AI-based euryale ferox processing process quality control system

Abstract

An AI-based euryale ferox processing process quality control system of the present invention belongs to the technical field of food processing and quality control, and comprises a quality perception quantification module used for carrying out real-time nondestructive detection on a material unit entering a processing flow, and determining the quality of the material unit through a pre-established chemometrics model. The system comprises a material unit acquisition module for acquiring original spectrum data, a quality dynamic evolution prediction module for analyzing and quantifying the acquired original spectrum data into an initial starch crystallinity index, and a quality dynamic evolution prediction module for predicting a quality evolution track of the material unit in future processing time through a dynamic evolution prediction model based on the initial starch crystallinity index, a current quality state and real-time process parameters. According to the method, different processing paths can be predictively planned for materials with different characteristics, so that disturbance caused by heterogeneity of the raw materials is effectively overcome.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of food processing and quality control, in particular to a gordon euryale processing process quality control system based on AI. BACKGROUND

[0002] The main challenge faced by gordon euryale deep processing industry is the consistency control of the quality of the final product. Traditional processing uses fixed process parameters, which cannot adapt to the significant quality heterogeneity caused by factors such as breeding and production place within the batch of raw materials, especially the difference in starch crystallinity. This difference directly affects the final texture characteristics of the product. The existing technology lacks the ability to identify the differences in raw materials online and adjust the process in real time, resulting in unstable product quality. Therefore, there is an urgent need in the field for an intelligent quality control method to solve this problem.

[0003] The above information disclosed in the above background section is only used to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0004] The purpose of the present application is to provide a gordon euryale processing process quality control system based on AI to solve the problems raised in the background.

[0005] The technical solution of the present application is to provide a gordon euryale processing process quality control system based on AI to solve the problems raised in the background.

[0006] The quality dynamic evolution prediction module is based on the initial starch crystallinity index, the current quality state and the real-time process parameters, and predicts the quality evolution trajectory of the material unit within the future processing time through a dynamic evolution prediction model.

[0007] The process parameter dynamic optimization module is used to compare the quality evolution trajectory with the preset ideal target quality trajectory, and calculates the optimal process parameter combination at the next time through a model predictive control optimization framework.

[0008] The closed-loop feedback control execution module is used to issue the optimal process parameter combination to the execution mechanism of the processing equipment to complete the real-time regulation and control of the processing process, and based on the quality state of the regulated material unit, a new round of quality evolution trajectory prediction, optimization and control is performed to form a closed-loop feedback.

[0009] Preferably, the quality perception and quantification module includes a near-infrared spectrum detection system deployed on the material conveying line, which is used to collect the original spectrum data of each material unit.

[0010] Preferably, the chemometrics model comprises: collecting Gorgon nut samples with different starch crystallinity; using X-ray diffraction method to determine the crystallinity value of the Gorgon nut samples, and synchronously collecting the near-infrared spectrum thereof; and using a partial least squares regression algorithm to train, taking the near-infrared spectrum as input and the crystallinity value as output.

[0011] Preferably, the dynamic evolution prediction model is composed of a quality state discrete time evolution equation and a quality instantaneous change rate constitutive model; the quality state discrete time evolution equation is used to iteratively calculate the quality state at the next time based on the current quality state and the quality instantaneous change rate.

[0012] Preferably, the quality instantaneous change rate constitutive model is used to construct a functional relationship between the quality instantaneous change rate and the real-time process parameters, the current quality state and the initial starch crystallinity index; the initial starch crystallinity index is introduced as a raw material quality correction term to represent the characteristic that the higher the initial crystallinity, the slower the raw material quality change rate.

[0013] Preferably, the model predictive control optimization framework calculates the optimal process parameter combination by solving a cost function; the cost function includes a tracking error term for minimizing the deviation between the quality evolution trajectory and the ideal target quality trajectory, and a control cost term for punishing process parameter changes.

[0014] Preferably, the control cost term is a weighted control increment penalty term, which uses the normal operating range of each process parameter to normalize the change amount of the control quantity, and adjusts the punishment degree of different control quantity changes through a weight matrix.

[0015] Preferably, the closed-loop feedback control execution module issues the optimal process parameter combination to a programmable logic controller, which drives the heater, valve or motor.

[0016] Preferably, the current quality state is obtained by real-time measurement through an online sensor deployed on the processing equipment, or when there is no online sensor, is obtained by a model-based state estimation method combined with the predicted value of the last control period and the model after correction.

[0017] The present application improves the quality control system of Gorgon nut processing process based on AI, which has the following improvements and advantages compared with the prior art:

[0018] 1. The revolutionary progress of the present application is that it refines the granularity of control from macro-batch to micro-material unit. Through the quality perception quantification module, the system no longer blindly processes all raw materials, but uses near-infrared spectroscopy detection system and pre-established chemometrics model to assign a quantitative initial starch crystallinity index to each material unit entering the processing flow , which is equivalent to issuing a unique quality ID for each material unit, enabling the system to predict its processing characteristics in advance. This is a fundamental difference from the passive quality control method of existing technologies, which only conducts spot checks after processing is completed;

[0019] 2. A core contribution of the present application is the creation of a quality dynamic evolution prediction module. The dynamic evolution prediction model within this module, especially its quality instantaneous change rate constitutive model, has profound practical significance. At the same time, the system can actively predict rather than passively respond, and can foreseeably plan different processing paths for materials with different characteristics, effectively overcoming the disturbance caused by raw material heterogeneity;

[0020] 3. Instead of using simple proportional-integral-derivative control, the present application introduces a more advanced model predictive control optimization framework through the process parameter dynamic optimization module, replacing the traditional reliance on trial and error and adjustment based on human experience, and transforming into an optimal pathfinding process that can prospectively plan a series of future control actions, making process control more accurate, stable, and efficient;

[0021] 4. Through the closed-loop feedback control execution module, the calculated optimal process parameters are sent to the underlying execution mechanism, and the regulated material quality state is used as the starting point for the next cycle, forming a complete cyber-physical closed loop. Compared to the open-loop, static control of existing technologies, the closed-loop system of the present application can suppress various unmodeled disturbances in real time, continuously stabilizing the processing process on the optimal path. BRIEF DESCRIPTION OF DRAWINGS

[0022] The present application will be further explained in conjunction with the accompanying drawings and examples:

[0023] Figure 1 is a flowchart of the system of the present application. DETAILED DESCRIPTION

[0024] To make the purpose, technical solutions and advantages of the present application clearer and more apparent, the present application will be further described in detail below in conjunction with specific examples.

[0025] Example 1:

[0026] Please refer to Figure 1The application provides an AI-based processing process quality control system for gualou, which comprises a quality perception and quantification module, a quality dynamic evolution prediction module, a process parameter dynamic optimization module and a closed-loop feedback control execution module.

[0027] The quality perception and quantification module is used for real-time nondestructive detection of material units entering the processing flow, and the collected original spectrum data are analyzed and quantified into an initial starch crystallinity index through a pre-established chemometrics model.

[0028] The quality dynamic evolution prediction module is used for predicting the quality evolution trajectory of the material units in future processing time through a dynamic evolution prediction model based on the initial starch crystallinity index, the current quality state and real-time process parameters.

[0029] The process parameter dynamic optimization module is used for comparing the quality evolution trajectory with a preset ideal target quality trajectory, and calculating the optimal process parameter combination at the next moment through a model predictive control optimization framework.

[0030] The closed-loop feedback control execution module is used for issuing the optimal process parameter combination to the execution mechanism of the processing equipment to complete real-time regulation and control of the processing process, and performing a new round of quality evolution trajectory prediction, optimization and control based on the quality state of the regulated material units, thereby forming a closed-loop feedback. The AI-based processing process quality control system for gualou is used for overcoming the technical problem of unstable product quality caused by the heterogeneity of raw materials.

[0031] Embodiment 2

[0032] The quality perception and quantification module comprises a near-infrared spectrum detection system arranged on a material conveying line, and the near-infrared spectrum detection system is used for collecting original spectrum data of each material unit.

[0033] The chemometrics model includes: collecting Gorgon nut samples with different starch crystallinity; using X-ray diffraction method to determine the crystallinity value of Gorgon nut samples, and synchronously collecting the near-infrared spectrum thereof; using a partial least squares regression algorithm to train, taking the near-infrared spectrum as input and the crystallinity value as output.

[0034] In the embodiment, the core of the quality perception quantification module is a near-infrared spectrum (NIRS) detection system deployed on the Gorgon nut slurry pipeline. The system performs non-invasive and high-throughput real-time scanning on each material unit flowing through, and captures its unique raw spectral data . The fundamental purpose of this is to provide the most essential initial state information of each material unit for subsequent precise control.

[0035] In order to convert high-dimensional spectral data into a single index with clear physical meaning, the quality perception quantification module uses a pre-constructed chemometrics model . The model construction process has scientific rigor: researchers collect a large number of Gorgon nut samples covering different origins, varieties and batches, and use X-ray diffraction (XRD), a recognized gold standard method, to accurately determine the crystallinity value of starch. At the same time of determining the crystallinity, the NIRS spectrum data of the corresponding sample is synchronously collected. Based on this, the collected multiple groups of spectral data are taken as input variables X, and the XRD-determined crystallinity value is taken as output variable Y. The partial least squares regression (PLS-R) algorithm is used for fitting training, and finally a calibration model capable of accurately inverting the crystallinity from the spectral data is obtained .

[0036] In the actual operation of the system, the module parses the real-time collected spectral data into the initial starch crystallinity index by the following formula :

[0037]

[0038] wherein, is a dimensionless relative index, the value of which directly reflects the ordering degree of the starch structure in the material unit, and further represents the processing difficulty, becoming the first key button connecting the heterogeneity of raw materials and the self-adaptive control of the process. This method enables the system to penetrate the appearance of the raw material appearance and understand the difference in its inherent quality, providing a solid data foundation for realizing quality control from the source.

[0039] Embodiment 3

[0040] The dynamic evolution prediction model is composed of a quality state discrete-time evolution equation and a quality instantaneous change rate constitutive model; the quality state discrete-time evolution equation is used for iteratively calculating the quality state at the next time based on the current quality state and the quality instantaneous change rate.

[0041] The quality instantaneous change rate constitutive model is used for constructing a functional relationship between the quality instantaneous change rate and real-time process parameters, the current quality state and an initial starch crystallinity index; the initial starch crystallinity index is introduced as a raw material quality correction term to represent the characteristic that the higher the initial crystallinity of the raw material, the slower the change rate of the quality of the raw material.

[0042] In the implementation of the embodiment, the quality dynamic evolution prediction module constructs a dynamic evolution prediction model capable of accurately depicting the quality evolution trajectory of the semen trapa material in the processing process; a key technical feature is that the model is not a single black box model, but is composed of two sub-models with clear physical connotations and mutual coupling: a macroscopic quality state discrete-time evolution equation and a microscopic quality instantaneous change rate constitutive model; the two models work together to completely describe the whole process dynamics from the raw material to the finished product.

[0043] The first component is the quality state discrete-time evolution equation; this equation is derived from the forward Euler method for solving ordinary differential equations in numerical analysis, and the technical motivation is to cleverly decompose the complex and nonlinear continuous quality evolution process into a series of linear changes in a very small time step , so that the computer can predict and simulate the long-term evolution trajectory of the material quality in the whole processing period through simple iterative operations; the equation constitutes the iterative framework of quality prediction, and in each control period, it recursively calculates the state at the next time using the current state and the instantaneous change rate, and gradually advances, providing the predicted value of the future state for model predictive control, which is the basis for forward-looking control, and the expression is:

[0044]

[0045] wherein, is the material quality state at the next time of the model prediction, which is the output of the equation; is the material quality state at the current time, which is the input of the model; is a preset time step for control and calculation of the system, which is a constant, and its selection needs to comprehensively consider the system computing power and process dynamic characteristics, and generally should be less than the main time constant of the controlled process to ensure the stability of numerical integration and the timeliness of control, and the specific value can be determined through simulation or field test, and its dimension is time ; is the quality instantaneous change rate at the current time, and its dimension is time inverse , which is not an independent parameter itself, but a bridge connecting macroscopic and microscopic models, and its value is calculated by the constitutive model described below;

[0046] The constitutive model is one of the core innovations of the present application, which is a quality instantaneous change rate constitutive model. This constitutive model deeply integrates scientific principles from multiple fields, and its form is inspired by the Arrhenius equation in chemical reaction kinetics, rheological theory and first-order reaction rate model. The technical motivation is to create a compact and clear mathematical expression that can capture the comprehensive influence of thermal coupling effects, the current state of materials and the most critical initial properties of raw materials on processing rate. The model receives real-time process parameters from sensors and raw material characteristic parameters from Example 2, and calculates the instantaneous change rate that drives the evolution of quality. It is the presence of the raw material quality correction term that enables the entire dynamic evolution prediction model to respond to the heterogeneity of raw materials. The prediction result is no longer uniform, but an individualized evolution trajectory for each material unit. The expression is:

[0047]

[0048] where the real-time process parameter vector is the control input of the model, including the effective processing temperature , the dimension of which is temperature and the equivalent mechanical force , which can be pressure or shear rate ;

[0049] For example, in a cooking pot with a stirrer, the equivalent mechanical force can mainly be represented as the average shear rate ; this value can be estimated by the following industry-recognized formula:

[0050]

[0051] where is the stirrer speed in revolutions per second, and is a dimensionless constant related to the geometry of the stirrer, such as diameter, blade type, which can be obtained by consulting relevant chemical engineering manuals or through fluid mechanics simulation;

[0052] is a material-specific property calibrated through experiments, representing the critical reference temperature at which starch quality begins to change significantly, with a dimension of temperature ; where is the temperature sensitivity coefficient, with a dimension of ; is the mechanical force sensitivity coefficient; when the equivalent mechanical force in terms of pressure, dimension measured in terms of pressure, dimension measured in terms of pressure, dimension ; when in terms of shear rate, dimension measured in terms of shear rate, dimension are dimensionless coefficients; their values are obtained by fitting training on a large amount of historical processing data; is a dimensionless state saturation term, which ensures that when the material quality tends to the ideal state (i.e. ), its change rate naturally slows down and tends to zero, which fully complies with the physical law; the crucial innovation lies in the dimensionless raw material quality correction term ; this term introduces the initial starch crystallinity index obtained in Example 2 as a key variable into the kinetic model, where is a dimensionless undetermined coefficient obtained by model training; the introduction of this correction term mathematically accurately depicts the core physicochemical property that the higher the initial crystallinity of the Ginkgo nut raw material, the more difficult the molecular chain deconstruction and rearrangement, and thus the slower the quality change rate;

[0053] To enable those skilled in the art to better understand and implement the present application, a set of exemplary parameter values calibrated through experiments in the Ginkgo nut powder processing scenario are provided here; these values can be used as initial references for implementing the present application: critical reference temperature : ; temperature sensitivity coefficient : ; mechanical force sensitivity coefficient : when the equivalent mechanical force is measured in terms of shear rate (unit ), dimension is a dimensionless coefficient, whose value can be taken as ; raw material quality correction coefficient : (dimensionless);

[0054] It should be emphasized that the optimal parameter values will vary with specific equipment and raw material batches, and those skilled in the art can fine-tune them based on the above reference values.

[0055] Example 4

[0056] The model predictive control optimization framework calculates the optimal process parameter combination by solving the cost function; the cost function includes a tracking error term for minimizing the deviation between the quality evolution trajectory and the ideal target quality trajectory, and a control cost term for penalizing the change of process parameters.

[0057] The control cost term is a weighted control increment penalty term, which normalizes the variation of control variables using the normal operating range of each process parameter and adjusts the penalty level of different control variable variations through a weight matrix.

[0058] In this embodiment, the function of the process parameter dynamic optimization module is to find the optimal path for the system to reach the ideal quality target based on the predicted quality evolution trajectory in Embodiment 3; to achieve this purpose, the module adopts a model predictive control (MPC) optimization framework; in each control cycle , the framework not only calculates the optimal action at the next instant, but also plans a complete set of optimal process parameter sequences within a finite time domain in the future through solving a carefully designed cost function ;

[0059] The design of this cost function aims to achieve a well-thought-out control strategy, which not only ensures that the future quality trajectory can track the preset ideal target as closely as possible, but also takes into account the process stability and safety, effectively suppressing the sharp fluctuations of process parameters, thereby avoiding shocks to the equipment or causing product quality shocks; in each control cycle, the optimization solver finds a set of future control sequences to minimize the total cost, which is essentially a dynamic trade-off process, thereby guiding the system to find a practical control strategy that can effectively correct quality deviations while maintaining smooth changes in process parameters; this directly ensures the stability, efficiency and safety of the entire processing process of the fruit; the expression of the model predictive control cost function is as follows:

[0060]

[0061] where, is the dimensionless total cost to be minimized; the first term of the cost function is the tracking error term; here, is the predicted quality state value at the future step obtained by repeatedly calling the prediction model of Embodiment 3 based on all information at the current time ; and is the ideal quality target value at the same future time defined by process experts based on historical data of producing golden batch products or texture design in the product development stage; the core role of this term is to drive the future output of the system to approach the desired target as closely as possible; the second term of the cost function is the weighted control increment penalty term, i.e., the control cost term; here, represents the variation of the process parameter vector at the future step. A sophisticated design is that the variation is normalized by the normal operating range of each process parameter The normalization is performed; The source is the physical specification and safety procedure of the processing equipment; this normalization makes the control variable changes of different physical dimension and order of magnitude be put on the same scale for comparison and weighting; in addition, is a diagonal weight matrix, which is used to adjust the punishment degree of different control variable amplitude respectively, and its specific value is a tunable hyperparameter, which is derived from on-site process debugging, aiming to balance the rapidity of control response and the stability of process operation; is the length of the prediction time domain, and its selection needs to balance between prediction accuracy and computational burden, and its value is usually finally determined through simulation or field test, so that the total prediction time can cover a major time constant of material quality change.

[0062] Embodiment 5

[0063] The closed-loop feedback control execution module issues the optimal process parameter combination to the programmable logic controller, which drives the heater, valve or motor.

[0064] In this embodiment, the function of the closed-loop feedback control execution module is to transform the decision in the information world into action in the physical world; in the process parameter dynamic optimization module of embodiment 4, the optimal process parameter sequence of the future N steps is obtained by solving the cost function After that, the closed-loop feedback control execution module strictly follows the basic principle of model predictive control, and only extracts and executes the first element in the sequence, i.e. the optimal process parameter combination at the current time ;

[0065] Sequentially, the module issues this optimal instruction , such as the optimal temperature and the optimal shear rate , in the form of a digital signal to the programmable logic controller (PLC) at the bottom of the factory; as the core control unit of industrial automation, the PLC will immediately analyze these instructions and accurately drive the physical actuators connected to it to respond; for example, the PLC will adjust the steam valve opening of the supply to the cooking pot to make the material temperature reach , and at the same time adjust the motor speed of the stirring system to apply a mechanical force equivalent to ; this completes the transformation from an optimal calculation result to an accurate physical operation, which constitutes the execution link of the control loop; after one control cycle , the system will obtain the new material quality state and start a new round of prediction-optimization-execution cycle based on it, thus forming a closed-loop feedback that is repeated and continuously corrected, ensuring that the processing process always runs on the optimal track.

[0066] Example 6

[0067] The current quality state, which is obtained by real-time measurement through online sensors deployed on the processing equipment, or, in the absence of online sensors, by a model-based state estimation method, combined with the predicted value of the previous control cycle and the model after correction.

[0068] In the implementation of this embodiment, it is crucial to accurately know the current quality state at the beginning of each control cycle in order to achieve effective closed-loop feedback; the present invention provides two highly flexible and adaptable technical paths for this purpose.

[0069] The first path, which is also the optimal path, is direct measurement; by deploying online sensors, such as online viscometers, rheometers, or specific optical probes, on core processing equipment such as cooking pots or maturation pipelines, the physical or chemical properties of the material are monitored in real time and continuously, and the measured values are converted into the current quality state through a calibration model .

[0070] In order to make the core indicator "quality state" have a clear physical meaning and measurability, the present invention defines it as the gelatinization degree of Gorgon nut material, which is a dimensionless index between 0 and 1; the calculation method can be associated with the measurement value of the online viscometer; for example, normalization can be performed through the following formula:

[0071]

[0072] where, is the quality state at the current time ; is the material viscosity measured by the online viscometer in real time; is the initial viscosity of the material entering the processing equipment; is the material viscosity at the ideal processing endpoint, i.e., the material viscosity of the golden batch product, determined through experiments; through this definition, indicates that the material is in the initial unprocessed state, while represents that the material has reached the ideal final quality state;

[0073] The obtained in this way is the most real and timely, and can provide the highest quality feedback signal for subsequent prediction and control;

[0074] In some existing production line modification or cost-limited scenarios, it may not be possible to install online sensors; for this situation, the present invention provides a second path, a model-based state estimation method; in this mode, the system will use the previous control cycle, i.e., time , to estimate the current quality state at time ​the optimal estimate of the current quality state is calculated by combining the quality prediction value of the model with a correction term based on the actual process parameters and the model bias;

[0075] The model-based state estimation method can adopt an observer structure with an intermittent correction term; in most control cycles without offline sampling, the system directly uses the model's own prediction as the optimal estimate of the current state, and when new offline measurement data is obtained, the model state is corrected once to eliminate accumulated errors, and the estimate of the current quality state is calculated by the following formula:

[0076] 1. At the time when there is no offline measurement value :

[0077]

[0078] 2. At a specific time when offline measurement values are obtained :

[0079]

[0080] wherein, is the estimate of the current quality state; is the prediction value of the model at the previous time for the current time quality state; the correction only occurs when offline measurement is obtained, and the correction term consists of two parts: is the corrected state estimate at a specific time ; is the true quality state of the material measured in the laboratory by offline sampling, for example, by measuring its viscosity and calculating it according to the supplementary formula of point 1; is the quality prediction value of the quality prediction model at the same time point as offline sampling ; is a preset observer gain coefficient, for example, between 0.1 and 0.5, used to adjust the strength of the correction; This method is indirect, but it fully utilizes the prediction ability of the existing high-precision dynamic model, and through periodic accurate offline measurement values, the model is anchored and calibrated, realizing soft measurement in the case of hardware missing, so whether through direct physical sensing or indirect model estimation, the system can reliably obtain the current quality state

[0081] necessary to start each closed-loop iteration, greatly enhancing the universality and robustness of the quality control system in different industrial environments.

[0082] ​It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. An AI-based quality control system for the processing of gorgon fruit, characterized in that: include: The quality perception and quantification module is used to perform real-time non-destructive testing of material units entering the processing flow. Using a pre-established chemometric model, the collected raw spectral data is analyzed and quantified into the initial starch crystallinity index; A quality dynamic evolution prediction module, which predicts the quality evolution trajectory of the material unit within the future processing time based on the initial starch crystallinity index, current quality status and real-time process parameters through a dynamic evolution prediction model; A dynamic process parameter optimization module is used to compare the quality evolution trajectory with the preset ideal target quality trajectory and calculate the optimal process parameter combination at the next moment through a model predictive control optimization framework; The closed-loop feedback control execution module is used to send the optimal process parameter combination to the actuator of the processing equipment to complete real-time regulation of the processing process. Based on the quality status of the material unit after regulation, a new round of quality evolution trajectory prediction, optimization and control is carried out to form a closed-loop feedback.

2. The AI-based quality control system for the processing of gorgon fruit according to claim 1, characterized in that: The quality perception and quantification module includes a near-infrared spectrum detection system deployed on the material conveying line, and the near-infrared spectrum detection system is used to collect the original spectrum data of each material unit.

3. The AI-based quality control system for the processing of gorgon fruit according to claim 2, characterized in that: The chemometric model includes: collecting gorgon fruit samples with different starch crystallinity; using an X-ray diffraction method to determine the crystallinity values ​​of the gorgon fruit samples, and simultaneously collecting their near-infrared spectra; using a partial least squares regression algorithm, with the near-infrared spectra as input and the crystallinity values ​​as output for training.

4. The AI-based quality control system for the processing of gorgon fruit according to claim 1, characterized in that: The dynamic evolution prediction model is composed of a discrete time evolution equation of quality state and a constitutive model of quality instantaneous change rate; the discrete time evolution equation of quality state is used to iteratively calculate the quality state at the next moment based on the current quality state and the quality instantaneous change rate.

5. The AI-based quality control system for the processing of gorgon fruit according to claim 4, characterized in that: The instantaneous quality change rate constitutive model is used to construct a functional relationship between the instantaneous quality change rate and the real-time process parameters, the current quality state and the initial starch crystallinity index; the initial starch crystallinity index is introduced as a raw material quality correction term to characterize the characteristic that the higher the initial crystallinity of the raw material, the slower the raw material quality change rate.

6. The AI-based quality control system for the processing of gorgon fruit according to claim 1, characterized in that: The model predictive control optimization framework calculates the optimal process parameter combination by solving a cost function; the cost function includes a tracking error term for minimizing the deviation between the quality evolution trajectory and the ideal target quality trajectory, and a control cost term for penalizing changes in process parameters.

7. The AI-based quality control system for processing of gorgon fruit according to claim 6, characterized in that: The control cost term is a weighted control increment penalty term, which uses the normal operating range of each process parameter to normalize the change in the control quantity and adjusts the penalty intensity for different control quantity changes through a weight matrix.

8. The AI-based quality control system for the processing of gorgon fruit according to claim 1, characterized in that: The closed-loop feedback control execution module sends the optimal process parameter combination to a programmable logic controller, and the programmable logic controller drives the heater, valve or motor.

9. The AI-based quality control system for processing of gorgon fruit according to claim 1, characterized in that: The current quality state is obtained by real-time measurement through online sensors deployed on the processing equipment, or when there are no online sensors, through a model-based state estimation method, combined with the predicted value of the previous control cycle and model correction.

Images