Method for driving pickled vegetable automatic production line through Internet of Things and intelligent control system

Abstract

The invention discloses a method for driving a pickled vegetable automatic production line through the Internet of Things and an intelligent control system, and the method comprises the steps: S1, carrying out the pretreatment of a raw material through a raw material module, and determining the type of a required strain and the volume of a bacterial suspension; s2, a bacterial suspension collection module monitors the growth speed of the bacterial strain and records the volume of the collected bacterial suspension and the type of the bacterial strain; s3, heating the culture medium of the corresponding strain according to the difference between the volume of the required bacterial suspension and the volume of the collected bacterial suspension; s4, the pickling module mixes the bacterial suspension with the pretreated raw materials, and predicts the pickling state based on a long-short-term memory network model; and S5, finely adjusting the long-short-term memory network model according to the total score of the pickled vegetables. Based on the difference between the demanded quantity and the collection quantity of the bacterial suspension, the corresponding culture medium is heated, and rapid collection of the bacterial suspension is achieved; the pickling state is predicted by adopting the long-short-term memory network model, and the model is corrected based on the quality of the pickled vegetables, so that the prediction accuracy of the pickling state is improved, and stable and high-quality production is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of pickled vegetable production, and particularly relates to a method and an intelligent control system for an Internet of Things-driven automated pickled vegetable production line. Background Art

[0002] Currently, pickled vegetable production enterprises generally introduce automated production lines. In the automated production line, vegetable raw materials are pre-treated by washing and cutting and then pickled. In the pickling process, parameters such as temperature and humidity in fermentation tanks and pickling ponds are continuously monitored and precisely controlled based on sensors and PLC control systems, thereby shortening the traditional natural fermentation cycle and achieving automated pickling. The use of automated production lines is beneficial to reducing the downtime in the production process, continuously operating in production, and improving the production efficiency of pickled vegetables; at the same time, the automated production line greatly reduces the manual operation links, saves a large amount of labor costs, processes raw materials more precisely, effectively utilizes raw materials, and reduces production costs.

[0003] Although the automated production line of pickled vegetables has significant advantages in terms of efficiency and cost, it exposes many problems when facing diverse pickled vegetable production demands. At the raw material level, vegetables from different origins vary significantly. For example, cucumbers from Shouguang, Shandong are famous for their straight melon strips, thick pulp, stable water content, and uniform size; cucumbers from Yuanmou, Yunnan are more slender in shape, their water content fluctuates greatly affected by the local climate, and their sizes are uneven. In the standardized process of automated equipment, using a fixed ratio of seasonings to process these two types of cucumbers often results in unsatisfactory results. Due to the different water contents of the above two types of cucumbers, the same strain type, strain quantity, and fermentation time will cause the cucumbers from Yunnan to be too salty or under-pickled, unable to ensure the stability of product quality and the consistency of flavor.

[0004] Secondly, the production requirements for different types of pickled vegetables vary greatly. Taking Sichuan pickles and Northeast sauerkraut as examples, their pickling processes are different. Sichuan pickles pursue a crispy texture and a unique sour and spicy taste. In a relatively low temperature environment of 15 - 20°C, a short pickling time is sufficient, usually it can be eaten in 3 - 5 days; Northeast sauerkraut needs to be fermented by lactic acid bacteria at a low temperature of 5 - 10°C for about 1 month to achieve the best flavor. It is very difficult for automated equipment to simultaneously meet different requirements for temperature, humidity, salinity, and pickling time. Moreover, during the fermentation process, the fermentation speed of Sichuan pickles is fast, and the heat generated will increase the local environmental temperature; the fermentation of Northeast sauerkraut is slow, and the temperature and humidity change relatively gently; these factors also affect the temperature measurement in automated equipment to a certain extent. It is very important to realize the production of multiple types of pickled vegetables in the same set of automated equipment on the premise of minimizing modification and debugging work.

[0005] Finally, the pickled vegetable industry lacks a unified quality standard, and enterprises mostly rely on traditional experience to adjust technical parameters. For example, an enterprise that produces various pickled vegetables plans to produce Beijing pickled cucumbers and Yangzhou pickled vegetables on the same set of automated equipment. Beijing pickled cucumbers emphasize a strong sauce flavor and a crispy texture; Yangzhou pickled vegetables, on the other hand, pursue a sweet, fresh, and refreshing taste with a moderate saltiness. Due to the lack of a unified standard, the enterprise can only rely on traditional experience and continuously try to adjust the fermentation parameters on the automated equipment. In actual production, due to the limitations of the automated equipment itself, it is difficult to accurately simulate the control of various subtle factors during traditional manual production, resulting in a significant difference between the flavor of the produced products and the traditional flavor, and unable to effectively improve the flavor of pickled vegetables and enhance the quality of pickled vegetables. Summary of the Invention

[0006] The purpose of the present invention is to provide a method and an intelligent control system for an Internet of Things-driven automated pickled vegetable production line, aiming to automatically adjust pickling formulas, microbial fermentation, and pickling processes through data analysis and intelligent algorithms; at the same time, provide a quality evaluation system for pickled vegetables to manage the pickling process based on the refined quality of pickled vegetables and ensure the best pickling effect.

[0007] The technical solution proposed by the present invention is a method for an Internet of Things-driven automated pickled vegetable production line, including the following steps:

[0008] S1: The raw material module preprocesses the raw materials required for pickling pickled vegetables to obtain the preprocessed raw materials, and determines the type of strain and the volume of bacterial suspension required for pickling according to the type, volume, and weight of the preprocessed raw materials;

[0009] S2: In the bacterial suspension collection module, collect the medium temperature, calculate and monitor the growth rate of the strain. When the strain grows mature, collect the bacterial suspension and record the volume of the collected bacterial suspension and the type of strain;

[0010] S21: Collect the medium temperature through a temperature sensor and calculate the growth rate of the strain:

[0011] grow t =(temp t -temp t-1 )·ρ·vincu·c ρ / (η·bacx t ·ΔH)

[0012] where grow t represents the growth rate of the strain at time t, temp t represents the medium temperature at time t, temp t-1 represents the medium temperature at time t - 1, ρ represents the air density, vincu represents the volume of the strain growth environment, c ρrepresents the specific heat capacity of air at constant pressure, η represents the heat utilization rate, bacx t represents the strain concentration at time t, and ΔH represents the heat generation coefficient per unit strain;

[0013]

[0014] wherein, represents the integration of the strain growth rate within the range of [0, t - 1];

[0015] S22: When the growth rate of the strain satisfies ∫grow t dt ≥ Grva, the strain grows to maturity, where Grva represents the growth value of the mature strain;

[0016] Collect the bacterial suspension through the bacterial suspension collection module, and record the volume Vstrain of the bacterial suspension and the strain type Tstrain. Among them, Vstrain represents the volume of the bacterial suspension that can be collected each time the strain grows to maturity;

[0017] S3: In the culture medium heating module, for the strain type required in step S1, heat the culture medium of the corresponding strain according to the difference between the required volume Vneed of the bacterial suspension and the volume Vstrain of the bacterial suspension collected in step S2;

[0018] S4: The pickling module is responsible for mixing and pickling the collected bacterial suspension with the pretreated raw materials, and predicting the pickling state based on the long short-term memory network model to obtain the pickled vegetables after pickling;

[0019] S5: Set the quality evaluation index set for the pickled vegetables, evaluate the quality of the pickled vegetables after pickling, and fine-tune the long short-term memory network model in step S4 according to the overall score of the pickled vegetables.

[0020] Optionally, S3 includes:

[0021] S31: Heat the culture medium using a heating rod, and randomly set the initial power Heat of the heating rod 0 ;

[0022] S32: Calculate the loss function J(Heat) of the heating rod power: J(Heat) = (Vneed - Vstrain) 2 ; where Heat represents the power of the heating rod;

[0023] S33: Calculate the gradient Gra(Heat) of the heating rod power: where represents the change rate of the loss function J(Heat) with respect to the power of the heating rod;

[0024] Adjust the power of the heating rod:

[0025] Heat′ = Heat - α·Gra(Heat)

[0026] Among them, Heat′ represents the power adjustment value of the heating rod, and α represents the learning rate;

[0027] S34: Measure the volume Vstrain′ of the bacterial suspension collected by the current bacterial suspension collection module;

[0028] Let Vstrain = Vstrain′, and repeatedly call step S32 and step S33 until J(Heat) < ε, where ε represents the measurement error of the bacterial suspension volume.

[0029] Optionally, the S4 includes:

[0030] S41: During the pickling process, record the pickled vegetable temperature, pickled vegetable PH value, and pickled vegetable nitrite content at fixed time intervals, and construct a feature vector:

[0031] pick n = [thermal n ,phvalue n ,nitrite n

[0032] Among them, pick n represents the feature vector, n represents the recording time step, thermal n represents the pickled vegetable temperature, phvalue n represents the pickled vegetable PH value, nitrite n represents the pickled vegetable nitrite content;

[0033] S42: Use the feature vector to construct a time series {pick n , n = 1, 2,..., m}, m represents the maximum time step, input the time series into the long short-term memory network model for pickling state prediction:

[0034] y n = δ(W y h n + b y )

[0035] Among them, y n = 1 indicates that pickling is completed, y n ≠ 1 indicates that pickling is not completed, W y represents the weight matrix of the output layer, b y represents the bias vector.

[0036] Optionally, the S5 includes:

[0037] ​S51: Set the quality evaluation index set of pickled vegetables judge = {x 1 , x 2 , x 3}, where x 1 represents taste, x 2 represents color, x 3 represents smell;

[0038] For taste x 1 , set the fuzzy subset set A 1 = {A 11 , A 12 , A 13}, where A 11 represents "light", A 12 represents "moderate", A 13 represents "strong";

[0039] For color x 2 , set the fuzzy subset set A 2 = {A 21 , A 22 , A 23}, where A 21 represents "dim", A 22 represents "normal", A 23 represents "bright";

[0040] For smell x 3 , set the fuzzy subset set A 3 = {A 31 , A 32 , A 33}, where A 31 represents "weak", A 32 represents "moderate", A 33 represents "strong";

[0041] S52: Set up an evaluation group, score according to the quality evaluation index set, and obtain the average score of pickled vegetables on the evaluation index;

[0042] S53: Calculate the overall score of pickled vegetables:

[0043]

[0044] where total represents the overall score of pickled vegetables, wei 1 , wei 2 , wei 3 represent the weights of taste, color, and smell respectively; j = 1, 2, 3; represents the average score of the taste of pickled vegetables, represents the average score of the color of pickled vegetables, Represents the average score of the smell of pickled vegetables;

[0045] If the overall score of the pickled vegetables satisfies |total - 1| ≤ ζ, it means that this batch of pickled vegetables meets the standard; where |total - 1| represents taking the absolute value of total - 1, and ζ represents the scoring error;

[0046] If the overall score of the pickled vegetables satisfies |total - 1| > ζ, it means that this batch of pickled vegetables does not meet the standard, and the long - short - term memory network model in step S4 is strengthened and trained, and the weight matrix W of the output layer is fine - tuned y .

[0047] Optionally, the S52 includes:

[0048] The evaluation group consists of k evaluators, and the evaluation group is:

[0049] E = {evalu 1 , evalu 2 , …, evalu i , …, evalu k}; where evalu i represents the i - th evaluator; i = 1, 2, …, k; construct a scoring matrix:

[0050]

[0051] where Score represents the scoring matrix, and s ij represents the score given by the evaluator evalu i to the pickled vegetables on the evaluation index x j , j = 1, 2, 3;

[0052] Calculate the average score of the pickled vegetables on the evaluation index:

[0053]

[0054] where, represents the average score of the taste of the pickled vegetables, represents the average score of the color and luster of the pickled vegetables, represents the average score of the smell of the pickled vegetables, represents the sum of s 1j , s 2j , …, s kj .

[0055] The present invention also provides an intelligent control system for an Internet - of - Things - driven automated production line of pickled vegetables, including:

[0056] Raw material module: Pretreat the raw materials required for pickled vegetables to obtain the pretreated raw materials, collect the type, volume and weight data of the pretreated raw materials, and determine the required strain type and the volume of the bacterial suspension;

[0057] Bacterial suspension collection module: Collect the medium temperature, calculate and monitor the growth rate of the strain. When the strain grows mature, collect the bacterial suspension and record the volume of the collected bacterial suspension and the strain type;

[0058] Medium heating module: Heat the medium according to the difference between the required volume of the bacterial suspension and the collected volume of the bacterial suspension;

[0059] Pickling module: Mix the bacterial suspension and the pretreated raw materials, regularly record the temperature of the pickled vegetables, the pH value of the pickled vegetables and the nitrite content of the pickled vegetables, construct a feature vector, construct a time series, and predict the pickling state based on the long short-term memory network model;

[0060] Quality evaluation module: Set the set of quality evaluation indicators for pickled vegetables, set the set of fuzzy subsets; Set up an evaluation group, score according to the quality evaluation indicators, calculate the overall score of the pickled vegetables and strengthen the training of the long short-term memory network model, and fine-tune the weight matrix.

[0061] Beneficial effects:

[0062] The method and intelligent control system for the Internet of Things-driven automated production line of pickled vegetables involved in the present invention utilize the characteristic that the growth of the strain generates heat, monitor the medium temperature in real time, convert the measurement of the medium temperature into the measurement of the growth rate of the strain, and realize the effective monitoring of the strain growth; When the strain grows mature, the strain concentration in the collected bacterial suspension reaches a unified standard; The bacterial suspension collected by this method is suitable for application in the automated production process of various pickled vegetables;

[0063] Based on the difference between the demand and the collection amount of the bacterial suspension, the present invention heats the corresponding medium, which is beneficial to the rapid collection of the bacterial suspension; During the heating process of the medium, according to the collected volume of the bacterial suspension, the power of the heating rod is gradually iteratively optimized to achieve accurate control of the power of the heating rod; During the pickling process, the present invention adopts a long short-term memory network model to predict the continuity of the pickling state based on the temperature of the pickled vegetables, the pH value of the pickled vegetables, and the nitrite content of the pickled vegetables, which is beneficial to obtaining pickled vegetables in time;

[0064] The present invention constructs a set of quality evaluation indicators for pickled vegetables and realizes an objective and comprehensive evaluation of the quality of pickled vegetables by constructing a scoring matrix; Modifying the long short-term memory network model based on the quality of pickled vegetables is beneficial to improving the prediction accuracy of the pickling state and realizing stable and high-quality production. Description of the drawings

[0065] Figure 1Schematic flow chart of a method for an Internet of Things-driven automated pickled vegetable production line provided by an embodiment of the present invention;

[0066] Figure 2 Relationship diagram of the raw material module, the bacterial suspension collection module and the pickling module in the present invention;

[0067] Figure 3 Graph of the change in the volume of the bacterial suspension collected during the growth of the strain. Detailed implementation manners

[0068] The present invention will be further described below with reference to the accompanying drawings, but the present invention is not limited in any way. Any transformation or replacement based on the teachings of the present invention falls within the protection scope of the present invention.

[0069] Example 1:

[0070] A method for an Internet of Things-driven automated pickled vegetable production line, as Figures 1 - 3 shown, includes the following steps:

[0071] S1: The raw material module preprocesses the raw materials required for pickled vegetables to obtain the preprocessed raw materials, and determines the type of strain and the volume of the bacterial suspension required for pickling according to the type, volume and weight of the preprocessed raw materials:

[0072] The raw material module preprocesses the raw materials required for pickled vegetables, removes impurities in the raw materials by using a sieve and a roller, removes sediment in the raw materials by using a shower, and selects the raw materials by using machine vision equipment to obtain pure raw materials to be pickled.

[0073] S2: In the bacterial suspension collection module, collect the temperature of the culture medium, calculate and monitor the growth rate of the strain, collect the bacterial suspension when the strain grows mature, and record the volume of the collected bacterial suspension and the type of strain:

[0074] S21: Collect the temperature of the culture medium through a temperature sensor and calculate the growth rate of the strain:

[0075] grow t =(temp t -temp t-1 )·ρ·vincu·c ρ / (η·bacx t ·ΔH)

[0076] where grow t represents the growth rate of the strain at time t, temp t represents the temperature of the culture medium at time t, temp t-1 represents the temperature of the culture medium at time t - 1, ρ represents the air density, vincu represents the volume of the strain growth environment, c ρIt represents the specific heat capacity of air at constant pressure, η represents the heat utilization rate, and bacx t It represents the strain concentration at time t, and ΔH represents the heat production coefficient per unit strain;

[0077]

[0078] Among them, It represents the integral of the strain growth rate within the range of [0, t - 1];

[0079] S22: When the growth rate of the strain satisfies ∫grow t dt ≥ Grva, the strain grows to maturity, where Grva represents the growth value of the mature strain;

[0080] Collect the bacterial suspension through the bacterial suspension collection module, and record the volume Vstrain of the bacterial suspension and the strain type Tstrain. Among them, Vstrain represents the volume of the bacterial suspension that can be collected each time the strain grows to maturity.

[0081] In the embodiments of the present invention, the heat production coefficient ΔH per unit strain is related to the strain type and fermentation process parameters (stirring rate, ventilation volume); the growth value Grva of the mature strain is related to the strain type and is determined according to experiments;

[0082] S3: In the culture medium heating module, for the strain type required in step S1, heat the culture medium of the corresponding strain according to the difference between the required volume Vneed of the bacterial suspension and the volume Vstrain of the bacterial suspension collected in step S2:

[0083] S31: Heat the culture medium using a heating rod, and randomly set the initial power Heat of the heating rod 0 ;

[0084] S32: Calculate the loss function J(Heat) of the heating rod power: J(Heat) = (Vneed - Vstrain) 2 ; where Heat represents the power of the heating rod;

[0085] S33: Calculate the gradient Gra(Heat) of the heating rod power: Among them It represents the change rate of the loss function J(Heat) with respect to the heating rod power;

[0086] Adjust the power of the heating rod:

[0087] Heat′ = Heat - α·Gra(Heat)

[0088] Among them, Heat′ represents the power adjustment value of the heating rod, and α represents the learning rate;

[0089] S34: Measure the volume Vstrain' of the bacterial suspension collected by the current bacterial suspension collection module;

[0090] Let Vstrain = Vstrain', and repeatedly call steps S32 and S33 until J(Heat) < ε, where ε represents the measurement error of the bacterial suspension volume.

[0091] S4: The pickling module is responsible for mixing the collected bacterial suspension with the pretreated raw materials, performing pickling, and predicting the pickling state based on the long short-term memory network model to obtain the pickled vegetables after pickling:

[0092] S41: During the pickling process, record the temperature, pH value, and nitrite content of the pickled vegetables at fixed time intervals, and construct a feature vector:

[0093] pick n =[thermal n ,phvalue n ,nitrite n

[0094] where pick n represents the feature vector, n represents the recording time step, thermal n represents the temperature of the pickled vegetables, phvalue n represents the pH value of the pickled vegetables, and nitrite n represents the nitrite content of the pickled vegetables;

[0095] S42: Use the feature vector to construct a time series {pick n , n = 1, 2,..., m}, where m represents the maximum time step, and input the time series into the long short-term memory network model for pickling state prediction:

[0096] y n =δ(W y h n +b y )

[0097] where y n =1 indicates that pickling is completed, y n ≠1 indicates that pickling is not completed, W y represents the weight matrix of the output layer, and b y represents the bias vector.

[0098] S5: Set the quality evaluation index set for the pickled vegetables. For the pickled vegetables after pickling, perform quality evaluation, and fine-tune the long short-term memory network model in step S4 according to the overall score of the pickled vegetables:

[0099] ​S51: Set the quality evaluation index set of pickled vegetables judge = {x 1 , x 2 , x 3}, where x 1 represents taste, x 2 represents color, and x 3 represents smell;

[0100] For taste x 1 , set the fuzzy subset set A 1 = {A 11 , A 12 , A 13}, where A 11 represents "light", A 12 represents "moderate", and A 13 represents "strong";

[0101] For color x 2 , set the fuzzy subset set A 2 = {A 21 , A 22 , A 23}, where A 21 represents "dim", A 22 represents "normal", and A 23 represents "bright";

[0102] For smell x 3 , set the fuzzy subset set A 3 = {A 31 , A 32 , A 33}, where A 31 represents "weak", A 32 represents "moderate", and A 33 represents "strong";

[0103] S52: Set up an evaluation group, score according to the quality evaluation index set, and obtain the average score of pickled vegetables on the evaluation index:

[0104] The evaluation group is expressed as E = {evalu 1 , evalu 2 , …, evalu i , …, evalu k}, where the evaluation group consists of k evaluators, and evalu i represents the i-th evaluator, i = 1, 2, …, k;

[0105] Construct a scoring matrix:

[0106]

[0107] Among them, Score represents the scoring matrix, and s ij represents the evaluator evalu i 's score for the pickled vegetables on the evaluation index x j , where i = 1, 2,..., k and j = 1, 2, 3;

[0108] Calculate the average score of the pickled vegetables on the evaluation index:

[0109]

[0110] Among them, represents the average score of the taste of the pickled vegetables, represents the average score of the color of the pickled vegetables, represents the average score of the smell of the pickled vegetables, represents the sum of s 1j , s 2j , …, s kj ;

[0111] S53: Calculate the overall score of the pickled vegetables:

[0112]

[0113] Among them, total represents the overall score of the pickled vegetables, and wei 1 , wei 2 , wei 3 represent the weights of taste, color, and smell respectively; j = 1, 2, 3; represents the average score of the taste of the pickled vegetables, represents the average score of the color of the pickled vegetables, represents the average score of the smell of the pickled vegetables;

[0114] If the overall score of the pickled vegetables satisfies |total - 1| ≤ ζ, it means that this batch of pickled vegetables meets the standard;

[0115] Among them, |total - 1| represents taking the absolute value of total - 1, and ζ represents the scoring error;

[0116] If the overall score of the pickled vegetables satisfies |total - 1| > ζ, it means that this batch of pickled vegetables does not meet the standard, and the long - short - term memory network model in step S4 is strengthened and trained, and the weight matrix W of the output layer is fine - tuned y .

[0117] It should be noted that in the embodiments of the present invention, the raw material types are divided into root and rhizome plants and leaf plants; for different raw material types, the volume and weight of the raw materials have different weights; according to the volume, weight, volume weight coefficient, and weight weight coefficient of the raw materials, the raw material coefficient is calculated; according to the raw material type and the raw material coefficient, the condiment configuration is carried out to determine the strain type and the volume of the bacterial suspension required for pickling; through multiple experiments by the experimenters, the raw material weight weight coefficient and the raw material volume weight coefficient in the case of different raw material types, and the condiment configuration standard applicable to the automated production line in the case of different raw material coefficients are determined.

[0118] For root and rhizome plants, the weight of the raw material has a high weight coefficient; for leaf plants, the volume of the raw material has a high weight coefficient; taking pickled vegetables with Chinese cabbage as the raw material and pickled vegetables with radish as the raw material as examples, the following is the calculation process of the strain type and the volume of the bacterial suspension required for pickling:

[0119] Weight Volume Weight Weighting Volume Weighting Chinese Cabbage 1000 grams 2000 cubic centimeters 0.4 0.6 Radish 1500 grams 1000 cubic centimeters 0.7 0.3

[0120] For pickled vegetables with Chinese cabbage as the raw material, the raw material coefficient = (1000 × 0.4) + (2000 × 0.6) = 1600;

[0121] For pickled vegetables with radish as the raw material, the raw material coefficient = (1500 × 0.7) + (1000 × 0.3) = 1350;

[0122] The strain types required for different raw materials are different, and the amount of strain used is proportional to the raw material coefficient; when the strain is mature, the concentration of the bacterial suspension is consistent, so the volume of the bacterial suspension required is proportional to the raw material coefficient.

[0123] For pickled vegetables with Chinese cabbage as the raw material, the strain type required is lactic acid bacteria, and the volume of the bacterial suspension required is 16 ml;

[0124] For pickled vegetables with radish as the raw material, the strain type required is acetic acid bacteria, and the volume of the bacterial suspension required is 13.5 ml;

[0125] In actual production, the weight weight and volume weight values of different raw materials are determined according to the laboratory experiment results; the strain types required for different types of raw materials are determined according to the laboratory experiment results; the proportional relationship between the raw material coefficient and the bacterial suspension is determined according to the laboratory experiment results.

[0126] In the embodiments of the present invention, the growth rate of the strain is calculated by collecting the temperature of the culture medium. In a microbial fermentation culture system, when the strain grows and metabolizes in the culture medium, a series of biochemical reactions will occur. These reactions belong to the heat generation process, causing the temperature of the culture medium to rise. At the same time, due to the ventilation conditions in the environment where the culture medium is located, heat will be dissipated to the surrounding environment through heat convection, heat conduction, etc. On the premise of knowing the heat utilization rate, according to the law of conservation of energy, that is, the total energy in the system remains constant, the input energy is equal to the sum of the output energy and the change in energy in the system. In this context, the input energy is the heat generated by the growth and metabolism of the strain, the output energy is the heat dissipated by ventilation, and the change in energy in the system is reflected in the change in the temperature of the culture medium. By accurately measuring the temperature change of the culture medium within a specific time period, combining relevant parameters such as the heat utilization rate and the specific heat capacity of the culture medium, and using mathematical models and physical formulas for calculation and derivation, the growth rate of the strain can be roughly estimated.

[0127] As a key parameter, the heat utilization rate η can be obtained by means of an experimental measurement process under controlled conditions. In a completely consistent experimental environment setting, including but not limited to the same culture medium composition and ratio, equivalent strain inoculation amount, consistent culture temperature, humidity, ventilation rate and other environmental factors, as well as the same culture vessel specifications and materials. Conduct multiple repeated culture experiments on the target strain, and accurately measure and record key data such as the heat production during the growth and metabolism of the strain, the heat absorption of the culture medium, and the heat dissipated to the environment in each experiment. Through systematic analysis and mathematical statistics processing of multiple groups of experimental data, using methods such as least squares fitting, average value calculation combined with standard deviation evaluation, etc., to reduce experimental errors and random factor interference, so as to accurately measure this important parameter of the heat utilization rate.

[0128] In the embodiments of the present invention, during the growth process of the strain, the change in the volume of the bacterial suspension collected is analyzed. The change law of the volume of the bacterial suspension with time is studied under two conditions: heating and not heating the strain culture medium, and then the influence of heating on the collection efficiency of the bacterial suspension is analyzed. Lactic acid bacteria are selected as the experimental object, and two experimental groups are set up, namely the heating group and the non-heating group. Each group of experiments is set with three biological replicates to improve the reliability of the experimental results. The non-heating group places the culture container containing the culture medium and the strain in a constant temperature incubator, keeping the ventilation conditions constant to simulate the natural heat dissipation environment; the heating group, on the basis of the non-heating group, uses a heating rod to heat the culture medium, and also keeps the ventilation conditions. At fixed time intervals (set to 0.5 hours in this experiment), accurately suck the bacterial suspension from each culture container, measure the volume of the bacterial suspension with a pipette, and record the data. Figure 3The figure shows the change in the volume of the bacterial suspension. It can be seen from the figure that the volume of the bacterial suspension in the non-heated group increases step by step with time. Within each time interval of 0.5 hours, the increase in the volume of the bacterial suspension is roughly the same. The volume of the bacterial suspension in the heated group also shows a step-by-step increase, but the volume increase within each time interval gradually increases. Thus, it can be seen that heating has an obvious promoting effect on the growth of the strain and the increase in the volume of the bacterial suspension. Under the condition of not heating, the growth of the strain is relatively stable, probably at a relatively basic growth rate. Limited by the environmental temperature, its metabolic activities are relatively stable, resulting in the volume of the bacterial suspension increasing at a relatively uniform rate. Under the heating condition, the appropriate temperature increase may promote the metabolism of the strain, accelerating the cell division speed, thus leading to a gradual increase in the volume increase of the bacterial suspension within each time interval.

[0129] In the embodiment of the present invention, in the long short-term memory network model, the temperature of the pickled vegetables, the pH value of the pickled vegetables, and the nitrite content of the pickled vegetables are selected to construct a feature vector. The determination of the above three features is derived from the measurement experiment of the fermentation state of the existing pickled vegetables by the experimenters (i.e., the pickling state in this embodiment), and can comprehensively and accurately reflect the fermentation state and quality characteristics of the pickled vegetables.

[0130] In the embodiment of the present invention, the fermentation state predicted by the long short-term memory network model can be regarded as the standard fermentation state. Considering the diversity and uniqueness of regional food cultures, there are slight differences in the flavor preferences of pickled vegetables among consumers in different regions. To make pickled vegetable products better meet the taste preferences of consumers in different regions, based on the fermentation state predicted by the long short-term memory network model, the present invention introduces an evaluation group to score the pickled vegetables. The evaluation group consists of professional food tasting personnel and representatives of typical local consumers. They evaluate the pickled vegetables from the aspects of taste, color, and smell according to the evaluation criteria and give corresponding scores. According to the overall score of the pickled vegetables, the long short-term memory network model is fine-tuned. This fine-tuning aims to optimize the prediction results of the model, enabling it not only to accurately predict the standard fermentation state but also to fully consider regional flavor differences, achieving precise control of the fermentation process, and thus producing pickled vegetable products that better meet the taste preferences of local consumers. For example, consumers in Shanghai generally prefer pickled vegetables with a sweet flavor. For the fermentation of lactic acid bacteria, as the fermentation time prolongs, the acidity will gradually increase, and some other flavor substances will also be produced. To obtain a sweeter taste, it is necessary to appropriately shorten the fermentation time of lactic acid bacteria. A shorter fermentation time can reduce the amount of sugars converted into acids, enabling the pickled vegetables to retain more sweet components. For a pickled vegetable with a conventional fermentation time of 7 days, when producing for the Shanghai market, the prediction of the fermentation state can be advanced. This can not only ensure a certain degree of fermentation, making the pickled vegetables have rich flavor layers, but also highlight the sweet taste, better meeting the taste preferences of Shanghai consumers.

[0131] Embodiment 2: The present invention also provides an intelligent control system for an Internet of Things-driven automated production line for pickled vegetables, including the following five modules:

[0132] Raw material module: Pretreat the raw materials required for pickled vegetables to obtain the pretreated raw materials, collect the type, volume, and weight data of the pretreated raw materials, and determine the required strain type and the volume of the bacterial suspension.

[0133] Bacterial suspension collection module: Collect the temperature of the culture medium, calculate and monitor the growth rate of the strain. When the strain grows to maturity, collect the bacterial suspension and record the volume of the collected bacterial suspension and the strain type.

[0134] Culture medium heating module: Heat the culture medium according to the difference between the required volume of the bacterial suspension and the volume of the collected bacterial suspension.

[0135] Pickling module: Mix the bacterial suspension and the pretreated raw materials, regularly record the temperature, pH value, and nitrite content of the pickled vegetables, construct a feature vector, construct a time series, predict the pickling state based on the long short-term memory network model, and obtain the pickled vegetables that are pickled to completion.

[0136] Quality evaluation module: Set a set of quality evaluation indicators for pickled vegetables, set a set of fuzzy subsets; set up an evaluation group, score according to the quality evaluation indicators, calculate the overall score of the pickled vegetables, and perform reinforcement training on the long short-term memory network model and fine-tune the weight matrix.

[0137] It should be noted that the serial numbers of the embodiments of the present invention above are only for description and do not represent the superiority or inferiority of the embodiments. And the term "including", "comprising" or any other variant thereof in this article is intended to cover a non-exclusive inclusion, so that a process, device, article or method including a series of elements not only includes those elements but also includes other elements not expressly listed, or also includes elements inherent to such a process, device, article or method. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of another identical element in the process, device, article or method including that element.

[0138] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium as described above (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions for causing a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in various embodiments of the present invention.

[0139] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A method for driving an automated pickled vegetable production line using the Internet of Things, characterized in that: The method comprises: S1: The raw material module pre-treats the raw materials required for pickling pickled vegetables to obtain pre-treated raw materials, and determines the strain type and bacterial suspension volume required for pickling according to the type, volume and weight of the pre-treated raw materials; S2: In the bacterial suspension collection module, the culture medium temperature is collected, the growth rate of the strain is calculated and monitored, the bacterial suspension is collected when the strain grows to maturity, and the collected bacterial suspension volume and strain type are recorded; S21: The temperature of the culture medium is collected through the temperature sensor to calculate the growth rate of the strain: grow t =(temp t -temp t-1 )·ρ·vincu·c ρ / (η·bacx t ·ΔH) Among them, grow t represents the growth rate of the strain at time t, temp t represents the culture medium temperature at time t, temp t-1 represents the culture medium temperature at time t-1, ρ represents the air density, vincu represents the volume of the strain growth environment, c ρ represents the specific heat capacity of air at constant pressure, η represents the heat utilization rate, bacx t represents the strain concentration at time t, ΔH represents the heat production coefficient per unit strain; in, It means to integrate the growth rate of the strain in the range of [0, t-1]; S22: When the growth rate of the strain satisfies ∫grow t When dt ≥ Grva, the strain grows to maturity, where Grva represents the growth value of the mature strain; The bacterial suspension is collected by the bacterial suspension collection module, and the bacterial suspension volume Vstrain and the strain type Tstrain are recorded, wherein Vstrain represents the volume of bacterial suspension that can be collected each time the strain grows to maturity; S3: In the culture medium heating module, for the strain type required in step S1, the culture medium of the corresponding strain is heated according to the difference between the required bacterial suspension volume Vneed and the bacterial suspension volume Vstrain collected in step S2; S4: The pickling module is responsible for mixing and pickling the collected bacterial suspension with the pretreated raw materials, and predicting the pickling state based on the long short-term memory network model to obtain the pickled vegetables; S5: Set a set of quality evaluation indicators for pickled vegetables, evaluate the quality of the pickled vegetables, and fine-tune the long short-term memory network model of step S4 according to the overall score of the pickled vegetables.

2. The method for driving the automated production line of pickled vegetables by the Internet of Things according to claim 1, characterized in that: The S3 includes: S31: Use a heating rod to heat the culture medium, and randomly set the initial power of the heating rod to Heat0; S32: Calculate the loss function J(Heat) of the heating rod power: J(Heat) = (Vneed-Vstrain) 2 , where Heat represents the power of the heating rod; S33: Calculate the gradient of the heating rod power Gra(Heat): in It represents the rate of change of the loss function J(Heat) relative to the power of the heating rod; Adjust the power of the heating rod: Heat′=Heat-α·Gra(Heat) Among them, Heat′ represents the power adjustment value of the heating rod, and α represents the learning rate; S34: measuring the volume Vstrain′ of the bacterial suspension collected by the current bacterial suspension collection module; Let Vstrain=Vstrain′, and repeat steps S32 and S33 until J(Heat)<ε, where ε represents the measurement error of the bacterial suspension volume.

3. The method for the automated pickled vegetable production line driven by the Internet of Things according to claim 1, characterized in that: The S4 includes: S41: During the pickling process, the temperature of the pickled vegetables, the pH value of the pickled vegetables, and the nitrite content of the pickled vegetables are recorded at fixed time intervals to construct a feature vector: pick n =[thermal n ,phvalue n ,nitrite n ] Among them, pick n represents the feature vector, n represents the recording time step, thermal n Indicates the temperature of pickled vegetables, phvalue n Indicates pH value of pickled vegetables, nitrite n Indicates the nitrite content of pickled vegetables; S42: Using feature vectors to construct time series {pick n ,n=1,2,…,m}, m represents the maximum time step, and the time series is input into the long short-term memory network model for pickling state prediction: y n =δ(W y h n +b y ) Among them, y n =1 means pickling is complete, y n ≠1 means pickling is not completed, W y represents the weight matrix of the output layer, b y Represents the bias vector.

4. The method for the automated pickled vegetable production line driven by the Internet of Things according to claim 3, characterized in that: The S5 includes: S51: Set the quality evaluation index set of pickled vegetables judge = {x1, x2, x3}, where x1 represents taste, x2 represents color, and x3 represents smell; For taste x1, set the fuzzy subset set A1 = {A 11 ,A 12 ,A 13 }, where A 11 A means "light". 12 Means "moderate", A 13 It means "rich"; For color x2, set the fuzzy subset set A2 = {A 21 ,A 22 ,A 23 }, where A 21 Means "dim", A 22 A means "normal", 23 It means "bright"; For smell x3, set the fuzzy subset set A3 = {A 31 ,A 32 ,A 33 }, where A 31 A means "weak". 32 Means "moderate", A 33 It means "strong"; S52: setting up an evaluation team to score based on the quality evaluation index set, and obtaining the average score of the pickled vegetables on the evaluation index; S53: Calculate the overall score of pickled dishes: Where total represents the overall score of pickled vegetables, wei1, wei2, wei3 represent the weights of taste, color and smell respectively; j = 1, 2, 3; Indicates the average taste score of pickled dishes. Indicates the average color score of pickled vegetables. Indicates the average score of the smell of pickled vegetables; If the overall score of the pickled vegetables satisfies |total-1|≤ζ, it means that the batch of pickled vegetables meets the standard; where |total-1| represents the absolute value of total-1, and ζ represents the score error; If the overall score of the pickled vegetables satisfies |total-1|>ζ, it means that the batch of pickled vegetables does not meet the standard, and the long short-term memory network model in step S4 is strengthened and the weight matrix W of the output layer is fine-tuned. y .

5. The method for the automated pickled vegetable production line driven by the Internet of Things according to claim 4, characterized in that: The S52 includes: The evaluation team is composed of k evaluators, and the evaluation team is E = {evalu1, evalu2, ..., evalu i ,…,evalu k }; where evaluate i represents the i-th evaluator, i = 1, 2, …, k; Construct the scoring matrix: Among them, Score represents the scoring matrix, s ij Indicates that the evaluator evaluates i Evaluation index x for pickled vegetables j The score on j = 1, 2, 3; Calculate the average score of pickled vegetables on the evaluation index: in, Indicates the average taste score of pickled dishes. Indicates the average color score of pickled vegetables. represents the average score of the smell of pickled vegetables, Indicates 1j ,s 2j ,…,s kj Sum.

6. An intelligent control system for an automated pickled vegetable production line driven by the Internet of Things, characterized in that: include: Raw material module: pre-process the raw materials needed for pickling pickled vegetables, obtain the pre-processed raw materials, collect the type, volume and weight data of the pre-processed raw materials, and determine the required strain type and bacterial suspension volume; Bacterial suspension collection module: collects the culture medium temperature, calculates and monitors the strain growth rate, collects the bacterial suspension when the strain grows mature, and records the collected bacterial suspension volume and strain type; Culture medium heating module: heats the culture medium according to the difference between the required bacterial suspension volume and the collected bacterial suspension volume; Pickling module: Mix the bacterial suspension and pretreated raw materials, regularly record the temperature, pH value and nitrite content of pickled vegetables, construct feature vectors, construct time series, predict the pickling state based on the long short-term memory network model, and obtain the pickled vegetables; Quality evaluation module: set the pickled vegetable quality evaluation index set and fuzzy subset set; An evaluation team was set up to score the pickled vegetables based on the quality evaluation indicators, calculate the overall score of the pickled vegetables and fine-tune the long short-term memory network model; To realize a method for an Internet of Things driven automated production line of pickled vegetables as described in any one of claims 1-5.

Images