Fresh milk pre-fermentation flavor regulation and control and multi-temperature-zone collaborative distribution method and system

Abstract

The invention provides a fresh milk pre-fermentation flavor regulation and control and multi-temperature-zone collaborative distribution method and system, belongs to the technical field of food supply chain intelligent control, and is used for solving the problem of unstable flavor quality after distribution caused by disconnection of production and logistics link control in a traditional fresh milk supply chain in related technologies. According to the method, a full supply chain state of fresh milk from production to distribution is uniformly mapped to a high-dimensional statistical manifold, a manifold geometric structure is updated through online learning, and an optimal geodesic line connecting a current state and a target state is calculated, so that a physical field of a production end and a temperature control field of a distribution end are cooperatively regulated and controlled, and a system is guided to evolve along an optimal path. The system comprises a holographic sensing layer, a field regulation and control execution layer and a calculation layer which correspond to each other. According to the invention, dynamic integrated collaborative optimization of fresh milk flavor formation and logistics environment is realized, and the flavor and taste of a terminal product are guaranteed.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology for the food supply chain, and in particular to a method and system for flavor regulation and multi-temperature zone collaborative distribution of fresh milk pre-fermentation. Background Technology

[0002] As consumers increasingly demand freshness and flavor from short-shelf-life foods like fresh milk, ensuring consistent quality throughout the entire process from production to consumption has become a key challenge for the industry. The flavor of fresh milk relies on the biochemical process of probiotic fermentation during production, while maintaining its quality during distribution is strictly constrained by the temperature environment at the logistics stage. These two stages have traditionally been managed by different systems, resulting in a natural disconnect in their control objectives and execution logic.

[0003] Current technologies largely focus on optimizing single stages. In production, flavor is mainly controlled by adjusting fermentation process parameters; in logistics, cold chain technology is primarily used to maintain a constant low temperature. However, these methods often treat fresh milk as a static, passive commodity, neglecting its essential nature as a continuously undergoing biochemical reactions within an active system. Even after pre-fermentation, fresh milk remains in a slow post-ripening stage during delivery, with its flavor composition and pH constantly changing, making it extremely sensitive to temperature.

[0004] The main drawback of existing technologies lies in the disconnect between production and logistics control. Static, constant temperature control strategies during delivery cannot adapt to the dynamic biochemical changes in fresh milk during transit, easily leading to excessive post-acidification or insufficient flavor development. Simultaneously, traditional methods lack unified modeling and real-time sensing capabilities for the entire supply chain, failing to dynamically adjust control strategies based on actual quality evolution, resulting in difficulty in guaranteeing consistent flavor and taste after long-distance delivery. Therefore, there is an urgent need for an intelligent solution that can connect production and logistics, enabling dynamic sensing and coordinated regulation of the active state of fresh milk. Summary of the Invention

[0005] This application provides a method and system for flavor regulation and multi-temperature zone collaborative distribution of fresh milk pre-fermentation. It can dynamically characterize and collaboratively optimize the entire process of fresh milk from fermentation to distribution using a unified mathematical model, thereby solving the problem of flavor quality fluctuations caused by the disconnect between production and logistics control.

[0006] In a first aspect, this application provides a method for flavor control and multi-temperature zone collaborative distribution of fresh milk pre-fermentation. The method includes the following steps: mapping the entire supply chain state of fresh milk from production to distribution to points on a high-dimensional statistical manifold; updating the geometric structure of the high-dimensional statistical manifold based on real-time sensing data, the geometric structure being characterized by a metric tensor and affine connections; calculating the optimal geodesic connecting points on the manifold representing the current supply chain state and points or regions representing the target supply chain state, the optimal geodesic being defined by the affine connections; and changing the geometric structure of the high-dimensional statistical manifold by controlling the physical field acting on the fresh milk production process and the temperature control field acting on the distribution process, thereby guiding the actual evolution trajectory of the supply chain system.

[0007] By adopting the above technical solution, this application creatively abstracts complex biochemical and physical processes into a geometric evolution problem on a high-dimensional manifold. By calculating and tracing the "optimal geodesic" connecting the current state and the target state, it provides accurate and globally optimal dynamic trajectory planning for the coordinated control of production and logistics, realizing a fundamental shift from static segmented control to dynamic integrated optimization.

[0008] Furthermore, the process of mapping the entire supply chain status of fresh milk from production to distribution to points on a high-dimensional statistical manifold includes: acquiring microbial metabolic state data from the production process and multiphysics field data from the distribution process; converting the microbial metabolic state data and the multiphysics field data into corresponding probability distribution representations respectively; mapping the converted probability distribution representations to a unified latent space through an encoder, and constructing the latent space as the high-dimensional statistical manifold.

[0009] By adopting the above technical solution, heterogeneous production biochemical data and logistics physical data are homogenized into probability distributions and mapped onto manifolds with clear information geometric meaning, laying a unified mathematical foundation for subsequent precise quantitative analysis and collaborative optimization based on manifold metric.

[0010] Furthermore, updating the geometric structure of the high-dimensional statistical manifold based on real-time sensing data includes: using the real-time acquired sensing data stream to adjust the parameters of the encoder online through neural manifold learning; and updating the metric tensor of the latent space by pulling back the mapping according to the adjusted encoder, thereby updating the geometric structure of the high-dimensional statistical manifold.

[0011] By adopting the above technical solution, online adaptive learning and updating of manifold geometry is realized, enabling the model to dynamically adapt to changes such as different batches of raw materials and environmental disturbances, ensuring the real-time accuracy of state representation and the robustness of control strategies.

[0012] Furthermore, the affine connection is decomposed into a production connection component, a logistics connection component, and a coupling connection component; wherein, the coupling connection component is described by a gauge field to characterize the dynamic interaction between the production process and the logistics process; the step of changing the geometric structure of the high-dimensional statistical manifold by adjusting the physical field and the temperature control field specifically includes changing the affine connection by adjusting the parameters of the physical field and the temperature control field.

[0013] By adopting the above technical solution, the concept of "coupling" in control theory is deepened into a "connection" decomposition on a manifold, and gauge field theory is introduced to provide a mathematical description. This makes it possible to model and actively control the complex interactions between production and logistics, enabling precise coordination of the dynamic behavior of the two subsystems.

[0014] Furthermore, it also includes state verification and repair steps: the sensor network deployed in the production and distribution process is regarded as the discretized boundary of the high-dimensional statistical manifold; the correlation function between sensor data on the boundary is calculated, and the correlation function is verified to satisfy the preset set of constraint equations; if the verification is not satisfied, the internal geometry of the high-dimensional statistical manifold is repaired by adjusting the control parameters on the boundary.

[0015] By adopting the above technical solution and drawing on the principles of holography, the complex verification problem of the internal high-dimensional manifold state is transformed into a lightweight verification of data correlations on its boundary (sensor network). This provides an efficient and reliable mechanism for system state monitoring and anomaly diagnosis.

[0016] Furthermore, the method of repairing the internal geometry by adjusting the control parameters on the boundary includes: identifying the boundary variables that cause the constraint equations to be violated; introducing corresponding adjustment terms into the boundary system model and optimizing their coefficients so that the new boundary correlation function satisfies the constraint equations again.

[0017] By adopting the above technical solution, closed-loop repair based on boundary feedback is achieved. When an anomaly is detected, the system can automatically perform a correction action at the boundary. According to the principle of holographic duality, this correction will equivalently repair the evolution trajectory of the internal high-dimensional state, ensuring the system's self-healing capability.

[0018] Furthermore, the regulation of the physical field acting on the fresh milk production process includes applying a programmable multiphysics field to the fermentation system to form a spatial interference pattern, thereby regulating the distribution of metabolic activity of the microbial community.

[0019] By adopting the above technical solution, the control of a single field strength is upgraded to the regulation of a "spatial interference pattern" formed by the synergy of multiple physical fields. This can provide precise spatial guidance for the microbial community in the fermenter and optimize its metabolic pathways and the efficiency of flavor substance synthesis.

[0020] Furthermore, the regulation of the temperature control field acting on the delivery process includes: controlling a temperature control unit with anisotropic thermal conductivity inside the delivery container to form a non-reciprocal heat flow path inside the container, thereby constructing the required dynamic temperature field distribution.

[0021] By adopting the above technical solution, the limitations of traditional temperature control based on symmetrical heat conduction are broken through. By using anisotropic materials to achieve "non-reciprocal" heat flow, a complex dynamic temperature field that is difficult to achieve by traditional methods can be constructed, which can accurately match the differentiated temperature control needs of fresh milk in different storage locations.

[0022] Furthermore, the real-time sensing data includes a first type of data reflecting the quantum coherence of microorganisms, a second type of imaging data reflecting the spatial distribution of the microscopic dielectric properties of fresh milk, and a third type of sensing data reflecting the spatial distribution of multi-physics fields within the delivery container.

[0023] By adopting the above technical solutions, cutting-edge sensing technologies such as quantum sensing, terahertz holographic imaging, and distributed fiber optic sensing are integrated, providing high-dimensional and high-precision multi-source data input for the complex models mentioned above, which constitutes the physical basis for the precise perception of the entire intelligent system.

[0024] Secondly, this application provides a fresh milk pre-fermentation flavor control and multi-temperature zone collaborative delivery system. The system is used to execute the method as described in any one of the first aspects, comprising: a holographic sensing layer, including a first sensing unit for acquiring the first type of data, a second imaging unit for acquiring the second type of imaging data, and a third distributed sensing unit for acquiring the third type of sensing data; a field control execution layer, including a multiphysics field modulation unit disposed in the fermentation system for generating programmable spatial interference patterns, and a temperature control unit disposed in the delivery container for implementing non-reciprocal heat flow paths; and a computation layer for constructing and updating the high-dimensional statistical manifold, calculating the optimal geodesic, and generating control instructions for the field control execution layer.

[0025] By adopting the above technical solution, a hardware entity system that is completely corresponding to the aforementioned method was constructed. Through a three-layer collaborative architecture of "perception-computation-control", the abstract algorithm model was transformed into executable physical control, realizing a closed loop from data perception to precise execution.

[0026] In summary, this application has at least the following beneficial effects: This paper presents an intelligent collaborative control paradigm for the fresh milk supply chain based on unified information geometry modeling, which realizes dynamic integrated optimization of production fermentation and logistics distribution, and fundamentally guarantees the flavor quality of the end product. By introducing a gauge field description coupling connection and a holographic boundary verification mechanism, an innovative theoretical framework and implementation path are provided for cross-scale collaboration and self-healing of complex industrial systems. By utilizing advanced execution methods such as programmable multiphysics interference and non-reciprocal heat flow control, an unprecedented level of precision control over the microenvironment of microbial metabolism and delivery has been achieved.

[0027] It should be understood that the description in the Summary Section is not intended to limit the key or essential features of the embodiments of this application, nor is it intended to restrict the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description

[0028] The above and other features, advantages, and aspects of the embodiments of this application will become more apparent from the accompanying drawings and the following detailed description. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1 The diagram illustrates the principle of a fresh milk pre-fermentation flavor control and multi-temperature zone collaborative delivery system according to an embodiment of this application.

[0029] Figure 2 A flowchart of a fresh milk pre-fermentation flavor control and multi-temperature zone collaborative delivery method is shown in an embodiment of this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0032] This application provides a method and system for flavor control and multi-temperature zone collaborative distribution of fresh milk pre-fermentation. By mapping the entire supply chain status to a high-dimensional manifold for geometric representation and dynamic optimization, it achieves deep collaboration and adaptive control between production fermentation and logistics distribution, thereby fundamentally ensuring the high consistency of flavor and taste of fresh milk after long-distance and complex distribution.

[0033] In the first aspect, embodiments of this application disclose a fresh milk pre-fermentation flavor control and multi-temperature zone collaborative delivery system.

[0034] Figure 1 The diagram illustrates the principle of a fresh milk pre-fermentation flavor control and multi-temperature zone collaborative delivery system according to an embodiment of this application.

[0035] Reference Figure 1 The system comprises a holographic perception layer, a field control execution layer, and a computing layer. These three layers are interconnected through a high-speed communication network, forming a closed-loop intelligent hardware environment that supports the implementation of the method. This high-speed communication network may include a wired industrial Ethernet network within the factory, a 5G or satellite communication link between the delivery vehicle and the cloud, and a CAN or TSN network within the vehicle, ensuring real-time and reliable transmission of control commands and sensing data. The end-to-end communication latency is required to be less than 100 milliseconds, preferably 50 milliseconds.

[0036] The holographic sensing layer constitutes the data acquisition front end of the system. This layer includes a first sensing unit and a second imaging unit deployed in the production stage, and a third distributed sensing unit deployed in the distribution stage. The first sensing unit is specifically a quantum sensing array based on materials such as diamond NV centers, which monitors the quantum spin state of microorganisms inside the fermenter in a non-invasive manner. In one specific implementation, this array consists of multiple diamond NV center sensor probes embedded in the fermenter jacket or placed in a sterile protective cover inside the tank in a mesh configuration. The probe spacing is adjustable between 10 cm and 50 cm, preferably 30 cm, and is used to measure the relaxation times T1 and T2 of the local micromagnetic field generated by microbial metabolic activity. Its measurement sensitivity needs to reach the nanotesla level, for example, 1 to 100 nanotesla, preferably 10 nanotesla, thereby obtaining first-type data reflecting the coherence and activity of metabolic pathways. The second imaging unit is specifically a terahertz holographic imaging system, which performs three-dimensional tomographic scanning of fresh milk. The system's operating frequency range can be selected between 0.1 and 3 THz, such as 0.5 THz, 1.5 THz, or 2.8 THz, preferably 1.0 THz, to balance penetration depth and resolution; the scanning resolution can reach sub-millimeter level, such as 0.2 mm to 2 mm, preferably 1 mm, thereby acquiring second-type imaging data reflecting the spatial distribution of microscopic dielectric constants such as water, fat globules, and protein networks in fresh milk. The third distributed sensing unit is specifically a multi-core fiber optic sensing network laid on the inner walls, partitions, and shelf surfaces of the delivery vehicle compartment. The number of fiber cores can be 4, 8, or 12, preferably 8, used for sensing temperature, strain, and vibration, respectively; the spatial sampling interval is 5 cm to 20 cm, preferably 10 cm, and the temperature measurement accuracy is ±0.1°C to ±0.5°C, preferably ±0.2°C, thereby synchronously and at high resolution acquiring spatial distribution data of multiple physical fields such as temperature, stress, and vibration acceleration inside the compartment, i.e., third-type sensing data.

[0037] The field modulation execution layer constitutes the physical control backend of the system. This layer includes a multiphysics field modulation unit installed in the fermentation system and a temperature control unit installed in the delivery container. The multiphysics field modulation unit integrates independently programmable electromagnetic field, acoustic field, or optical field generators to generate composite physical fields with specific spatial interference patterns. For example, the electromagnetic field can be generated by multiple Helmholtz coil pairs arranged on the outer wall of the fermenter, with an adjustable field strength range from 0.1 to 10 mT and an adjustable frequency range from 0 to 100 Hz; the acoustic field can be generated by an array of ultrasonic transducers arranged at the bottom of the tank, with a sound pressure level range from 130 to 150 dB and a frequency range from 20 to 40 kHz; the optical field can be generated by a multi-wavelength LED array at the top, with an adjustable light intensity range from 10 to 1000 μmol·m⁻²·s⁻¹ and wavelengths covering 450 nm for blue light and 660 nm for red light. By coordinating and controlling the intensity, frequency, and phase of these fields, standing wave or traveling wave interference patterns that promote the growth of specific probiotics (such as Streptococcus thermophilus) can be formed within the container, thereby regulating the spatial distribution of metabolic activity in the microbial community. The temperature control unit employs a smart panel based on topological insulator materials or phase change memory alloys to achieve non-reciprocal heat flow paths. The topological insulator panel, for example, is composed of a doped Bi2Te3 thin film, and its in-plane thermal conductivity anisotropy is altered by applying a directional electric field (e.g., 5 to 20 V / cm, preferably 10 V / cm), causing heat flow to be preferentially conducted in a specific direction, with a thermal conductivity anisotropy ratio of 1.5:1 to 3:1, preferably 2:1. Phase change memory alloy panels, for example, are woven from NiTi alloy wires. By applying a pulsed current (current density 1 to 10 A / mm², preferably 5 A / mm²), a reversible phase transformation between austenite and martensite is triggered, thereby dynamically changing the physical shape of the local thermal resistance and the cabin space partition, realizing the dynamic reconstruction of the temperature field. The temperature setpoint adjustment resolution can reach 0.5°C.

[0038] The computational layer constitutes the core decision-making center of the system. This layer consists of neuromorphic computing chips and optical computing coprocessors deployed at the edge, and quantum simulation clusters and high-performance computing servers deployed in the cloud. Edge computing devices (such as chips based on Intel Loihi or IBM TrueNorth architectures) are responsible for preprocessing the raw high-dimensional data uploaded from the holographic perception layer and performing lightweight inference of manifold encoding, with a response time requirement of less than 10 milliseconds. The optical computing coprocessor uses photonic integrated circuits to perform intensive operations such as matrix multiplication and Fourier transform to accelerate the forward propagation process of the encoder and decoder in the neural manifold model. The high-performance computing cluster in the cloud is responsible for performing large-scale training tasks, such as training an initial model on historical datasets using the neural manifold learning methods disclosed in the aforementioned embodiments, and simulating gauge field evolution using quantum simulation clusters (such as Schrödinger equation solvers based on classical computers) to optimize the parameters of the coupled connection components. The core function of the computational layer is to receive and process multi-source data streams from the holographic sensing layer, perform the construction and online updating of high-dimensional statistical manifolds, solve for the optimal geodesic, and generate control commands, and then send the commands to the field control execution layer. Specifically, the algorithm software module in the computational layer takes real-time sensing data (first, second, and third types of data) as input, updates the manifold metric through online learning, and solves for the geodesic from the current state point to the target flavor state region on the updated manifold. This target region is defined by the numerical range of the target flavor substances (such as ethyl butyrate concentration of 2-4 mg / L and acetaldehyde concentration of 10-20 mg / L). Then, based on the ideal geometric structure changes corresponding to the geodesic, the algorithm reverse-engineers the precise control parameter sequence required by the multiphysics modulation unit and the temperature control unit.

[0039] The raw data collected in real time by the holographic perception layer is uploaded to the computing layer for processing and analysis via a high-speed network. The computing layer generates precise control commands based on the analysis results, which are then distributed to the corresponding units in the field control execution layer via the same network. The effects of the execution layer's actions, such as changes in metabolite concentration in the fermenter or redistribution of the temperature field within the vehicle, are captured in real time by the perception layer and fed back to the computing layer. For example, when the temperature control unit is adjusted, the distributed fiber optic network monitors the new temperature distribution. The computing layer compares this measured data with the temperature field corresponding to the expected geodesic. If the deviation exceeds a threshold (e.g., ±0.5°C), it triggers fine-tuning of control parameters or relearning of the manifold model, thus forming a complete closed loop from perception and decision-making to execution, and then to verification and correction. This closed-loop intelligent hardware environment jointly supports the physical realization and stable operation of the aforementioned fresh milk pre-fermentation flavor control and multi-temperature zone collaborative delivery method.

[0040] Secondly, this application discloses a method for flavor control and multi-temperature zone collaborative delivery of fresh milk pre-fermentation. Figure 2A flowchart of a fresh milk pre-fermentation flavor control and multi-temperature zone collaborative delivery method is shown in an embodiment of this application.

[0041] Reference Figure 2 The method is implemented by constructing a closed-loop intelligent decision-making system of "perception-modeling-planning-control-verification". Its core inventive concept is to regard the entire supply chain of fresh milk from pre-fermentation production to multi-temperature zone distribution as a dynamic and evolving complex system. It innovatively adopts information geometry theory to represent the state and goal of the system on a high-dimensional statistical manifold. By calculating and tracking the optimal geodesic on the manifold, the biochemical process at the production end and the physical process at the logistics end are coordinated and optimized in real time under a unified mathematical framework.

[0042] The method begins with multi-source sensing and unified manifold coding of the entire supply chain status. The system acquires three types of heterogeneous sensing data in real time. The first type of data comes from quantum sensing arrays deployed in the fermenter, such as sensors based on diamond NV centers, which measure the local magnetic field fluctuations generated by microbial metabolic activity, specifically outputting the spin relaxation time. and . (Longitudinal relaxation time, typically ranging from 1 to 10 milliseconds, such as 2 milliseconds, 5 milliseconds, or 8 milliseconds) reflects energy dissipation. (The transverse relaxation time, typically ranging from 0.1 ms to 2 ms, e.g., 0.5 ms, 1 ms, or 1.5 ms) reflects the ability to maintain quantum coherence, and together they constitute a vector reflecting metabolic activity. The second type of data comes from a terahertz holographic imaging system, which scans the fresh milk inside the tank to acquire each voxel. Complex permittivity at ... The real part The imaginary part is related to the density of matter. Reflecting absorption characteristics, a three-dimensional data volume is formed. The third type of data comes from a multi-core distributed optical fiber sensor network installed inside the delivery vehicle, which synchronously measures spatial location. Temperature at the location (Accuracy ±0.2°C) and micro-strain , constitute physical field vectors Spatial distribution.

[0043] Next, the system performs data homogenization and manifold mapping. Due to the aforementioned data... , , The dimensions and structures are vastly different, requiring them to be converted into mathematical objects that can be processed uniformly. For continuous scalar data (such as...), this is particularly important. This is transformed into a probability density function using nonparametric kernel density estimation. Specifically, for Temperature values ​​at each spatial sampling point Its probability density estimate is:

[0044] in, For kernel functions (such as Gaussian kernels) ), This is a bandwidth parameter, and its value is determined by Silverman rules, for example... , This is the sample standard deviation. For discrete or vector data (such as metabolic state vectors) Then, construct an empirical probability distribution. Represent the probability distribution of all data. The vectors are concatenated into a high-dimensional vector, which is then input into a depth autoencoder. This encoder... Map the input to Dimensions (e.g.) The hidden space Encoder parameters By minimizing the loss function Training was conducted, among which It is the mean square error between the input and the decoder's reconstructed output. It is a manifold regularization term that encourages latent spaces. It has a smooth manifold structure. After training, the latent space... It is itself constructed as the high-dimensional statistical manifold. One of the points It uniquely encodes the overall state of the entire supply chain at a given moment.

[0045] In manifold After establishment, the method enters the online adaptive learning phase of the manifold geometry. The system receives real-time data streams and updates the encoder parameters online through neural manifold learning. Thus, the manifold is dynamically adjusted. The local geometry of a manifold is determined by its metric tensor and affine connections. Fully characterized. The metric tensor defines the distance between two infinitely close points on a manifold. In the hidden space In this context, the metric is obtained from the data space through pullback mapping. Assume the data space uses a standard Euclidean metric. The encoder is mapped to Then the metric of the latent space is:

[0046] in, It is a decoder At point Find the (pseudo) inverse elements of the Jacobian matrix, or calculate them using automatic differentiation techniques. Metric It determines the information distance between state points.

[0047] More importantly, it relates to affine communication. Modeling and decomposition. Connections define how vectors move parallel to curves on a manifold, determining the shape of geodesics. In this method, It is explicitly decomposed into the sum of three components with definite physical meaning:

[0048] To produce the connecting components, it is constrained by a microbial metabolic kinetic model. For example, its components in certain directions can be constrained based on mass conservation and enzyme kinetics. As a logistics linking component, it is constrained by the laws of thermodynamics and fluid dynamics equations. The most innovative is the coupled linking component. It is modeled as a gauge potential (Yang-Mills field). The expression:

[0049] in, The selected Lie algebra (e.g.) The structure constants of ) It is a normative potential In manifold coordinates The component in the direction. This gauge field. It is dynamic, and its field strength It intuitively describes the "strength" and "curvature" of the interaction between the two subsystems of production and logistics. Adjusting the normative potential through online learning... The system can then dynamically adjust the coupling method between the two subprocesses.

[0050] Based on real-time updated manifold geometry The method performs geometric programming of the optimal evolutionary trajectory, i.e., calculates the optimal geodesic. The target state is defined as a region on the manifold. This region is mapped from target flavor quality indicators, for example, corresponding to all possible state points with pH values ​​between 6.4 and 6.6 and ethyl butyrate concentrations between 2.5 and 3.5 mg / L. Let the current state point be... The goal is to find a connection arrive curve This makes the curve a locally shortest path on the manifold, i.e., it satisfies the geodesic equation:

[0051] in, These are curve parameters (which can correspond to actual time). It is the tangent vector of the curve. This is a system of second-order ordinary differential equations, combined with boundary conditions. and This constitutes a two-point boundary value problem. Numerical methods (such as the target method or relaxation method) are used to solve it, and the solution obtained is... This refers to the theoretically optimal path from the current state to the target flavor state on the entire supply chain state manifold. This geodesic... Its tangent vector field completely defines the future time period. Within, how should each subsystem of production and logistics coordinate and change?

[0052] Subsequently, the method performs a reverse translation and control from geometric planning to physical execution. This step involves abstracting the geodesic... This is translated into specific actuator control instructions. For the physical field control of the production process, the system needs to solve an inverse problem: to ensure that the system's state evolution follows... That is, connections on a manifold (in particular How should the complex physical fields within the fermenter be configured to change in a specific manner? This involves solving the inverse problems of coupled Maxwell's equations and the acoustic wave equations. For example, in order to... This induces a metabolic gradient corresponding to the tangential direction of the geodesic, and the system calculates a set of coil currents. and ultrasonic drive voltage This allows the energy standing wave nodal surface of the electromagnetic-acoustic interference pattern synthesized within the container to match the spatial distribution of the desired metabolically active region.

[0053] For temperature control during the delivery process, the system uses geodesics... The implicit, time-evolving ideal temperature field distribution is used to calculate the control signal for the temperature control unit. Taking a topological insulator panel as an example, the system solves the inverse problem of the heat conduction equation: given the desired temperature distribution... and external environmental heat disturbance By deducing the voltage distribution required to be applied to the different electrodes of the panel, we can deduce the voltage distribution that needs to be applied. This utilizes the anisotropic thermal conductivity of the material to guide heat flow. Asymmetric flow, precise construction and maintenance For example, the control voltage can be adjusted between 5V and 20V to produce a thermal conductivity anisotropy ratio of 1.5:1 to 3:1.

[0054] To ensure the robustness and self-healing capability of the entire system, the method incorporates a real-time verification and repair mechanism based on holographic principles. The system abstracts the sensor network in the physical world into a high-dimensional state manifold. boundary Based on the concept of holographic duality, the data correlation at the boundary should reflect the geometric health of the internal manifold. Specifically, the system calculates sensor readings at different locations and of different types at the boundary. The time delay correlation function between them:

[0055] in It's a time window. When the internal manifold geometry is correct and the system is operating normally (i.e., near a "healthy" conformal field-theoretic fixed point), these infinitely many correlation functions... A set of strong constraints called conformal bootstrap equations must be satisfied between them. A simplified exemplary constraint is: for three operators, their correlators should satisfy... A specific proportional relationship. The system continuously checks whether these constraints are satisfied.

[0056] If a constraint is detected to be significantly violated (e.g., the calculated proportion deviates from the theoretical value by more than a threshold, such as 10%), it means that the internal manifold... The geometry is "distorted" due to unknown perturbations. At this point, a repair mechanism is activated. First, by analyzing the constraint violation patterns, the most significant boundary variables violating the rules are identified, such as temperature sensor readings at a specific location. Then, in the effective field theory describing the boundary system, a marginal perturbation term coupled to this variable is introduced: Then, the system automatically adjusts the disturbance coefficient. (For example, minimizing the total residual of the constraint equations using gradient descent), so that in the new effective action... Below, the calculated boundary correlation function The bootstrap equations are satisfied again. According to the AdS / CFT duality principle, such a marginal perturbation to the boundary theory strictly corresponds to a conformal deformation of the internal manifold geometry on the gravitational side. Therefore, by adjusting the boundary parameters... The system mathematically "repairs" the internal high-dimensional state manifold. The geometry, thus pulling the actual evolution path of the system back to the optimal geodesic. Within its neighborhood.

[0057] In summary, this method, through the closed-loop operation of the above five core steps, achieves cross-scale, integrated, and adaptive intelligent control of the complex and active system of the fresh milk supply chain, from microscopic biochemical reactions to macroscopic logistics physics, thereby fundamentally ensuring the flavor consistency and high quality of fresh milk products after long-distance, multi-temperature zone delivery.

[0058] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to the embodiments of this application, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0059] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for flavor control and multi-temperature zone coordinated delivery of fresh milk pre-fermentation, characterized in that, Includes the following steps: The entire supply chain status of fresh milk, from production to distribution, is uniformly represented by points on a high-dimensional statistical manifold. Based on real-time sensing data, the geometric structure of the high-dimensional statistical manifold is updated, and the geometric structure is characterized by the metric tensor and affine connection. Calculate the optimal geodesic between a point on the connecting manifold representing the current supply chain state and a point or region representing the target supply chain state, the optimal geodesic being defined by the affine connection; By regulating the physical field acting on the fresh milk production process and the temperature control field acting on the distribution process, the geometric structure of the high-dimensional statistical manifold is changed to guide the actual evolution trajectory of the supply chain system.

2. The method according to claim 1, characterized in that, The points that map the entire supply chain status of fresh milk from production to distribution onto a high-dimensional statistical manifold include: Acquire microbial metabolic state data from the production process and multiphysics data from the distribution process; The microbial metabolic state data and the multiphysics data are respectively converted into corresponding probability distribution representations; The transformed probability distribution is mapped to a unified latent space by an encoder, and the latent space is constructed as the high-dimensional statistical manifold.

3. The method according to claim 2, characterized in that, The updating of the geometric structure of the high-dimensional statistical manifold based on real-time sensing data includes: The encoder parameters are adjusted online using real-time acquired sensor data streams through neural manifold learning; Based on the adjusted encoder, the metric tensor of the latent space is updated by pulling back the mapping, thereby updating the geometry of the high-dimensional statistical manifold.

4. The method according to claim 1, characterized in that, The affine communication is decomposed into a production communication component, a logistics communication component, and a coupling communication component. The coupling connection component is described by a gauge field, which is used to characterize the dynamic interaction between the production process and the logistics process; The method of altering the geometric structure of the high-dimensional statistical manifold by regulating the physical field and the temperature control field specifically includes changing the affine connection by regulating the parameters of the physical field and the temperature control field.

5. The method according to claim 1, characterized in that, It also includes status verification and repair steps: The sensor network deployed in the production and distribution process is regarded as the discretized boundary of the high-dimensional statistical manifold; Calculate the correlation function between sensor data on the boundary and verify whether the correlation function satisfies the preset set of constraint equations; If the verification is not satisfied, the internal geometry of the high-dimensional statistical manifold is repaired by adjusting the control parameters on the boundary.

6. The method according to claim 5, characterized in that, The method of repairing the internal geometry by adjusting control parameters on the boundary includes: Identify the boundary variables that cause the constraint equations to be violated; Introduce corresponding adjustment terms into the boundary system model and optimize their coefficients so that the new boundary correlation function can once again satisfy the constraint equations.

7. The method according to claim 1, characterized in that, The regulation of the physical fields acting on the fresh milk production process includes applying programmable multiphysics fields to the fermentation system to form spatial interference patterns, thereby regulating the distribution of metabolic activity of the microbial community.

8. The method according to claim 1, characterized in that, The regulation applied to the temperature control field during the delivery process includes: A temperature control unit with anisotropic thermal conductivity is used to control the delivery container to form a non-reciprocal heat flow path inside the container, thereby constructing the required dynamic temperature field distribution.

9. The method according to claim 1, characterized in that, The real-time sensing data includes a first type of data reflecting the quantum coherence of microorganisms, a second type of imaging data reflecting the spatial distribution of the microscopic dielectric properties of fresh milk, and a third type of sensing data reflecting the spatial distribution of multi-physics fields within the delivery container.

10. A fresh milk pre-fermentation flavor control and multi-temperature zone coordinated delivery system, used to perform the method as described in any one of claims 1 to 9, characterized in that, The system includes: The holographic sensing layer includes a first sensing unit for acquiring the first type of data, a second imaging unit for acquiring the second type of imaging data, and a third distributed sensing unit for acquiring the third type of sensing data. The field control execution layer includes a multiphysics field modulation unit installed in the fermentation system for generating programmable spatial interference patterns, and a temperature control unit installed in the delivery container for realizing non-reciprocal heat flow paths. The computational layer is used to construct and update the high-dimensional statistical manifold, calculate the optimal geodesic, and generate control instructions for the field control execution layer.

Images