Method for optimizing technological parameters in soaking stage in production of fried potato chips

Abstract

The invention provides a method for optimizing technological parameters in a soaking stage in production of fried potato chips, belongs to the technical field of production technologies of the fried potato chips, and realizes accurate classification and parameter management of potato raw materials by constructing a comprehensive potato raw material database. According to the method, potato raw materials are classified by adopting a multi-dimensional clustering technology, and a soaking change empirical equation set containing an adjustment factor is established, so that the quality parameters of a potato sample can be accurately predicted. The innovative process comprises the steps of performing parameter detection on a potato sample to be treated, predicting quality, optimizing soaking parameters, and monitoring and dynamically adjusting the parameters in real time in the soaking process. And by comparing the process parameters with a preset target value, continuously obtaining optimal soaking parameters, and finally completing quality evaluation of the potato sample. The accuracy, efficiency and scientificity of potato soaking treatment are remarkably improved, and the problems that in the prior art, refined management and control cannot be conducted according to the difference of raw materials, and the product quality stability is difficult to guarantee are solved.

Description

Technical Field

[0001] The invention belongs to the technical field of fried potato chips production process, and in particular relates to a method for optimizing process parameters in a soaking stage of fried potato chips production. Background Art

[0002] Fried potato chips, as a popular snack food, play an important role in the food industry. The production process usually includes washing, slicing, soaking, frying and other steps. Among them, the soaking process is the key link, which can significantly affect the taste, color and nutritional value of the finished product.

[0003] The traditional potato soaking process mainly relies on the workers' experience to set parameters, such as soaking temperature, time and soaking medium. However, due to the natural differences of potato raw materials, this "empirical" process control often cannot ensure the stability and consistency of product quality. The physical and chemical properties of different batches of potato raw materials, such as starch content, cell wall integrity, surface area, etc., will significantly affect the hydration, softening, water absorption and other behaviors during the soaking process, thus causing fluctuations in the quality of the final product.

[0004] In order to solve this problem, there have been some attempts in the industry. Some companies use experimental design to optimize the soaking parameters, but this method requires a large number of repeated experiments, which is costly and inefficient. Other researchers have tried to establish a mathematical model for the potato soaking process, but most of them only consider the linear influence of a single factor, which makes it difficult to accurately describe the complex nonlinear behavior in actual production. In addition, these models usually lack a feedback update mechanism for actual production data, making it difficult to adapt to long-term changes in raw material properties.

[0005] In short, the existing potato soaking process control technology generally has the problem of being unable to finely control the differences in raw materials and difficult to ensure the stability of product quality. Therefore, there is an urgent need for a new process parameter optimization method that can realize intelligent and adaptive control of the soaking process and effectively improve product quality and production efficiency. Summary of the invention

[0006] In view of this, the present invention provides a method for optimizing process parameters in the soaking stage of fried potato chips production, which can solve the problem that the prior art cannot perform fine control on raw material differences and is difficult to ensure product quality stability.

[0007] The present invention is achieved in that:

[0008] The present invention provides a method for optimizing process parameters in the soaking stage of fried potato chips production, comprising the following steps:

[0009] S10, establishing a potato raw material database to record raw material parameters of potato raw materials;

[0010] S20, using multi-dimensional clustering analysis on the raw material parameters in the potato raw material database to classify the potato raw materials into multiple categories, and determining initial soaking parameters of the potato raw materials in each category;

[0011] S30, detecting the raw material parameters of the potato sample to be processed and inputting a potato soaking change empirical equation group including an adjustment factor function term to obtain predicted quality parameters;

[0012] S40, optimizing the initial soaking parameters based on the predicted quality parameters to obtain optimized soaking parameters;

[0013] S50, soaking the potato sample to be processed according to the optimized soaking parameters;

[0014] S60, during the soaking process, monitoring process parameters;

[0015] S70. According to the deviation between the process parameter and the preset target value, the optimized soaking parameter is adjusted in real time to obtain the optimal soaking parameter.

[0016] On the basis of the above technical solution, the method for optimizing process parameters in the soaking stage of fried potato chips production of the present invention can also be improved as follows:

[0017] The empirical equation group of potato soaking changes includes: starch hydration equation, structure change equation, soaking water absorption dynamics equation and quality prediction equation.

[0018] Furthermore, the starch hydration equation takes the starch content and pH value in the raw material parameters and the soaking temperature and soaking time in the soaking parameters as input parameters, and outputs the degree of hydration of starch molecules in the process parameters.

[0019] Furthermore, the structural change equation takes the cell wall integrity and moisture content in the raw material parameters and the soaking medium composition in the soaking parameters as input parameters, and outputs the cell gap change and tissue structure softening degree in the process parameters.

[0020] Furthermore, the soaking water absorption kinetic equation takes the cell wall integrity and surface area in the raw material parameters, the degree of tissue structure softening in the process parameters and the soaking temperature in the soaking parameters as input parameters, and outputs the water absorption rate in the process parameters.

[0021] Furthermore, the quality prediction equation takes the enzyme activity in the raw material parameters and the process parameters as input parameters and outputs the predicted quality parameters.

[0022] Furthermore, the raw material parameters include starch content, moisture content, pH value, cell wall integrity, surface area and enzyme activity.

[0023] Furthermore, the soaking parameters include soaking temperature, soaking time and soaking medium composition.

[0024] Furthermore, the quality parameters include taste, color and nutrient retention rate; the process parameters include the degree of hydration of starch molecules, changes in cell gaps, degree of softening of tissue structure and water absorption rate.

[0025] Optionally, the method further includes S80, performing a quality assessment on the potato sample to be processed that has completed the soaking treatment to obtain actual quality parameters;

[0026] Optionally, it also includes S90, using the raw material parameters, the optimal soaking parameters, and the actual quality parameters using a machine learning algorithm to update the potato raw material database and the adjustment factor function items of the potato soaking change empirical equation group.

[0027] Further, the equations or formulas involved are described in detail as follows:

[0028] 1. The starch hydration equation is specifically expressed as follows:

[0029]

[0030] In the formula, H s is the hydration degree of starch molecules (%), T is the soaking temperature (K), t is the soaking time (min), S0 is the initial starch content (%), k1(pH) is the pH-dependent hydration rate constant (min^{-1}), α1(T) is the temperature-dependent maximum hydration degree coefficient, and f1 is the adjustment factor function term;

[0031] The parameter acquisition method is:

[0032] The expression of α1(T) is: Where a1 is the pre-exponential factor, E a is the activation energy (J / mol), R is the gas constant;

[0033] The expression of k1(pH) is: k1(pH) = k0·(1+b1|pH-pH opt | 2 ), where k0 is the base rate constant, pH opt is the optimal pH value;

[0034] f1 is the first adjustment factor function term obtained by neural network training. The input is temperature, pH, initial starch content and time, and the output is the correction coefficient.

[0035] 2. The structural change equation is specifically expressed as follows:

[0036]

[0037] Where, G is the cell gap change rate (%), S is the softening degree of tissue structure (%), I is the cell wall integrity index, W0 is the initial moisture content (%), C is the ion concentration vector of the immersion medium, M(C) is the medium composition influence matrix, and f2 is the second adjustment factor function term;

[0038] The parameter acquisition method is:

[0039] I was obtained by observing the integrity of cells under a microscope, ranging from 0 to 1;

[0040] The expression of M(C) is: Among them C i is the concentration of the i-th ion;

[0041] The coefficient matrix is ​​obtained by regression of a large amount of experimental data.

[0042] 3. The immersion water absorption dynamics equation is specifically expressed as follows:

[0043]

[0044] Where, W is the moisture content (%), W e is the equilibrium moisture content (%), A is the surface area (m 2 ), V is the volume (m 3 ), β(T) is the temperature-dependent mass transfer coefficient, γ is the softening influence coefficient, and f3 is the third adjustment factor function term;

[0045] The parameter acquisition method is:

[0046] The expression of β(T) is: Where β0 is the mass transfer coefficient at the reference temperature T0;

[0047] W e Determined by long-term immersion experiment;

[0048] A / V was obtained through image analysis.

[0049] 4. The quality prediction equation is specifically expressed as follows:

[0050]

[0051] Where, Q1 is the taste score, Q2 is the color score, Q3 is the nutrient retention rate (%), E is the enzyme activity, p ij is the weight coefficient, and f4 is the fourth adjustment factor function term.

[0052] Explanation of the equation construction principle:

[0053] 1. The starch hydration equation is based on a first-order kinetic model, taking into account the exponential effect of temperature on the maximum hydration degree (Arrhenius relationship) and the parabolic effect of pH on the hydration rate;

[0054] 2. The structural change equation uses a matrix form to describe the coupled effects of cell wall integrity and moisture content on gaps and softening, and considers the complex effects of medium composition through the ion concentration matrix;

[0055] 3. The kinetic equation of water absorption by immersion is based on Fick's diffusion law, taking into account the specific surface area, concentration difference driving, structural softening promotion effect and cell wall barrier effect;

[0056] 4. The quality prediction equation uses a linear matrix model to associate process parameters and quality indicators, and each adjustment factor function term is obtained by neural network training to describe nonlinear effects.

[0057] Implementation scheme of the first adjustment factor function term f1:

[0058] 1. A three-layer feedforward neural network structure is used, with 4 neurons in the input layer (corresponding to temperature, pH, initial starch content and time), 8 neurons in the hidden layer, and 1 neuron in the output layer (corresponding to the correction coefficient);

[0059] 2. The training data set was established in the following way: different varieties of potato samples were selected, and the actual hydration degree was measured under different temperature (278K-328K), pH value (4.0-8.0), and soaking time (0-120min) conditions, and the ratio of the actual hydration degree to the theoretical calculated value was used as the target output value;

[0060] 3. The training process uses the back propagation algorithm, the loss function is the mean square error, the initial value of the learning rate is 0.01, the Adam optimizer is used, the batch size is 32, and the training rounds are 1000 rounds;

[0061] 4. The default adjustment factor function term is initialized as: f1 = 0.2sin(0.01T) + 0.1(pH-7) 2 -0.05ln(t+1)+0.1S0-0.15.

[0062] Implementation scheme of the second adjustment factor function term f2:

[0063] 1. A convolutional neural network structure is used to process the multidimensional features of the immersion medium composition. The backbone network contains 3 convolutional layers and 2 fully connected layers. The input is the cell wall integrity index, initial moisture content and ion concentration vector, and the output is a two-dimensional correction vector.

[0064] 2. The training data set includes: using soaking media of different concentrations and compositions (NaCl, CaCl2, citric acid, etc.) to treat potato samples with different integrity and water content, measuring the cell gap change rate and softening degree, and collecting processing conditions and result data;

[0065] 3. The training process uses small batch stochastic gradient descent, the loss function combines the mean square error and Huber loss function, the learning rate uses the cosine annealing strategy, and the training rounds are 2000 rounds;

[0066] 4. The default adjustment factor function term is initialized as:

[0067] Implementation scheme of the third adjustment factor function term f3:

[0068] 1. The long short-term memory network (LSTM) structure is adopted to consider the temporal characteristics of the water absorption process. The network contains 2 layers of LSTM units (64 hidden units in each layer) and 1 fully connected layer;

[0069] 2. The training data set is obtained through continuous monitoring: under different temperature conditions, the sample moisture content, softening degree and other parameters are recorded every 5 minutes to construct a time series data set;

[0070] 3. The training process adopts a sequence-to-sequence approach, uses a teacher-forcing strategy, and the loss function is a weighted mean square error. The weight decays over time, and the number of training rounds is 1500.

[0071] 4. The default adjustment factor function term is initialized as:

[0072] Implementation scheme of the fourth adjustment factor function term f4:

[0073] 1. Adopt ensemble learning method, combining random forest and gradient boosting tree. Each base learner targets a quality indicator, with enzyme activity and process parameters as input and the correction vector of quality indicator as output.

[0074] 2. The training data set includes: collecting samples under different process parameters, having an expert group score the taste and color, testing the nutritional content, and establishing a parameter-quality correspondence database;

[0075] 3. The training process adopts cross-validation method, optimizes each base learner independently, and finally integrates them through voting or weighted average. The training rounds are 500 rounds;

[0076] 4. The default adjustment factor function term is initialized as: f4 = (0.1E + 0.2H s -0.1G+0.15S,0.15E+0.1H s+0.2g-0.1S,0.2E-0.1H s +0.15G+0.1S).

[0077] Update strategy for all adjustment factor function items:

[0078] 1. After processing each batch of samples, the deviation between the actual measured quality parameters and the predicted values ​​is back-propagated to update the network parameters;

[0079] 2. Use a sliding time window to retain only the most recent N (default N=100) processing data for model fine-tuning to ensure the model's adaptability to changes in raw material characteristics;

[0080] 3. Set the model update threshold, and trigger the complete retraining process when the prediction deviation exceeds the set value;

[0081] 4. Regularly (e.g. monthly) retrain the model with all the data to prevent catastrophic forgetting of the model.

[0082] Compared with the prior art, the beneficial effects of the method for optimizing process parameters in the soaking stage of fried potato chips production provided by the present invention are:

[0083] 1. Establish a potato raw material database to record the physical and chemical properties of potatoes from different origins and varieties, laying the foundation for subsequent process optimization. Through multi-dimensional cluster analysis, the raw materials are divided into several categories, and the initial soaking process parameters for each type of raw material are determined. This creates conditions for targeted optimization of the soaking process.

[0084] 2. A set of empirical equations for potato soaking changes, including starch hydration, cell structure changes, soaking water absorption dynamics and final product quality prediction, was established. This set of equations integrates the nonlinear effects of multiple factors such as temperature, pH, cell integrity, and ion concentration, and better simulates the complex behavior in actual production.

[0085] 3. A process parameter optimization strategy based on predicted quality parameters was proposed. Through numerical optimization algorithms, driven by product quality goals, precise control of immersion temperature, time and medium composition was achieved, greatly improving the stability of product quality.

[0086] 4. During the soaking process, key process parameters such as starch hydration degree and cell structure changes are monitored in real time, and the process parameters are automatically adjusted using an adaptive PID control algorithm. This feedback-based real-time optimization ensures that the soaking process is always maintained at the best state.

[0087] In summary, the potato soaking process parameter optimization method proposed in the present invention makes full use of key technologies such as raw material data management, process modeling, online optimization control and machine learning, realizes refined control of the soaking process, and solves the problem that the existing technology is unable to perform refined control over raw material differences and is difficult to ensure product quality stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 A flow chart of the method provided by the present invention. DETAILED DESCRIPTION

[0089] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0090] like Figure 1 FIG. 1 is a flow chart of a method for optimizing process parameters in the soaking stage of fried potato chips production provided by the present invention. The method comprises the following steps:

[0091] S10, establishing a potato raw material database to record raw material parameters of potato raw materials;

[0092] S20, using multi-dimensional clustering analysis on raw material parameters in a potato raw material database, classifying the potato raw materials into multiple categories, and determining initial soaking parameters for each category of potato raw materials;

[0093] S30, detecting the raw material parameters of the potato sample to be processed and inputting a potato soaking change empirical equation group including an adjustment factor function term to obtain predicted quality parameters;

[0094] S40, optimizing the initial soaking parameters based on the predicted quality parameters to obtain optimized soaking parameters;

[0095] S50, soaking the potato sample to be processed according to the optimized soaking parameters;

[0096] S60, during the soaking process, monitoring process parameters;

[0097] S70, adjusting the optimized soaking parameters in real time according to the deviation between the process parameters and the preset target values ​​to obtain the optimal soaking parameters.

[0098] The specific implementation methods of the above steps are described in detail below:

[0099] The specific implementation method of step S10 is: first, establish a potato raw material database to record the characteristic parameters of different potato raw materials. These raw material parameters include starch content, moisture content, pH value, cell wall integrity, surface area and enzyme activity. The purpose of establishing this database is to provide basic data support for subsequent raw material classification and process parameter optimization. The specific method is:

[0100] 1) Collect potato samples from different origins and varieties, and after pretreatment such as washing and slicing, measure various raw material parameters. Use instrumental analysis to measure starch content, moisture content and pH value; evaluate cell wall integrity index through microscopic observation; use image analysis software to calculate surface area; and use enzyme activity detection kit to measure enzyme activity.

[0101] 2) The collected raw material parameter data is established in the form of a spreadsheet or database, with each record corresponding to a potato sample. The record includes information such as sample number, origin, variety, and the measured values ​​of the above 6 raw material parameters. At the same time, a reasonable data management mechanism is set up, including regular backup and data verification.

[0102] 3) For samples with missing parameter data, interpolation estimation or machine learning methods are used to complete them to ensure the completeness of database information. Data with obvious outliers need to be further verified and eliminated.

[0103] 4) In order to meet the needs of different users, the database can be structured and a multi-level index can be established to support rapid retrieval and screening based on parameter ranges, origins, varieties and other conditions.

[0104] 5) Update the database regularly and input the newly collected raw material parameter information in time. When the distribution of a raw material parameter changes significantly, notify the relevant personnel in time to adjust the process parameters.

[0105] In general, the establishment of such a detailed potato raw material database provides reliable basic data support for subsequent process optimization. It records the physical and chemical properties of different potato samples and lays the foundation for studying the influence of raw materials on the process.

[0106] The specific implementation method of step S20 is: using a multi-dimensional cluster analysis method to classify the potato raw materials in the database and determine the initial parameters of the soaking process for each type of raw material. The specific method is as follows:

[0107] 1) First, choose a suitable clustering algorithm. Here we recommend the k-means algorithm because it is simple, efficient, and can handle continuous data well. It should be noted that the algorithm's hyperparameters, such as the number of clusters k, need to be tuned to obtain more reasonable classification results.

[0108] 2) The six raw material parameters in the database are used as input features for clustering. Since the dimensions of each parameter are different, they need to be standardized first and mapped to the [0,1] interval.

[0109] 3) Run the k-means clustering algorithm and divide the potato raw materials into several categories according to the clustering results. Each category represents a relatively uniform potato raw material characteristic. By analyzing the distribution of the eigenvalues ​​of each category, the corresponding initial soaking process parameters are determined.

[0110] 4) The soaking process parameters include soaking temperature, soaking time and soaking medium composition. For temperature and time parameters, 3-5 different levels can be set, such as temperature (293K, 303K, 313K) and time (30min, 45min, 60min). The soaking medium can be water, salt water, acidic solution, etc., and the initial ion concentration range can be determined.

[0111] 5) Record each potato raw material category and its corresponding initial soaking parameters in the database for subsequent process optimization. It should be noted that these initial parameters are only a starting point and need to be further optimized.

[0112] In this way, the potato samples in the raw material database were divided into several categories according to their physical and chemical properties, and the corresponding initial soaking process parameters were determined for each type of raw material. The purpose of this is to better optimize the process according to the characteristics of different raw materials and improve the stability and controllability of product quality.

[0113] The specific implementation method of step S30 is: firstly, the raw material parameters of the potato sample to be processed are detected, and then these parameters are input into the pre-established potato soaking change empirical equation group to calculate the predicted product quality parameters. The specific method is as follows:

[0114] 1) Determine six raw material parameters of the potato sample to be processed, including starch content, moisture content, pH value, cell wall integrity, surface area and enzyme activity. The determination methods of these parameters are similar to those described in step S10.

[0115] 2) Substitute the measured raw material parameters into the empirical equations of potato soaking changes. This equation group includes 4 sub-equations, which respectively describe the starch hydration process, cell structure change process, soaking water absorption dynamics process, and the prediction of the final product quality.

[0116] 3) The starch hydration equation takes into account the effects of temperature, pH and initial starch content on the degree of starch hydration, and is described by a first-order kinetic model including an adjustment factor.

[0117] 4) The structural change equation is in matrix form. The input parameters include cell wall integrity, initial moisture content and ion concentration of the immersion medium. The output is the cell gap change rate and the degree of tissue structure softening.

[0118] 5) The immersion water absorption kinetic equation is based on Fick's law and takes into account the effects of factors such as specific surface area, concentration gradient, structural softening and cell wall integrity on the water absorption rate.

[0119] 6) The quality prediction equation adopts a linear matrix form to associate the above process parameters with product quality indicators (taste, color, and nutrient retention rate).

[0120] 7) In these sub-equations, multiple adjustment factor function terms are also set to describe some complex nonlinear effects. These adjustment factor function terms will be optimized and updated in subsequent steps.

[0121] By substituting the raw material parameters of the potato sample to be processed into the above empirical equations, the product quality index of the sample under the soaking process can be predicted, which provides a basis for the subsequent optimization of process parameters.

[0122] The specific implementation of step S40 is: according to the product quality parameters predicted in step S30, the initial soaking process parameters determined in step S20 are optimized to obtain more optimized soaking parameters. The specific method is as follows:

[0123] 1) First, set the target values ​​for product quality. For example, taste score ≥ 4.5 points, color score ≥ 4.0 points, and nutrient retention rate ≥ 90%. These target values ​​can be determined based on market demand and corporate standards.

[0124] 2) Compare the quality parameter predicted in step S30 with the above target value and calculate the deviation between the two.

[0125] 3) Using a numerical optimization algorithm, with minimizing the deviation of quality parameters as the objective function, the initial soaking temperature, time and medium composition in step S20 are optimized. Here, optimization methods such as gradient descent method and genetic algorithm can be used.

[0126] 4) During the optimization process, the importance weights of various quality parameters need to be considered simultaneously. For example, the weight of taste score can be set to 0.4, the weight of color score can be set to 0.3, and the weight of nutrient retention rate can be set to 0.3.

[0127] 5) When the optimization algorithm converges, the optimal soaking process parameters for the potato raw material can be obtained. These parameters include soaking temperature, soaking time and soaking medium composition.

[0128] 6) Record the optimized immersion parameters in the database for subsequent use in actual production. At the same time, it is also necessary to retain the intermediate results of the optimization process to provide a reference for further process improvement.

[0129] In this way, according to the predicted product quality indicators, the initial soaking process parameters were optimized by numerical optimization method, and more satisfactory soaking process parameters were obtained, which laid the foundation for the subsequent actual soaking treatment.

[0130] The specific implementation method of step S50 is: according to the soaking parameters optimized in step S40, the potato sample to be processed is actually soaked. The specific method is as follows:

[0131] 1) First, prepare a sufficient amount of potato samples to be processed. An automatic slicer can be used to cut the potato blocks into uniform thin slices.

[0132] 2) Put the cut potato slices into a soaking tank and add the soaking medium optimized in step S40. The temperature, pH value and ion concentration of the soaking medium should be consistent with the optimized parameters.

[0133] 3) Start the stirring device and stir slowly at an appropriate speed (such as 50-100 rpm) to ensure that the potato slices are fully soaked. At the same time, strictly control the soaking time to ensure that it is consistent with the optimized parameters.

[0134] 4) During the soaking process, monitor and record the actual process parameter changes, including the degree of hydration of starch molecules, changes in intercellular spaces, degree of softening of tissue structure, and water absorption rate. These data will be used for further optimization in subsequent steps.

[0135] 5) When the soaking time is over, immediately remove the soaking tank from the heating device and quickly cool it to room temperature. Then remove the potato slices from the soaking medium and rinse them with tap water.

[0136] 6) The potato slices that have completed the soaking process are properly dried or dehydrated to reduce the moisture content and prepare for the subsequent frying process.

[0137] Through this step, the actual soaking operation of the potato samples to be treated was carried out according to the optimized soaking parameters. During this process, some key process parameter changes were also recorded to provide data support for further optimization.

[0138] The specific implementation method of step S60 is: during the soaking treatment process of step S50, some important process parameters are monitored and recorded in real time to provide a basis for subsequent process optimization. The specific method is as follows:

[0139] 1) First, set up the necessary online monitoring equipment, including temperature sensors, pH meters, salinity meters, etc., to collect key parameter data in real time during the immersion process. At the same time, appropriately install sampling devices to facilitate regular collection of samples for offline analysis.

[0140] 2) The monitored process parameters include: the degree of hydration of starch molecules, changes in cell gaps, the degree of softening of tissue structure, and the rate of water absorption.

[0141] 3) The degree of hydration of starch molecules can be estimated by measuring the reducing sugar content or viscosity change in the solution. Changes in intercellular spaces and softening of tissue structure can be quantitatively evaluated with the help of microscopic observation and image analysis techniques. The water absorption rate can be measured by periodic weighing.

[0142] 4) In order to ensure the accuracy and reliability of monitoring data, monitoring equipment needs to be calibrated and maintained regularly. At the same time, a complete data collection, processing and storage mechanism should be established to ensure the integrity of monitoring data.

[0143] 5) During the entire immersion process, the changes in the above process parameters are continuously recorded to form time series data. These data will provide an important basis for the subsequent optimization of process parameters.

[0144] 6) When there is a large deviation between the monitoring data and the expected target value, promptly notify relevant personnel to make process adjustments. For example, if it is observed that the degree of starch hydration is significantly lower than expected, the soaking temperature can be appropriately increased or the soaking time can be extended.

[0145] By real-time monitoring and recording the changes of key process parameters, the dynamic characteristics of potato soaking process can be fully understood, providing a scientific basis for further optimization of process parameters. These monitoring data can also be used to update and improve the empirical equations of potato soaking changes.

[0146] The specific implementation of step S70 is: according to the changes in the process parameters monitored in step S60, the soaking process parameters optimized in step S40 are adjusted in real time to achieve the best soaking effect. The specific steps are as follows:

[0147] 1) First, set the target value range for each process parameter. For example, the starch hydration target is 80%-90%, the cell gap change rate is controlled at 20%-30%, the tissue structure softening degree does not exceed 40%, and the water absorption rate is maintained at 0.5%-1.0% / min. These target values ​​can be determined based on experience or product requirements.

[0148] 2) Real-time monitoring of the process parameter data recorded in step S60, comparing it with the above target value, and calculating the current deviation.

[0149] 3) Adopt adaptive proportional-integral-differential (PID) control algorithm to dynamically adjust parameters such as immersion temperature, time and medium composition according to the deviation. The amplitude and frequency of adjustment are proportional to the deviation.

[0150] 4) For example, if the starch hydration degree is found to be low, the soaking temperature can be appropriately increased; if the cell gap changes too much, the calcium ion concentration in the soaking medium can be appropriately reduced.

[0151] 5) At the same time, in order to prevent excessive adjustment from causing serious deviation of quality indicators, reasonable adjustment range and rate limit conditions need to be set. For example, the temperature change should not exceed 2 degrees Celsius / min, and the time change should not exceed 5 minutes.

[0152] 6) Record and archive the real-time adjusted soaking process parameter values ​​to form a complete process data log. These data can not only be used for post-analysis, but also provide reference for process optimization of the next batch.

[0153] Through this real-time adjustment method based on feedback control, it can ensure that the soaking process is always maintained in the optimal state, maximizing the stability and consistency of product quality. This automated adjustment method greatly reduces the dependence on manual experience.

[0154] The specific implementation method of step S80 is: perform actual quality assessment on the potato samples that have completed the soaking treatment to obtain the measured values ​​of various quality parameters. The specific method is as follows:

[0155] 1) First, determine the quality indicators that need to be evaluated based on the product quality standards, including taste score, color score and nutrient retention rate.

[0156] 2) For taste and color scores, experienced sensory evaluation experts can be invited to score. They need to conduct a careful sensory evaluation of the samples based on the pre-established scoring criteria.

[0157] 3) The nutrient retention rate needs to be determined through laboratory analysis. The content of major nutrients such as protein, vitamin C, starch, etc. in the sample can be measured and calculated by comparing with the content of the raw materials.

[0158] 4) In order to ensure the reliability of the test results, the samples need to be tested repeatedly and the average value and standard deviation calculated. At the same time, the test equipment must be calibrated and maintained regularly.

[0159] 5) The actual quality parameter values ​​obtained are recorded in the database and compared with the quality indicators predicted in step S30. If there is a large deviation, it is necessary to promptly feedback to relevant personnel for process improvement.

[0160] 6) In addition to regular quality testing of samples, special inspections can also be carried out at irregular intervals to comprehensively evaluate the stability of product quality levels.

[0161] Through this step, the quality performance data of the potato samples that have completed the soaking treatment under actual production conditions were obtained. These data can not only be checked with the predicted results, but also provide important basis for subsequent process optimization.

[0162] The specific implementation method of step S90 is: the raw material parameters, optimized soaking parameters and actual quality parameters obtained in steps S10 to S80 are continuously optimized and updated by machine learning algorithm for the potato raw material database and the potato soaking change empirical equation group. The specific method is as follows:

[0163] 1) First, build a comprehensive machine learning model to analyze the relationship between raw material characteristics, process parameters and product quality. The model should include multiple algorithms, such as neural networks, decision trees, support vector machines, etc., to capture various complex nonlinear relationships.

[0164] 2) The raw material database established in step S10, as well as the raw material parameters, process parameters and quality parameter data obtained in steps S30 and S80, are used as training samples for the machine learning model.

[0165] 3) During the training process, it is necessary to optimize the model parameters for different adjustment factor function items (such as f1, f2, f3, f4) to improve the prediction accuracy. At the same time, it is also necessary to use cross-validation and other methods to evaluate the generalization performance of the model.

[0166] 4) After the training is completed, the optimized machine learning model is embedded into the empirical equations of potato soaking changes to replace the original adjustment factor function terms. This can greatly improve the prediction accuracy of the equations.

[0167] 5) At the same time, the trained machine learning model is also used to update and expand the raw material database. For example, the model can be used to predict the raw material parameters of unknown samples to fill in the gaps in the database.

[0168] 6) In order to ensure the timeliness of the model, a rolling update mechanism needs to be established. That is, after processing each batch of samples, the newly acquired data will be included in the training samples to continuously optimize the model. At the same time, a certain update threshold is set, and when the prediction deviation exceeds the standard, the retraining of the full data is triggered.

[0169] 7) In addition, the entire machine learning model needs to be fully retrained and evaluated regularly (e.g., monthly) to prevent model performance degradation due to long-term changes in feedstock properties or process conditions.

[0170] Specifically, the principle of the present invention is to establish a complete potato soaking process modeling and optimization control framework. The key is to make full use of the potato physical and chemical property information recorded in the raw material database, classify the raw materials through multi-dimensional cluster analysis, and lay the foundation for targeted process optimization.

[0171] For each type of potato raw material, the present invention constructs a set of empirical equations including starch hydration, cell structure change, soaking water absorption dynamics and final product quality prediction. These sub-equations fully consider the nonlinear coupling effects of multiple factors such as temperature, pH, ion concentration, cell integrity, etc., and better describe the complex soaking behavior in actual production.

[0172] On this basis, the present invention proposes a process parameter optimization strategy based on predicted quality parameters. Through the numerical optimization algorithm, driven by the product quality goal, the precise control of the soaking temperature, time and medium composition is achieved. This goal-oriented optimization method ensures the stability and controllability of the final product quality.

[0173] A specific embodiment 1 of the present invention is provided below. The specific implementation of the steps of this embodiment 1 is described in detail as follows: The specific implementation of step S10 is:

[0174] First, a potato raw material database is established to record the characteristic parameters of different potato raw materials. These raw material parameters can be expressed as: X = {x1, x2, ..., x6}, where x1 represents starch content, %, x2 represents moisture content, %, x3 represents pH value, x4 represents cell wall integrity index (0-1), and x5 represents surface area, m 2 ,x6 represents enzyme activity, U / g.

[0175] The specific steps are as follows:

[0176] 1) Collect potato samples from different origins and varieties and measure various raw material parameters. Starch content can be measured by potassium dichromate-sulfuric acid colorimetry; moisture content can be measured by drying and weighing method; pH value can be measured by pH meter; cell wall integrity can be evaluated by microscopic observation as an index between 0 and 1; surface area can be measured by image analysis software; enzyme activity can be measured by enzyme activity detection kit.

[0177] 2) The collected raw material parameter data is established in the form of a spreadsheet or database, with each record corresponding to a potato sample. At the same time, a reasonable data management mechanism is set up, including regular backup, data verification, etc.

[0178] 3) For samples with missing parameter data, interpolation estimation or machine learning methods can be used to complete them. For data with outliers, further verification and elimination are required.

[0179] 4) In order to meet the needs of different users, the database can be structured and a multi-level index can be established to support rapid retrieval and screening based on parameter ranges, origins, varieties and other conditions.

[0180] 5) Update the database regularly and input the newly collected raw material parameter information in time. When the distribution of a raw material parameter changes significantly, notify the relevant personnel in time to adjust the process parameters.

[0181] The specific implementation method of step S20 is:

[0182] The potato raw material database established in step S10 is classified into multiple dimensions using the k-means clustering algorithm. The input feature of the clustering is the raw material parameter X, and the clustering result divides the potato raw materials into K categories. For the kth category of raw materials, its initial soaking process parameters can be expressed as: k ={T k ,t k ,C k}, where T k Indicates the soaking temperature, K, t k Indicates soaking time, min, C k ={c k1 ,c k2 ,…,c kn} represents the ion concentration vector of the immersion medium, mol / L.

[0183] The specific steps are as follows:

[0184] 1) Select a suitable k-means clustering algorithm and standardize the input feature X to map it to the interval [0,1].

[0185] 2) According to the clustering results, the potato raw materials are divided into K categories. The eigenvalue distribution of each category is analyzed to determine the corresponding initial soaking process parameters.

[0186] 3) Temperature parameter T k It can be set to 3-5 levels such as 293K, 303K, 313K, etc.; the time parameter t k It can be set to 3-5 levels such as 30min, 45min, 60min, etc.; the immersion medium can be water, salt water, acid solution, etc., and the initial ion concentration range can be determined.

[0187] 4) Record each potato raw material category and its corresponding initial soaking parameters in the database for subsequent process optimization.

[0188] The specific implementation method of step S30 is:

[0189] Firstly, the raw material parameter X of the potato sample to be processed is detected, and then it is input into the pre-established potato soaking change empirical equation group to calculate the predicted product quality parameter Q = {Q1, Q2, Q3}, where Q1 represents the taste score, dimensionless, Q2 represents the color score, dimensionless, and Q3 represents the nutrient retention rate, %.

[0190] The empirical equations for potato soaking changes include the following four sub-equations:

[0191] 1) Starch hydration equation:

[0192]

[0193] Among them, H s represents the degree of hydration of starch molecules, %, T represents the soaking temperature, K, t represents the soaking time, min, S0 represents the initial starch content, %, k1(pH) represents the pH-dependent hydration rate constant, min -1 ,α1(T) represents the temperature-dependent maximum hydration degree coefficient, dimensionless, and f1 is the adjustment factor function term.

[0194] 2) Structural change equation:

[0195]

[0196] Where G represents the rate of change of intercellular space, %, S represents the degree of softening of tissue structure, %, I represents the cell wall integrity index, dimensionless, W0 represents the initial moisture content, %, C represents the ion concentration vector of immersion medium, mol / L, M(C) represents the medium composition influence matrix, and f2 is the second adjustment factor function term.

[0197] 3) Immersion water absorption dynamics equation:

[0198]

[0199] Where W represents the moisture content, %, W e represents the equilibrium moisture content, %, and A represents the surface area, m 2 ,V represents volume,m 3 , β(T) represents the temperature-dependent mass transfer coefficient, γ is the softening influence coefficient, and f3 is the third adjustment factor function term.

[0200] 4) Quality prediction equation:

[0201]

[0202] Where E represents enzyme activity, U / g, p ij is the weight coefficient, and f4 is the fourth adjustment factor function term.

[0203] These sub-equations fully consider the complex influence of raw material characteristics and process parameters on the soaking process, and describe the nonlinear effects by adjusting the factor function terms f1, f2, f3, and f4.

[0204] The specific implementation of step S40 is:

[0205] According to the product quality parameter Q predicted in step S30, the initial soaking process parameter P determined in step S20 is k Optimize and get the optimized soaking parameters

[0206] The specific steps are as follows:

[0207] 1) Set product quality target values For example

[0208] 2) Calculate the prediction quality parameter Q and the target value Q target Deviation between:

[0209] ΔQ=QQ target ;

[0210] 3) Using gradient descent method to optimize the immersion parameter P k , with minimizing the quality parameter deviation ΔQ as the objective function:

[0211]

[0212] 4) During the optimization process, the weight factor of the quality parameter ω = {ω1, ω2, ω3} can be introduced to minimize the weighted deviation:

[0213]

[0214] Where ⊙ represents the Hadamard product.

[0215] 5) Optimize the soaking parameters Including temperature time and medium composition Record it in the database for later use.

[0216] Through this optimization method, the optimal soaking process parameters that meet the quality goals can be obtained for different types of potato raw materials.

[0217] The specific implementation method of step S50 is:

[0218] According to the soaking parameters optimized in step S40 The potato samples to be treated are subjected to actual soaking treatment. The specific steps are as follows:

[0219] 1) Put the cut potato slices into a soaking tank, add the soaking medium optimized in step S40, and set the temperature to Soaking time is

[0220] 2) Start the stirring device and stir slowly at an appropriate speed (such as 50-100 rpm) to ensure that the potato slices are fully soaked.

[0221] 3) During the soaking process, monitor and record the actual process parameter changes, including the degree of hydration of starch molecules H s , intercellular space change rate G, tissue structure softening degree S and water absorption rate wait.

[0222] 4) When the soaking time is over, quickly cool down to room temperature, remove the potato slices from the soaking medium, and rinse with water.

[0223] 5) The soaked potato slices are properly dried or dehydrated to reduce the moisture content.

[0224] Through this step, the samples to be treated were actually immersed according to the optimized immersion parameters, and the changes of some key process parameters were recorded. These data will provide support for subsequent process optimization.

[0225] The specific implementation of step S60 is:

[0226] During the soaking process in step S50, some important process parameters are monitored and recorded in real time to provide a basis for subsequent process optimization.

[0227] The specific steps are as follows:

[0228] 1) Degree of hydration of starch molecules H s It can be estimated by measuring the reducing sugar content in the solution or the change in viscosity.

[0229] 2) The intercellular gap change rate G and the degree of tissue softening S can be quantitatively evaluated with the help of microscopic observation and image analysis technology.

[0230] 3) Water absorption rate It can be measured by periodic weighing.

[0231] 4) Record the process parameter data obtained from real-time monitoring into a time series to form a complete process monitoring log.

[0232] 5) When there is a large deviation between the monitoring data and the expected target value, promptly notify relevant personnel to make process adjustments.

[0233] Through this real-time monitoring method, the dynamic characteristics of potato soaking process can be fully understood, providing a basis for further optimizing process parameters. At the same time, these monitoring data can also be used to update and improve the empirical equations of potato soaking changes.

[0234] The specific implementation of step S70 is:

[0235] According to the process parameter changes monitored in step S60, the adaptive PID control algorithm is used to adjust the soaking process parameters optimized in step S40 in real time. In order to achieve the best soaking effect.

[0236] The specific steps are as follows:

[0237] 1) Set the target value range for each process parameter, such as G target ∈[20%,30%],S target ≤40%,

[0238] 2) Real-time monitoring of process parameters H s , G, S, and Calculate the deviation from the target value:

[0239]

[0240] ΔG=GG target ;

[0241] ΔS=SS target ;

[0242]

[0243] 3) Adopt adaptive PID control algorithm to dynamically adjust the soaking parameters according to the above deviations

[0244]

[0245] Among them, K p , K i , K d are the proportional, integral and differential coefficients of the PID controller.

[0246] 4) To prevent excessive adjustment, reasonable adjustment range and rate limit conditions need to be set, such as temperature change not exceeding 2 degrees Celsius / min and time change not exceeding 5 minutes.

[0247] 5) The real-time adjusted soaking process parameter P * Records are archived to form a complete process data log.

[0248] In order to further better understand and implement the present invention, the following provides Example 2 of a specific application scenario of the present invention: A food company produces and sells various fried potato chips, among which classic potato chips are the best-selling. In order to improve the stability and consistency of product quality, the company decides to adopt the potato soaking process parameter optimization method proposed in the present invention.

[0249] First, the company established a comprehensive potato raw material database. They collected 200 potato samples from different origins and varieties. These samples were tested for comprehensive physical and chemical indicators, including starch content, moisture content, pH value, cell wall integrity, surface area and enzyme activity. All test data were entered into a spreadsheet to form a standardized raw material parameter database, as shown in Table 1.

[0250] Table 1 Raw material parameter information of some potato samples:

[0251]

[0252] Next, the company used the k-means clustering algorithm to analyze these raw material parameters and divided the 200 potato samples into 5 categories. By analyzing the characteristic value distribution of each type of sample, the initial soaking process parameters were determined as shown in Table 2 below:

[0253] Table 2 Initial soaking process parameters

[0254]

[0255]

[0256] In this way, the company has established a database covering different types of potato raw materials and their corresponding initial soaking parameters.

[0257] Next, the company conducted a series of experiments and established an empirical equation group for the changes in potato soaking. The specific process is as follows:

[0258] 1. Starch hydration equation:

[0259] They selected potato samples of different varieties and measured the actual starch hydration degree under the conditions of temperature (278K-328K), pH value (4.0-8.0), and soaking time (0-120min). These measured data were compared with the theoretical calculated values, and a three-layer feedforward neural network model was trained as the adjustment factor function f1 of the starch hydration equation. The input layer of the neural network has 4 neurons, corresponding to temperature, pH value, initial starch content and soaking time, 8 neurons in the hidden layer, and 1 neuron in the output layer (corresponding to the correction coefficient). The training process uses the back propagation algorithm, the loss function is the mean square error, the initial value of the learning rate is 0.01, the Adam optimizer is used, the batch size is 32, and the training rounds are 1000 rounds.

[0260] 2. Structural change equation:

[0261] In order to establish the structural change equation, the company conducted a large number of experiments: using soaking media of different concentrations and compositions (NaCl, CaCl2, citric acid, etc.) to treat potato samples with different integrity and water content, and measuring the cell gap change rate and softening degree. These treatment conditions and result data were collected as training samples. The company used a convolutional neural network model to fit the second adjustment factor function f2 of the structural change equation. The input was the cell wall integrity index, initial moisture content and ion concentration vector, and the output was a two-dimensional correction vector. The training process used small batch stochastic gradient descent, the loss function was a combination of mean square error and Huber loss function, the learning rate used a cosine annealing strategy, and the training round was 2000 rounds.

[0262] 3. Immersion water absorption dynamics equation:

[0263] In order to obtain the time series data of the soaking water absorption dynamics process, the company recorded the moisture content, softening degree and other parameters of the samples every 5 minutes under different temperature conditions to form a time series data set. Based on these data, the company trained a long short-term memory network model containing 2 layers of LSTM units (64 hidden units in each layer) and 1 layer of fully connected layer as the third adjustment factor function f3. The training process adopts a sequence-to-sequence method, using a teacher forcing strategy, the loss function is a weighted mean square error, the weight decays over time, and the training round is 1500 rounds.

[0264] 4. Quality prediction equation:

[0265] Finally, the company collected samples under different process parameters, and the expert group scored the taste and color, and tested the nutritional content at the same time, and established a parameter-quality correspondence database. Based on these data, the company used an ensemble learning method, combining random forests and gradient boosting trees as the fourth adjustment factor function f4 of the quality prediction equation. Each base learner optimizes a quality indicator, with enzyme activity and process parameters as input and a correction vector of the quality indicator as output. The training process uses cross-validation, and the training rounds are 500 rounds.

[0266] Through the above experiments and modeling work, the company finally built a complete set of empirical equations for potato soaking changes. This set of equations can better describe the complex behavior in actual production and provide strong support for subsequent process optimization.

[0267] With the support of the empirical equations, the company began to optimize the soaking process parameters. Taking a batch of potato samples (No. A003) as an example, the specific process is as follows:

[0268] 1. First, the raw material parameters of this batch of potatoes were measured: starch content 15.6%, moisture content 77.9%, pH value 6.8, cell wall integrity 0.78, surface area 0.032m 2 , enzyme activity 2.5U / g.

[0269] 2. Substitute these parameters into the empirical equations to calculate the predicted product quality indicators:

[0270] Taste score Q1 = 4.1 points;

[0271] Color score Q2 = 3.8 points;

[0272] Nutrition retention rate Q3 = 87%;

[0273] 3. Compare the predicted quality indicators with the target values ​​set by the enterprise (taste ≥ 4.5 points, color ≥ 4.0 points, nutrient retention rate ≥ 90%) and calculate the deviation:

[0274] ΔQ1=4.1-4.5=-0.4;

[0275] ΔQ2=3.8-4.0=-0.2;

[0276] ΔQ3=87-90=-3%;

[0277] 4. The weighted gradient descent method is used to optimize the soaking parameters, with the objective function of minimizing the deviation of quality indicators:

[0278]

[0279] The weight factor ω={0.4,0.3,0.3};

[0280] The optimized soaking parameters are:

[0281] temperature

[0282] time

[0283] The soaking medium was 2% citric acid solution;

[0284] 5. Apply the optimized soaking parameters to actual production. During the soaking process, the company monitored key process parameters such as starch hydration degree, cell gap changes, tissue structure softening degree and water absorption rate in real time, as shown in Table 3.

[0285] Table 3 Monitoring data

[0286]

[0287] 6. Based on the deviation between the monitoring data and the preset target, the company adjusted the immersion parameters in real time:

[0288] In the first 40 minutes, the temperature and time were appropriately increased according to the deviation, so that the starch hydration degree and water absorption rate met the standards;

[0289] In the last 15 minutes, the temperature and time were appropriately lowered to control the changes in cell gaps and the softening of tissue structure within a reasonable range.

[0290] 7. Finally, after the dynamically adjusted soaking process, the quality indicators of the potato chips samples produced are as follows:

[0291] Taste score Q1 = 4.6 points;

[0292] Color score Q2 = 4.2 points;

[0293] Nutrition retention rate Q3 = 92%;

[0294] These indicators are in line with the company's target requirements, and product quality has been significantly improved compared to before.

[0295] 8. The company will feed back the raw material parameters, soaking parameters, process parameters and quality parameters recorded in this production process to the potato raw material database and empirical equation group, and use machine learning technology to continuously optimize and update. In this way, for different batches of potato raw materials encountered in subsequent production, the company can quickly determine the initial soaking parameters based on the database information and adjust them online in real time to ensure the stability of product quality.

[0296] It should be noted that the explanations of the variables involved in the present invention are shown in Table 4 below.

[0297] Table 4 Variable explanation table

[0298]

[0299]

[0300] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for optimizing process parameters in the soaking stage of fried potato chips production, characterized in that: The following steps are involved: S10, establishing a potato raw material database to record raw material parameters of potato raw materials; S20, using multi-dimensional clustering analysis on the raw material parameters in the potato raw material database to classify the potato raw materials into multiple categories, and determining initial soaking parameters of the potato raw materials in each category; S30, detecting the raw material parameters of the potato sample to be processed and inputting a potato soaking change empirical equation group including an adjustment factor function term to obtain predicted quality parameters; S40, optimizing the initial soaking parameters based on the predicted quality parameters to obtain optimized soaking parameters; S50, soaking the potato sample to be processed according to the optimized soaking parameters; S60, during the soaking process, monitoring process parameters; S70. According to the deviation between the process parameter and the preset target value, the optimized soaking parameter is adjusted in real time to obtain the optimal soaking parameter.

2. The method for optimizing process parameters in the soaking stage of fried potato chips production according to claim 1, characterized in that: The potato soaking change empirical equation group includes: starch hydration equation, structure change equation, soaking water absorption dynamics equation and quality prediction equation.

3. The method for optimizing process parameters in the soaking stage of fried potato chips production according to claim 2, characterized in that: The starch hydration equation takes the starch content and pH value in the raw material parameters and the soaking temperature and soaking time in the soaking parameters as input parameters, and outputs the degree of hydration of starch molecules in the process parameters.

4. The method for optimizing process parameters in the soaking stage of fried potato chips production according to claim 3, characterized in that: The structural change equation takes the cell wall integrity and moisture content in the raw material parameters and the soaking medium composition in the soaking parameters as input parameters, and outputs the cell gap change and tissue structure softening degree in the process parameters.

5. The method for optimizing process parameters in the soaking stage of fried potato chips production according to claim 4, characterized in that: The soaking water absorption kinetic equation takes the cell wall integrity and surface area among the raw material parameters, the degree of tissue structure softening among the process parameters and the soaking temperature among the soaking parameters as input parameters, and outputs the water absorption rate among the process parameters.

6. The method for optimizing process parameters in the soaking stage of fried potato chips production according to claim 5, characterized in that: The quality prediction equation takes the enzyme activity in the raw material parameters and the process parameters as input parameters and outputs the predicted quality parameters.

7. The method for optimizing process parameters in the soaking stage of fried potato chips production according to claim 6, characterized in that: The feedstock parameters include starch content, moisture content, pH, cell wall integrity, surface area and enzyme activity.

8. The method for optimizing process parameters in the soaking stage of fried potato chips production according to claim 7, characterized in that: The soaking parameters include soaking temperature, soaking time and soaking medium composition.

9. The method for optimizing process parameters in the soaking stage of fried potato chips production according to claim 8, characterized in that: The quality parameters include taste, color and nutrient retention rate; the process parameters include starch molecule hydration degree, cell gap change, tissue structure softening degree and water absorption rate.

10. The method for optimizing process parameters in the soaking stage of fried potato chips production according to claim 9, characterized in that: The process further comprises S80, performing a quality assessment on the potato sample to be processed after the soaking treatment to obtain actual quality parameters; S90, using the raw material parameters, the optimal soaking parameters, and the actual quality parameters using a machine learning algorithm to update the potato raw material database and the adjustment factor function items of the potato soaking variation empirical equation group.

Images