 Validate the prediction model and establish the following judgment condition formula: judge If yes, proceed to S5; otherwise, return to S4.3 for supplementary experiments and re-predict. Furthermore, step S5 also includes the following sub-steps: S5.1 Constructing the comprehensive fitness function : in As the initial weights, , , are the predicted values ​​of the RSM model, respectively.

[0007] Where R1 is the predicted whiteness value, R2 is the predicted tannin residue conversion value, and R3 is the predicted starch purity value. S5.2, Based on the comprehensive fitness function To calculate the fitness value, and finally output The optimal solution of the model , .

[0008] Furthermore, step S6 also includes the following sub-steps: S6.1 Optimal Solution of Input RSM The prediction variance and initial parameter range matrix of the RSM model at each parameter. To calculate the overall confidence level of the RSM model at the optimal solution. : in It is the prediction variance of the RSM model at the optimal solution. It is the largest prediction variance among all experimental points.

[0009] S6.2 Calculate the model sensitivity in the direction of the parameters: S6.3 Constructing the dynamic shrinkage coefficient: S6.4 Calculate the parameter range after shrinkage: in It is the width of the initial range parameter; S6.5. Perform validity processing on the shrunken parameter range to obtain the final search space: Output the final search space. .

[0010] Furthermore, S7 also includes the following sub-steps: S7.1 Setting the initial population size within the search space. ; S7.2 Calculate the shrinkage ratio r of the initial search space P1: The initial search space volume, The volume of the shrunken search space; S7.3. Construct a fitness function formula based on the initial search space size: Condition A: Where F(x) is the original objective function, and Fmin and Fmax are the minimum and maximum fitness values ​​of the current population, respectively. To explore the weighting coefficients, and to see how they increase as r increases; in To prevent positive numbers from being divided by zero, k is the pressure coefficient. It decreases as r increases. This is the initial adjustment coefficient; S7.4 Calculate the fitness value of an individual based on the judgment results; S7.5 Perform genetic iteration on each individual and output the optimal solution and fitness value for each generation.

[0011] Furthermore, step S8 also includes the following sub-steps: S8.1 Construct the judgment formula based on the optimal solution output by RSM: in Let the fitness value be the current optimal individual. It is the theoretically optimal fitness value predicted by the RSM model. This represents the current iteration number. This is the preset reference iteration number; S8.2 The expression for determining the termination condition of the iteration is as follows: in The convergence threshold, This represents the minimum number of iterations.

[0012] Compared with the prior art, the technical solution of this application has the following beneficial effects: I. This method for extracting starch from wild acorns utilizes a deep integration of Response Surface Methodology (RSM) and Genetic Algorithm (GA) to form an efficient collaborative optimization framework. RSM establishes an accurate quadratic polynomial model through carefully designed experiments, providing a reliable fitness prediction function for GA and avoiding numerous time-consuming physical experiments. GA, on the other hand, leverages its powerful global search capability to efficiently explore optimal solutions within the prediction space constructed by the RSM model. This collaborative mechanism fully utilizes the local modeling accuracy of RSM and the global search advantages of GA, significantly improving optimization efficiency while ensuring the engineering practicality of the solution.

[0013] II. This method for extracting starch from wild acorns innovatively utilizes the prediction results of the RSM model to dynamically shrink the search space of the acorn extractor (GA). The optimal solution of the RSM model is used to determine the center point, and a reasonable search radius is set according to the model's confidence interval, intelligently narrowing the GA search range from an initial broad space to the most promising region. This statistical confidence-based search space shrinkage strategy avoids prematurely falling into local optima and significantly improves search efficiency, enabling the GA to perform a fine-grained search in the most promising region and significantly reducing unnecessary computational overhead.

[0014] Thirdly, this method for extracting starch from wild acorns employs an RSM model as an intelligent guidance tool for convergence judgment. By comparing the deviations between RSM predictions and actual experimental values ​​in real time, the optimization process is dynamically adjusted. When the RSM predictions are highly consistent with the actual situation, it indicates high model reliability and accelerates convergence; when the deviation is large, a model update mechanism is triggered. This intelligent iterative control strategy based on model accuracy ensures that the optimization process proceeds efficiently under the premise of model reliability, avoiding suboptimal solutions caused by premature termination and preventing resource waste caused by excessive iteration, thus achieving the best balance between optimization efficiency and accuracy. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the algorithm flow of the present invention; Figure 2 This is a diagram illustrating the fitness value determination process of the GA model in this invention. Figure 3 This is a diagram illustrating the process of determining the number of iterations in the GA model of this invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1-3 The method for extracting starch from wild acorns in this embodiment is characterized by comprising the following steps: S1. Conduct preliminary experiments to obtain the process parameter set. To set process parameters Chromosome encoding yields process parameter set Based on process parameter set Used as input to construct a genetic algorithm model; A set of process parameters can be set during actual experiments. ,in For the soaking process, soaking temperature, The pH value during the soaking process. The soaking time is the time required for the soaking process. The concentration of H2O2 during the bleaching process; preferably, a real number encoding method is used for the process parameter set. Encode; S2, and according to the process parameter set To construct the parameter range matrix As the initial search space of the genetic algorithm model ; Determine the physical feasible range of each parameter through preliminary experiments, such as The temperature range is 30℃-60℃. The pH value ranged from 4 to 8, the soaking time from 12 to 36 hours, and the H2O2 concentration was set between 0.05% and 2.5% to construct a parameter range matrix. as follows The initial population was generated as follows: S3. Define the multi-objective optimization problem. And construct a dynamic fitness function; Among them, multi-objective optimization problems , For starch whiteness, This refers to the residual tannin content. For starch purity; Based on multi-objective optimization problem The dynamic fitness function is constructed as follows; Among them These are the normalized whiteness, tannin residue, and starch purity, respectively. These are the initial weighting coefficients.

[0018] S4, using the parameter range matrix As an experimental factor Experiments are used to construct the experiment matrix D. The response value matrix was obtained by measurement. A quadratic regression model is then established to obtain the predicted response value matrix. and to The model was validated. S4.1. A five-factor central composite design is used to generate an experimental matrix to obtain N sets of parameter combinations Z, and the parameter combinations Z are encoded and transformed into actual physical values. S4.2. Conduct starch extraction experiments on N sets of parameter combinations Z, and measure the experimental products of each starch extraction experiment. Given several response values, where k = (1, 2, 3), the response value matrix is ​​obtained. ; S4.3 For each response variable Three prediction models were obtained by fitting quadratic polynomial models respectively. The prediction model formula is as follows: Where yk is the predicted value of the k-th response variable. For constant terms, The sum of all linear terms. The sum of all quadratic terms. It is the sum of all interactive items.

[0019] S4.4, The response value matrix is ​​analyzed using three prediction models respectively. Three predicted response value matrices were obtained through prediction. ; S4.5 Validate the prediction model and establish the following judgment condition formula: judge If yes, proceed to S5; otherwise, return to S4.3 for supplementary experiments and re-predict. in The coefficient of determination represents the degree to which the model explains the corresponding variable. 0.9 is a judgment threshold set based on past experience, and its value ranges from [value missing]. When it is greater than 0.9, the model has strong interpretability and can be adjusted according to actual needs in practical use. RMSE stands for Root Mean Square Error, which represents the average deviation between the model's predicted values ​​and the actual values. Its value range is... When RMSE is less than 0.2, it indicates that the prediction error is small. In actual use, it can be further reduced or increased.

[0020] By constructing accurate mathematical models, the intrinsic relationship between process parameters and quality indicators can be revealed, providing accurate fitness prediction functions for genetic algorithms and significantly reducing the number of experiments.

[0021] S5. Find the optimal solution on the validated RSM model based on the genetic algorithm model; S5.1 Constructing the comprehensive fitness function : in As the initial weights, , , are the predicted values ​​of the RSM model, respectively.

[0022] Where R1 is the predicted whiteness value, R2 is the predicted tannin residual conversion value, and R3 is the predicted purity value. S5.2, Based on the comprehensive fitness function To calculate the fitness value, and finally output The optimal solution of the model , .

[0023] S6, according to The optimal solution is used to construct a mapping function to shrink the initial search space P to obtain the search space P1; S6.1 Optimal Solution of Input RSM And the prediction variance of the RSM model at each parameter, and the initial parameter range matrix. To calculate the overall confidence level of the RSM model at the optimal solution. : in It is the prediction variance of the RSM model at the optimal solution. It is the largest prediction variance among all experimental points.

[0024] S6.2 Calculate the model sensitivity in the direction of the parameters: S6.3 Constructing the dynamic shrinkage coefficient: S6.4 Calculate the parameter range after shrinkage: in It is the width of the initial range parameter; S6.5. Perform validity processing on the shrunken parameter range to obtain the final search space: Output the final search space. .

[0025] S7. Generate a new population P2 in the updated final search space, and perform crossover and mutation on the new population P2. Output the optimal solution and fitness value for each generation. S8. Set the termination condition based on the optimal solution output by RSM. Once the termination condition is met, the algorithm stops running and outputs the optimal combination of process parameters.

[0026] S8.1 Construct the judgment formula based on the optimal solution output by RSM: in Let the fitness value be the current optimal individual. It is the theoretically optimal fitness value predicted by the RSM model. This represents the current iteration number. This is the preset reference iteration number; S8.2 The expression for determining the termination condition of the iteration is as follows: in The convergence threshold can be set according to the actual usage. This represents the minimum number of iterations.

[0027] Table 1 Please refer to Table 1. Compared with traditional GA and RSM models, this scheme, which combines the GA model with the RSM model, significantly improves search efficiency, avoids wasting computational resources in non-optimal regions, and makes the optimization process more focused on potential optimal solution regions, achieving efficient global search. While ensuring population diversity, it quickly approaches the optimal solution. By guiding the response surface model, it avoids local optima, significantly improves optimization efficiency while reducing the number of experiments, and finally obtains reliable optimal process parameters verified by experiments.

[0028] Table 2 L* represents brightness, with a larger L* indicating a brighter color and a smaller L* indicating a darker color; a larger a* indicates a reddish tint and a smaller a* indicates a greenish tint; b* indicates a yellowish tint and a bluish tint; whiteness is the sample's whiteness compared to a standard whiteness plate, with a maximum of 100; color difference is the difference between the sample and a reference sample; a color difference between 0 and 0.25 indicates a very small or no difference between the sample and the reference sample, 0.25–0.5 indicates a slight difference, 0.5–1.0 indicates a slight to moderate difference, 1.0–2.0 indicates a moderate difference, 2.0–4.0 indicates a significant difference, and above 4.0 indicates a very large difference. In the above tests, sample A was used as the reference sample to calculate the color difference. Sample E, after process optimization using this algorithm, showed the greatest improvement in whiteness during starch extraction from acorns.

[0029] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0030] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for extracting starch from wild acorns, characterized in that, Includes the following steps: S1. Conduct preliminary experiments with wild acorns to obtain a set of process parameters. To set process parameters Chromosome encoding yields process parameter set Based on process parameter set Used as input to build the GA model; S2, and according to the process parameter set To construct the parameter range matrix As the initial search space of the GA model ; S3. Define the multi-objective optimization problem. And construct the initial fitness function; S4, using the parameter range matrix As an experimental factor Experiments are used to construct the experiment matrix D. The response value matrix was obtained by measurement. A quadratic regression model is then established to obtain the predicted response value matrix. and to The model was validated. S5. The validated model is tested using a genetic algorithm. The model seeks the optimal solution; S6, according to The optimal solution is used to construct the mapping function on the initial search space. To obtain the search space, we need to shrink the space. ; S7, in the updated search space A new population is generated, a fitness function judgment formula is constructed, the fitness value of the new population is calculated based on the judgment result, and the optimal solution and fitness value of each generation are output through genetic variation. S8, Settings based on The optimal solution output is used to construct the termination condition of the GA model. Once the termination condition is met, the algorithm stops running and outputs the optimal combination of process parameters.

2. The method for extracting starch from wild acorns as described in claim 1, characterized in that, S4 further includes the following sub-steps: S4.

1. A five-factor central composite design is used to generate an experimental matrix to obtain N sets of parameter combinations Z, and the parameter combinations Z are encoded and transformed into actual physical values. S4.

2. Conduct starch extraction experiments on N sets of parameter combinations Z, and measure the experimental products of each starch extraction experiment. Given several response values, where k = (1, 2, 3), the response value matrix is ​​obtained. ; S4.3 For each response variable Three prediction models were obtained by fitting quadratic polynomial models respectively. The prediction model formula is as follows: S4.4, The response value matrix is ​​analyzed using three prediction models respectively. Three predicted response value matrices were obtained through prediction. ; S4.5 Validate the prediction model and establish the following judgment condition formula: judge If yes, proceed to S5; otherwise, return to S4.3 for supplementary experiments and re-predict.

3. The method for extracting starch from wild acorns as described in claim 1, characterized in that, S5 further includes the following sub-steps: S5.1 Constructing the comprehensive fitness function : in As the initial weights, , , are the predicted values ​​of the RSM model, respectively; Where R1 is the predicted color value, R2 is the predicted tannin residue conversion value, and R3 is the predicted starch purity value. S5.2, Based on the comprehensive fitness function To calculate the fitness value, and finally output The optimal solution of the model , .

4. The method for extracting starch from wild acorns as described in claim 1, characterized in that: S6 further includes the following sub-steps: S6.1 Optimal Solution of Input RSM The prediction variance and initial parameter range matrix of the RSM model at each parameter. To calculate the overall confidence level of the RSM model at the optimal solution. : in It is the prediction variance of the RSM model at the optimal solution. It is the largest prediction variance among all experimental points; S6.2 Calculate the model sensitivity in the direction of the parameters: S6.3 Constructing the dynamic shrinkage coefficient: S6.4 Calculate the parameter range after shrinkage: in It is the width of the initial range parameter; S6.

5. Perform validity processing on the shrunken parameter range to obtain the final search space: Output the final search space. .

5. A method for extracting starch from wild acorns as described in claim 1, characterized in that, The S7 process further includes the following sub-steps: S7.1 Setting the initial population size within the search space. ; S7.2 Calculate the shrinkage ratio r of the initial search space P1: The initial search space volume, The volume of the shrunken search space; S7.

3. Construct a fitness function formula based on the initial search space size: Condition A: Where F(x) is the original objective function, and Fmin and Fmax are the minimum and maximum fitness values ​​of the current population, respectively. To explore the weighting coefficients, and to see how they increase as r increases; in To prevent positive numbers from being divided by zero, k is the pressure coefficient. It decreases as r increases. This is the initial adjustment coefficient; S7.4 Calculate the fitness value of an individual based on the judgment results; S7.5 Perform genetic iteration on each individual and output the optimal solution and fitness value for each generation.

6. The method for extracting starch from wild acorns as described in claim 1, characterized in that, S8 further includes the following sub-steps: S8.1 Construct the judgment formula based on the optimal solution output by RSM: in The fitness value of the current best individual. It is the theoretically optimal fitness value predicted by the RSM model. This represents the current iteration number. This is the preset reference iteration number; S8.2 The expression for determining the termination condition of the iteration is as follows: in The convergence threshold, This represents the minimum number of iterations.

Images